[Debian-ha-commits] [pacemaker] annotated tag Pacemaker-1.1.7 created (now b5b0a7b)
Christoph Berg
myon at debian.org
Mon Jan 4 22:02:45 UTC 2016
This is an automated email from the git hooks/post-receive script.
myon pushed a change to annotated tag Pacemaker-1.1.7
in repository pacemaker.
at b5b0a7b (tag)
tagging ee0730e13d124c3d58f00016c3376a1de5323cff (commit)
tagged by Andrew Beekhof
on Thu Mar 29 08:52:11 2012 +1100
- Log -----------------------------------------------------------------
Pacemaker 1.1.7 - Final
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)
iEYEABECAAYFAk9ziAsACgkQ7DWE79RJ5ZrhZACglLFuE+bUF6WBVGJfK0oSmm2S
I54Anj2OqJb8sgcDTwTOwT19aFo1chog
=clQ+
-----END PGP SIGNATURE-----
Alan Robertson (217):
Put in some comments in the cts README about coordinating versions of
Put in a bunch of nice CTS improvements from Mi, Jun <jun.mi at intel.com>
Added a real replacement strnlen function.
Incorporated a large set of new function for the linux-ha project.
Put in various small fixes necessary to make an rpm properly.
Put in some minor CTS fixes.
Put in a few minor fixes and some slight code formatting changes.
Changed a distribution file from being README to README.in.
Minor cleanup, turning off some troublesome tests, and put in some temporary kludges to
Fixed a convenience target that alanr uses to update the web site image on his machine.
Added some code to allow a single machine to be able to monitor simultaneous HA
Removed a funny character Python is complaining about.
Got rid of some ARP kludge stuff.
Changed the CTS tests so that they take command line arguments (Gasp!)
Changed the type of strerror to not return a const string.
Committed some CTS test improvements from Mi Jun <jun.mi at intel.com>
Added a new lookforall() member function for the LogWatcher class.
Commented out bandwidthtest and split brain test for now, since they
Tweaked the fast failure detection test.
More corrections to the CTS test code for fast failure detection.
Added Start and Stop operations to the Resource class and subclasses...
Minor CTS enhancements for the LogWatcher class
Changed the name and messages of the FailureDetection test to not
Put in code to work around a no-such-child error which lmb seems to see...
Reordered events in the arp-capture-on-monitor-error code
Various mostly-minor CTS test fixups.
Indentation changes.
Moved libraries from heartbeat to lib directory
Moved heartbeat libraries to 'lib'
Added the Local Resource Manager from Zhen Huang and Jiang Dong Sun to CVS.
Put in the latest version of the CTS tests from Jun, Mi plus
Minor bandwidth measurement bug fixes.
Minor message reformatting...
Tweaked the 'bandwidth' test to use a unique temp file name, so that I could monitor more than
Put in some more failure debugging for the split braint test.
Updated CTS README
Put in some fixes into the cts readme.
Moved cl_status and cl_respawn to /usr/bin/, and made cl_status setuid.
Put in code to activate glibc's malloc integrity checking during tests...
Added more "normal errors" to the split-brain case...
Put in code to make bandwidthtest work with different version of tcpdump...
Made DoStonith default to On for CTS tests...
Fixed two lines of code that didn't compile:
Changed cl_status from setuid to setgid
Put in a commented workaround for what looks very much like a BEAM bug.
Straightened out some CTS messages...
Added core dump directories, and a bunch of code to cd into the
Fixed malloc/free prototypes in PILS to match glib conventions, not libc conventions...
Fixed a weird bug in making RPMS.
Undid the RPM 'specfile' change to the makefile. Apparently it wasn't needed.
Disambiguated a couple of macros in the CRM xml.h file.
Put in a fix for a memory overwrite problem found by BEAM.
Fixed a memory leak pointed out by BEAM.
Put in an obviously-missing break called out by BEAM.
Put in a hack which allows this code to compile without warnings on 64-bit platforms
Fixed a 'passing null object' BEAM problem.
Fixed a minor BEAM error. (passing null object)
Updated some flags for release 2 testing w/CTS
Minor changes to CTS - to enhance it slightly (due to a self-inflicted wound)
Added a non-debug message to trigger off so CTS can
Fixed the start/stop patterns to match Andrew's recommendation.
Turned off tracing of regular expression matching.
Upped the "jackpot" threshold for stopping testing from 20 to 100
Fixed a minor BEAM error, and removed duplicate debug statement.
BEAM warning suppression...
Removed erroneous messages-to-ignore.
Put in a largish number of signed/unsigned fixes
Switched an echo statement which actually does something to a
Fixed compilation errors: signed/unsigned comparisions
Put in a FIXME comment asking why the redundant path test was commented out.
BUG 346: Disabled discontinued code, turn on CRM, LRM by default, etc.
Fixed bug 427
Replaced a strlen("foo") with a STRLEN_CONST
Converted a number of strncmp(foo, "", strlen("")) to use STRNCMP_CONST
Converted a number of strncmp(foo, "", strlen("")) to use STRNCMP_CONST
Put in code to pick up the right syslog level from ha.cf (if present)
Put in the --trunc flag for CTS testing (truncate log files before starting)
Added new extracttests python script for giving Andrew (and others) test logs they're happy with.
Changed extracttests so that:
Changed the way the SNMP code is excluded from being built
Apparently we forgot to include cib_private.h in the Makefile.
Fixed an infinite loop bug in extracttests (not detecting EOF properly)
Got rid of the semi-ugly stat size on the output from CTS because we no
Fixed the STONITH test so it gives a legitimate list of hosts to the stonith command -- so it will work ;-)
Fixed a declartion placement problem (non-ANSI construct).
Removed an extra ; which terminated the declartion-portion prematurely.
Fixed an ANSI-compile-error problem.
BEAM FIX: NULL printed string.
Very minor output text changes in CTS.
Raised the timeout for the IPaddr resources we create by default
Added SpecialTest1 for Andrew.
Attempting to fix a problem in the ResourceRecover test.
Added a little additional debugging for some quorum problems.
BEAM bugs: operating on NULL - crm/crmd/lrm.c
Fixed a wee bug in my previous CRM/CCM debug code
Corrected previous correction to debug code :-(
Put in an additional test in a print statement.
Fixed a fairly major bug in the CRM where a long-lived malloced structure
Fixed a bug in some of Andrew's debugging code - it always thinks the CIB
Fixed some logic in extracttests that required the ranges in test numbers to be last
I put in comments on delaying pengine calls until the CIB and CRM have
Bugzilla bug 544: Converted LRM to return the IPC Channel rather than the
CTS fix: s/self.rsh/self.CM.rsh/
Changed self[foo] to self.CM[foo]
Fixed some more of the sloppy logging code which causes tests to crash
Fixed error in StopTest
Changed the SWIG directory to always be included in builds - even
Changed a #if 1 into just comments so it compiles on FreeBSD.
Fixed a bug - where a test hadn't been updated to correspond
Changed the R1 tests to also start the logging daemon.
The uuid code didn't make an RPM correctly, and when I got to looking into
Replaced uuid_t stuff with cl_uuid stuff...
Converted from nanosleep to mssleep() - missing include - so I changed it while I was here...
Fixed a couple minor uuid things.
Disabled a broken test.
Huge number of signed/unsigned comparisons.
Fixed the spelling of dependency and also made CTS aware that
Disabling stonithd test as per bug 692 comments.
Removed lib/bindings from the build process...
disabled bindings code.
Made stonithd applicable again.
Changed stonith test to not run when R2 is enabled.
Left out a line necessary for keeping the web site up to date.
Fixed signed/unsigned comparision
BEAM fixes of various sorts.
Re-alphabetized the list of scripts.
Disable curses code if some kind of curses libraries are not present.
Updated copyright dates.
Added code to explicitly test logging - and nothing else - before starting up.
Changed test logging to be non-blocking - to speed up startup.
Added disk free checks to CTS.
Moved a line to the right place.
Put in two missing '%' operators for CTS error messages.
Changed it so that breaking remote logging will keep the tests from
Increased timeout for testing logging to 30+#nodes seconds.
Added code to print the stack traceback on abnormal termination and to
Fixed the ResourceRecover test
Updated the history information on scandir.c from Rich Salz.
Put in some minor changes - including a copyright: none, and license: none statement.
Added final authorship verification email from Rick Salz to the file.
Put in a fix from Andrew for FC4 compiling...
Reverting two source files to 2.0.1 versions so that I can apply a patch
Applied a couple of patches that Andrew wanted to go into
I must have somehow botched Andrew's patch...
Made DoFencing on by default (same as --fencing yes)
Put in code to change to using the external/ssh plugin instead of the semi-built-in ssh plugin.
Fixed the syntax error from 948
Put (back?) import of sys module.
Minor English phrasing changes.
Made heartbeat-style RAs consider that they're stopped unless they say "running" or "OK"
Put in additional debug information for heartbeat RA debugging...
Added a debug printout to the LRM to print out the debug level currently in effect...
Changed the code to treat monitor as the same as status - since what the CRM
Minor code reformatting and simplification.
Separated out the "wait for all nodes to come up" function into
Put in a little code to make sure that we wait for nodes to come up
Changed Andrew's code to use the new Gsource for heartbeat API connections.
First version of cibadmin man page.
Added cibadmin man page.
Put in more examples, and added a missing id string.
Minor improvements in the man page.
Changed a format string to be a string literal - for GCC parameter checking reasons.
Changed a format string to be a string literal - for GCC parameter checking reasons.
Update to cibadmin man page.
Put some missing header files into a couple of makefiles.
Put in a patch due to Dejan Muhamedagic which enhances the
Upped some test timeouts (again)
Added w3m to a warning error message for building Linux-HA
Bugzilla Bug 1445: unsetenv has wrong return type in replace function
Merge??
Put in changes to add explicit timeouts for all the operations in CTS.
Increased timeouts for operations and added XML for the operations so they
Merged changes in from upstream
This is piece of: LinuxFoundation Bug 1534 compiling/installation error on OpenBSD
Cleaned up more of the mess left by a sloppy developer in this changeset:
Fixed a bug reported by Max Hofer <max.hofer at apus.co.at> - crm_uuid didn't return proper exit codes.
Fixed a couple of minor spelling errors.
Fixed a tengine compile error.
Bugzilla Bug 1612 haresources2cib @substitution@ broken (probably by a global change)
LF Bugzilla # compiling/installation error on OpenBSD
LF Bug # 1617 - Miscellaneous RPM and source cleanups
Merged in upstream changes...
Merge from upstream.
Patch applies to bug LF # 1534 - compiling/installation error on OpenBSD
Upstream merge
LF bug # 1650 - heartbeat should put scripts/non arch specific things in /usr/share
LF bug 1650: heartbeat should put scripts/non arch specific things in /usr/share
LF Bug 1576: shutdown hangs under certain shells // LF Bug 1534: compiling/installation error on OpenBSD
Merged in some changes from two different scripts from the 'test' branch.
Added code to cause logging to be restarted if it dies.
hg: branch merge from 'dev' branch
Put in a change to allow CTS to run. There is a sort-of-bug (in BSC/CTS) underlying this that I'll have to file a bug on.
LF bug # 1661 - crm_verify aborts checking and reoccurring crm_abort errors in logs
LF bug # 1661 - move symlink creation out of specfile and into Makefile
Updated changelog
Put back a missing pathname fix
More missing file relocation changes
re-enabled code to make symlinks for moved files.
Fixed something in the symbolic link creation which Horms mentioned.
Fixed a few more test-only path errors, and marked a test as R2-only
More file location tweaks - mostly GUI-related.
Merge from upstream.
hg: upstream merge
Fixed some link code which shouldn't have been conditional on building the mgmt code
Bugzilla 1657: Speed up BasicSanityCheck and also make logging inheritance more uniform.
hg: merge from 'dev' upstream
LF bug # 1662 - massive heartbeat specfile update - to make it more usable
branch merge on configure.in
hg: merge from upstream
hg: Merged changes in from upstream + merged in some changes in configure.in. Everything else was unchanged by me.
Merged in a patch with one from Dejan to make RPMs build a little better
Added a script for creating the configuration for a single node
hg - branch merge from upstream ('dev')
minor: CTS: changed the code to wait for nodes to come up before we try and get uuid list
bug impact: testing only; fix risk: minimal - raised timeouts in CTS to account for a slow machine I own
LF bug 1762: ComponentFail test results in: Resource ocf::IPaddr:rsc_ibm1 appears to be active on 2 nodes. IMPACT: critical; Risk: minimal
LF bug 1690: there should be a tool to audit (validate) node names in constraints among other things...
Andrew Beekhof (8156):
Rewrite so that admin clients use IPC to talk to the CRMd instead of
Fixed a couple of silly mistakes.
Added admin_cib_test.sh and all a few bug fixes required to make it work.
- addition of memory deallocation code which required a little bit of
- general cleanups, removing unused code, irrelevant comments etc
- some extra memory deallocation
Conversion to the new CIB Schema as suggested by lmb on the ha-dev list.
- factored out some common code in cib/cibprimatives.c
- fixed memory leak/seg fault bug by following my own instructions in
- removed unnecessary call to ha_strdup
* A description of the states, inputs and actions used as part of the
* Added some explanitory text
+ crmd_dsa.dot can be used to generate a graphical representation of
Moved to an simplified/less-redundant and therfore easier to use message
* Converted the existing functionality to be driven by a Finite State Machine.
Improved the information in log messages.
Fixed the startup proceedure.
Corrected minor "include" slip up
* Fixed control of client sub-systems
Now that we are spawned by the CRMd, the loss of comms should be fatal.
Minor formatting and documentation changes
* Minor format changes
* Improved the reference generator to include human readable elements
Updated function prototype for the reference generator
Disabled some unduly verbose logging in stable parts of the code.
* Rejigged the script so that fsa_inputs.dot produces a "graph" in which
Election, Join Protocol, Message Processing
Removed overly verbose logging
Dont hang on to pointers to old IPC_Sockets after they have disconeected.
Miscelaneous cleanups
Dont update the timstamp on save. This is so that they are preserved when
* Added the "bump" operation which modifies a simple generation counter and
General cleaups
* CRM clients now treat a failed connection to the CRMd as fatal
* Dont set xml properties if the value is empty
Added a register flag to the subsystem struct
* Changes required for CIB calculation and distribution
* Uppped some warnings to errors
Formatting
* Update of the status section is now correctly done as a CIB_OP_MODIFY for
* Moved the status pieces to the end of the file
* Logging changes
Logging change
Removed inappropriate action
* Filter out inappropriate messages
* Deal with the result of "store" operations correctly (make sure they get
Finally made the CRM operations into #defines
* Perform an ACK for accepted clients
* Make the CRMd quieter during normal operations
fix install warning
debug for lmb
Cut down the chatter during normal operations
Be a little more graceful about disconnecting clients.
* Recieve followup CCM events
Stop lmb bitchin about the word FIFO :)
Formatting of long lines
Followed suggestion from http://kernel.dk/
Assume the caller is an idiot and check /everything/ (Fix segfault)
Allow use of ctrl-c during development
Make the debug statements dependant on XML_TRACE being defined (off by default)
shhhhh
* Less but better logging
* Changed A_CIB_DISTRIBUTE -> A_CIB_BUMPGEN to reflect the new symatics of
* s/fifo/socket/g
* Moved some common strings to be #defines
Only log on failure
Only log on failure (timers)
Altered log levels to be more appropriate
Subsystems with PIDs <= 0 are considered stopped
* s/recieve/receive/g
Correct a dumb mistake (check that we ARE the DC, not that we arent)
* Dumb down the CIB to make it more flexible (forwards and backawards
- Dont log CRM-level heartbeats anymore
- Removed duplicate #defines
- Removed uneeded debugging
- Removed old code now that uname is a permanent feature of HA
Fix an error in the election logic by which it was possible for everyone
* Formatting
+ Log election events
+ Formatting
* Preparation for the CIB to be aggregated into the CRMd process
A better place to put #defines
Link against the new cib library
Use the new and improved CIB interface functions
+ Roll the CIB into the CRMd process (but keep them as distinct as possible)
+ Fix a shutdown problem whereby the stop command would fail if the subsystem
Moved the (cleaned up) "interface" to the CIB here
Dont use "default" in switches so that the compiler will tell us if an
CIB Cleanup:
This is a better place for this information
- Moved common/crm.h common/xmltags.h common/xmlvalues.h to the include
Receive the arguments in the right order
+ Use our own functions for reading xml from files and strings. This way we
Added some new error codes
A better and more flexible way of maintaining the CIB
+ Added some useful functions for setting and getting xml attributes in
Mainly just collapsed the *_op tags into operation.
Revert to regular updates for the status section. This will probably need to
- Remove old crud
verbose has restricted values
Use the better xml util functions
Empty out clearly unused crud
Try and deal with the fact that shutdown_escalation_timmer wont actually
- CCM uname is here to stay... no need for #ifdefs
+ Placeholder for CCing the TE on events
Messages are now saved where they are produced (which avoids having to
- Integrated CIB is here to stay, get rid of ifdefs
No need for this anymore
Dont be so extreme in reaction to welcome messages frrom other DCs
Make the new cl_malloc header available
+ Use the new location for libhbclient.la
First take at adding LRM support.
People in glass houses shouldnt throw stones...
+ Split out the starting and stopping of the election timer
+ More informative error messages
* Fix errors in the calculation of the expected election winner
Return the calculated validity of the CIB since we bothered to figure it out.
* Store and "Replace all" now take the whole CIB verbatum
Just tidying a few things up
* Rename fsa_lrm_connection to something shorter
Rename fsa_cluster_connection to something shorter (fsa_cluster_conn)
Missed one
Some much needed documentation on XML utility functions uteful for the CRM
*** empty log message ***
More LRM sketches... some of it even compiles
Get rid of a redundant parameter to copy_xml_node_recursive() that was
Ok, just about there. Remaining tasks are to receive the results via the
Distribute the CIB as a replace operation rather than a STORE
Remove an unsed exec so that make -jN, N>1 will succeed
Or just specifying the library without the full directory seems to work too
Formatting
++ Finalize LRM support
Use defines... thats what they're there for
I thought i fixed this already...
Be more informative about exactly what it is we cant find.
+ Creative formatting (with probably a few bug fixes hiding in there)
Actually invoke the LRM functions
Default to TRUE for the moment. Generation counters are checked
Invoke A_UPDATE_NODESTATUS and A_LRM_EVENT at various points
Change "return -1" for the child process to exit(-1)
Commit on behalf of sunjd...
Convience function for getting ints into XML
Allow clients to query for the contents of LRM (supported and active resources).
Start conforming to lmb's schema
Updates based on IRC discussions with lmb.
Makefile for CRM headers
Sick of checking for NULL every 5 seconds
Make setting an attribute to NULL equivalent to removing the attribute
Updating the properties of <XML_CIB_TAG_STATE>s must be handled specially
New helper function
Rejig the cib updates from the LRM so that the same function can be used by
Produce CIB updates for <node_state> based on CCM events (updates are only
Use ordered messages for everything - for the moment. Specific messages
Update the join protocol based on the fallout from the last few days of
Split up the second graph so that it will be small enough for firebird not to
* Implement the shutdown process as described at
* Debug the updated join protocol
* Re-jig to be more in keeping with the original design - ie. return control
Further changes to allow the next-gen pieces to compile on MacOS X
The initial phases of the Policy Engine's processing.
Split out rsc_to_node and rsc_to_attr (aka. rsc_to_node_with(out)_attr(s) )
Actually make some stuff work. Currently:
dev checkpoint
dev checkpoint
Missed a file
break, fix, break, fix...
* detect shutdowns also
* Avoid coloring un-runnable resources (due to unclean parent nodes) and
Finally get rid of those stupid macros and do it properly
Fix the CRM_DEBUG group of macros
Fix a memory leak picked up by kvn
re-jig in preparation for ordering constraints
Implement order processing algorithm
Ordering constraints appear to be working... commit before i break them again
Incorporate shutdown and stonith actions
Create the CRM wrapper for the PE so that it can be invoked by the CRMd.
Remove some very ugly global variables
General cleanup
Some helpful comments
Try and re-use common functionality
Move soem definitions from crm/pengine/pengine.h to include/crm/crm.h where
Initial crack at a transitioner
Missed the function prototype
Fill in the noinst_HEADERS target so that make rpm will succeed
Integrate the TE and PE with the CRMd.
Mods, fixes, refinements to allow a resource to be "started" in a
More closely simulate what the LRM would do so I can remove a hack in the TE
Bump the CIBs generation counter /before/ CC'ing the TE
Bunch of fixes for things that showed up when moving the "hello" messages
Move the "hello" message into the proper crm_message format.
Streamline the TE / PE startup
Make the top-level clean target recursive like all the others
Do not treat replies to CIB ops as errors
Add the correct version of the constraint to the correct resource
Code consolidation. Processing information.
Make sure the crm dir is also processed if necessary
Re-set all node states to "down" after an election, the join protocol
precautionary change... possibly were not having siblings coppied
Convert the cached CCM data to use GHashTables to speedup some operations.
Dont die if someone passes in an empty update
Identify the start of the search when the search fails
Only send an update to the CIB when there is actually anything to update.
Print the full transition grph on stdout (for passing to the tengine test prog
If an action is optional, it doesnt matter if it cant be started.
Unpack resource status correctly.
Added a test prgram for the transitioner.
Ignore optional actions until the first non-optional action is processed.
Dont log the whole IPC message anymore.
Only take action on the results of the last command sent to the PE.
Leave the intelligence to the crmd.
Fix up scheduling of STONITHs, make sure "stops" that will fail dont
Revert an overzealous cut.
Formatting and the ability for the pretend LRM to fail aoocasionally
Rename action->replaced_by_stonith to be more flexible.
Some macros for switching the CRM to vanilla malloc (for mtrace)
Reuse the xml utility function for parsing XML
Reuse the xml util function for parsing XML stings
Make sure xml buffers are cleaned up
Cleanup and use the new malloc macros
Use the new malloc macros
Use the new malloc macros
The other case where a node is no longer unclean
Use g_str_hash instead of g_direct_hash to allow us to find nodes without
Improved running information and warnings
Tweak the priority of actions, remove some redundant ones.
Be more selective about taking actions for PE and TE messages.
Re-enable automatic starting of subsystems
Make sure the "no color" is added all resources.
Handle NULL objects in the color_t matching function.
Move CIB input/output logging to a more central location.
* Resolve issues causing delays in the join protocol - incorrect usage of
Split out lrm functionality... its was getting to big/interetsing
Move all the LRM related code to lrm.c
Get the node information as assigned to the action (since rsc->color may not
Formatting
Formatting and some debug
Increase timeout during development
Tweak the action order so that the shutdown req comes after we have
Make sure the update from the last node to integrate isnt lost.
Keep track of our DC (when in slave mode). Add some error checks using this
Remove some crud, replace with strings that are actually used :)
Cleanups
Fix up a silly typo, and a couple of missing entries
Dont filter out all replies by default... the TE needs soem of them
Misc debug changes
Misc loggin changes
Revert to using the real libxml unlink function as the wrapper appears
Make use of the common ipc message function
Reset the join status on shutdown
Misc logging changes
Make sure the new .h is included in the rpm
Ignore all replies, we dont care about them yet.
Misc logging and formatting changes
Remove obsolete code
Put headers that might be used more widely in the appropriate place
Re-org including
Use the new common headers
Use the new common headers
Clean up the sorry excuse for macros that were being used for logging.
Add missing copyright notices
Switch to GLists instead of GSLists
Remove early debug macros
Make lrm wrapper functions more manageable
Be more intelligent about the processing of CIB answers
Get rid of an unnecessary typedef
Use complete path for local includes
Initial pass at cleaning up the PE and avoiding monolithic .c files.
Dont always compile in mtrace
This update probably undoes some of msoffen's changes, but as indicated, the
Test script for the Policy Engine
Inputs and expected results for Simple PE tests
"node" --> "member"
Use more appropriate #define name
Make sure new_con->node_list_rh is always initialized
Diff args
make sure the lower priority resource stops before
Tests for rsc_to_rsc constraints
Test bad data
Add new tests
Fix errors resulting from bad inputs
Fix empty transition graph: test rsc_rsc5
Clean up and consolidate some of the object creations
Add "only" as a modifier. This allows us to more easily restrict a resource
Implement the new location constraint configuration
Make the "replace" idea more generic
Fix error in logging calculation - valid messages werent being logged
Pull out more of the fake LRM - do the query for supported RAs
Avoid sending two copies of CIB updates to the TE in the case that the TE
Actions are optional by default and marked otherwise as necessary
No longer required
Slight variation on horms' approach which also allows for a centrally
Slight variation on horms' approach that allows for a cetral definition
Updated regression tests now that stonith (in response to node and
Add a tag for the new config section
Moral of the story: pay attention to the output from commits
Add the CRM config section to the CIB
Fix a switch that was behaving oddly
Observe contents of CRM config section in the CIB if present
Support STONITH as a result of failed nodes and resource stops (fencing
Support the new node status fields
Fix a conflict with configure.in - didnt change all cases last time - appologies
Try and bring the DTD inline with reality
Change tags and attributes to match their designed names
Consolidate where the merging of node weights is performed
Implement the one true algorithm
Closed a memory leak
Some random updates so lmb doesnt complain about things that are already "fixed" :)
Make sure it compiles...
Add CRM version string
Add test cases for the rsc_location (running|not_running) constraints
Add a function for comparing version strings
Send the CRM version string as part of the vote
Some new stonith tests
Stonith ops as resource actions
Testing updates
Support STONITH resource agents with a bit of trickery
Make tengine.c compile again
Split the admin client into control and CIB parts
A couple of new xml tags
Fix the impl of get_crm_log_level()
Formatting
Fix a potential shadow variable
Cleanups and refactoring
Ooops, copyright notices...
Potentially do something special on erase (like regenerate the status section)
Send ID and UNAME in status section updates (allows the PE to use either)
Switch to the thread safe version of ctime() - makes a
Add in break
Avoid use of deprecated diff option
Remove unneeded log commands
Use UNAME instead of ID (for now at least)
Make sure all cases have uname and id specified for <node_state> tags
General trace information
New README
Some new functions and a couple of signal aliases (for debugging)
Ooops, remove some work in progress stuff
Move logging to somewhere we'll have write access to
Change soem logging around
Preserve the status section during an erase op
Find the section object for the new constraint types
Remove this test... it doesnt make sense anymore
Some changes so that the rpm builds when --enable-crm is specified
Make sure the LRM updates are in place before we ACK the join
Add in an error message
Interpret the version comparision correctly
Some extra conditions to avoid those nasty NULL pointers
Split some code out into a function so it can be reused
Host_from isnt always set
Provide some real option processing
Remove a duplicate definition
New utility function for adding a node entry to the CIB
Make sure --enable-crm compiles against the new LRM API
Move some logging around
Build LRM event updates correctly (an extra node_state tag was being added)
Allow the IP addresses for the resourvces to be easily changed
Put headers in the right place
Update the README to reflect current project realities.
Install the other parts of the tests as well
Only declare a CIB update if we are the DC
Wrong test
If a no-vote condition is met, then we loose the election
Move the "heartbeat died on us" code to the destroy function where it belongs
Avoid init script joy, just use -k
Switch to using the root user for the tests (by default)
Dont log this message if we are in test mode (when the stream will be NULL)
Dont create nodes, allow the CRM to do that (for now) so the UUIDs are correct
Provide information on how many nodes have joined the CRM
Make sure it will always be the right value
Fix comparision
Make sure all shutdowns floow the same code path.
Disable the erase op until i think about it more
Make sure node and node_state entries have a correct UUID
Disable the erase command for now, it is un-needed anyway
New attribute
Gather the status of active resources from the LRM correctly
Move remaining uses of id=uname to UUID
Consolidate debugging into a single location for easy reporting.
Changes required for 2-node testing to complete:
Make sure the runtime lrm dir is created (I still think its the wrong place)
Make mention of some people's past issues with syslog.
Some new defines
Clean up xml->children calls and filter unexpected child nodes
Use the new library functions (oops)
Update testcase to reflect changes made for alpha2
TEngine 2.0
These files moved but I forgot to add them before committing.
These files are obsolete
Ok, that was *really* sloppy. 2 more files
Fix a thinko spotted by lmb.
Disable the pid check and trust heartbeat to do the right thing :)
Bump the version number
Make sure the timers are stopped after they complete.
Formatting
NULL events arent just for testing
Continue loops safely (Ie. so they actually exit)
Continue the loop safely
Fix accounting error
At the end of the integration phase the PE needs to be invoked
Default to some really verbose logging
Memory leak fixing
Only mark the action not ready to initiate if there are some inputs
Address various issues raised by BEAM. Mostly:
Treat resources on a failed node as if the resource itself had failed.
Handle node failures like resource failures (observe the ignore|block|fence
The expected outputs and the ammended regression test file
Incorporate malloc feedback
Sketches of the 3node tests - never been run
Do the quickfire start claimed by the previous log message
Avoid using g_list_length() in the list iterator, occasional weirdness was
Restructure the iterator so that normal loop instructions behave as expected
Add the config section
Get the config section from the place specified in the CIB
Recoverable, so only a warning
Drop the debug level of this message
Make sure memory returned from dump_xml_formatted() is from the same pool
Formatting
Name change, seems like it might be likely to be used elsewhere
Updated test suite for TE2.0
Lots of changes...
Delete then copy, copy then delete... I wonder which is right.
Pass logfile in to testutils.pl
Create the crm_config in the new/correct place
Give some feedback when a client connection cant be made
Remove PID check... trust the caller
Actually assign the result to the buffer (which we then try to print out!)
Add two functions for the new join protocol
Provide more and better details about update failures
Stub for compare_cib_generation()
Make sure memory is freed when re-setting attributes
Move cleanup code into a common function
Prepare for config driven timeouts
Remove 2 unused hacks from actions in the transition graph
Make a huge mess of the CRM (aka. start to rewrite the join protocol)
No longer existed on the server side, not sure what it even did
Implement a half-assed generation check
Allow the DC to retrieve a remote CIB
Switch to macros to make my life easier with the params
Determine if the DC answer was a nack or an ack
Extra logging
Some extra cases for join related messages
Handle the CRM_OP_RETRIVE_CIB case correctly
Remove some outdated OPs, add a new one
Use the new inplace macros
Generate these so they can be moved (and still find the right helper scripts)
Implement the rest of the new join protocol and make sure it works for 1node.sh
Now generated from .in files
Oops, missed one.
Obsolete - split into client and dc parts
Better logging
Remove logging
Initial stages of 2node testing passes
BEAM Fix (printing NULL)
BEAM Fix (dereferencing NULL)
Whitespace is wonderful
CIB Ops need to occur first so that the RETRIEVE_CIB op can complete and
We shouldnt we sending events to the TE in S_INTEGRATION (and possibly not
The join has not been finalized at this point... make sure we come back here.
BEAM Fix (Printing NULL)
Not 100% sure this is required but it is consistent will all other DC comms
Shhhh...
Work with copies of the messages so that our modifications dont affect the real thing
block comment instead of #if 0 so it compiles on all machines
In S_FINALIZE_JOIN we are not in a transition, there is therefor no point
Because the debug logs just werent big enough
Remove duplicate "else if"
Make sure the expected state is populated for every node that goes through
An insurance policy, allow the slaves to re-announce themselves if the DC
Turn up the default debug level
Put in a trtansition counter to help know where events are comming from.
The need for the sleeps has been fixed.
The PE should also be triggered in this case
- Clean up the tests so that even their creator understands them again
Clean up the use of timers in the TE
Re-jig some helper functions
Split up the subsystems .c file
Set a value for the finalization timer... otherwise it wont run.
This action should only be run as a result of I_STARTUP... period.
BEAM Fix (I think): Printing NULL.
The integration timer is started at the start, and stopped at the end...
Add some missing asserts
Rewrite this test - it now passes.
Add a timeout for the case when the CRM doesnt respond
Use the new message identifier
Feed node leaving events back into the FSA
Delay "started" until we have various pieces of data.
Handle a LRM that might not have started yet
Dont route messages that didnt come from IPC or HA.
Pass data around as a structure that also tracks where the input came on
Print in color so I can tell whats going on
Add the same "no response" timer to cibadmin.
BEAM Bug: Printing NULL
BEAM Bugs: Printing NULL, Freeing unalloc'd memory.
Various tweaks
Re-work such that inputs are read from a queue and are always available with
Back out an overzealous BEAM fix.
Support the case where the CCM hasnt started up sufficiently to connect yet.
Support the case where the LRM may not have started up suffiently yet.
Store deffferred messages at the end of the queue.
Debugging.
Factor out copying a CCM node
Allow tests to run for multiple iterations
Make fsa_actions available to the FSA.
Add a couple of input varients including stalling the fsa until further inputs
Dont vote if we're starting up
Unpack the first message earlier in the loop
A_TE_CANCEL will now be scheduled from in the FSA for these cases.
Formatting
Join offers are sent ordered again (thanks gshi) and tracked so that we can
Implement the two exit functions
Fix Typos
Dont loose inputs that come as return values from fsa functions.
A function for stopping all running resources
Clean up logging and do a better check for running timers.
Be less strict on allowing these inputs into the FSA (which we can do now that
Use the I_SUCCESS replacements
I_SUCCESS is now deprecated
Results of "internal" calls to the CIB shouldnt be feed back into the FSA
Fix transition logging
Stop all local resources before exiting - do now so that a DC might hear us.
Split I_SUCCESS up. Clean up the inputs a little.
Ignore resource entries that arent ours
Various tweaks
Starting a started subsystem is not an error
Move timers into FSA state transition code to be sure all cases are caught.
Make sure the DC has its value for XML_CIB_ATTR_CRMDSTATE set correctly.
Move declarations to the top for Solaris compaitibility.
Add the S_STARTING state to make startup a little cleaner (and dfferentiate
Add the ability to set CRM options
Make the message more informational and more important.
Take notice of LRM return codes.
Fix unclean calculations
Supply expected state as a separate parameter instead of infering it
Add verbosity to the string searching
These attributes should never make it into the CIB
Make failure tests easier to manage
Correct the logic for nodes that drop out of the CCM
Treat C_FSA_INTERNAL data as XML
Rename some callback and dispatch methods that were anoying me
Add failback tests
Put an int rather than a string in the LRM operation user_data field
Make sure the join hashtables are always initialized at the appropriate time
A little overzealous going into S_RECOVERY, tone it down a little
Update test cases
Handle dates with days < 10
Support adding rsc_dependancy and rsc_order to the CIB
Make sure all constraint types are listed
Remove obsolete code
Add PE plugin "architecture"
Initial implementation of the group plugin
Second take at the group plugin - this one stands a chance of running
Fix the direction of unpacked ordering constraints
The resource itself is freed in common_free()
Get variant data safely
Hideously verbose debug info
Get variant data safely
Add tests for rsc_order and the group plugin
Add the ability to find resources in complex PE objects
Add a group-group constraint test
Convert the group plugin to use pseudo events
The correct place for this action is do_ccm_update_cache() not the callback
These should not be fatal
Alternative and possibly working method of checking a child left us.
Alternative and possibly working way of determining if a child has left us.
Also test the TE with the results of the PE tests
Tweak a few macros
Make the logging more useful - log the whole action
Minor logging improvements
Allow an xml input to be specified on the commandline
Move timer declarations to init step.
Make rsc_order a little more flexible and possibly intuitive.
Fix the creation of correct ordering constraints
Set an id for each synapse for clarity
Remove unused counters. They're defined by the slist_iter macro now.
These counters are defined by the slist_iter macro now.
Free message data only after being sent.
Dont assume any step of the signup with the CCM wont fail.
Disconnecting from the cstatus callbacks is no longer appropriate.
Make sure we invoke A_SHUTDOWN before we disconenct from the LRM - A_SHUTDOWN
80 character lines
Allow stop dependancies too.
Kill processes as ADMIN_USER
Remove obsolete code from #if 0 blocks
Raise the input AND invoke the FSA... forgot the last bit :)
Pending actions while moving to the S_IDLE state are soemthign to worry about
Losing a subsystem in an error that the FSA needs to know about
Pass per-action timers to the TE
Pass timeouts through to the LRM for actions
Update with the latest info about complex PE objects
stuff
Tabify
When started using a HA respawn directive, HA might get to wait() on our
Tweak the exec'ing of subsystems to allow for argument passing
Allow the TE to generate coredumps
Allow the PE to generate core dumps
action->node might be NULL. for the moment just log a warning.
tweaks
Initial cut of the API for the new standalone CIB.
Some misc cleanups
Some things just became "common" again - used by the CIB too.
API for the new standalone CIB
Misc.
Remove a bunch of stuff now in lib/crm/cib
Some thing sjust became common again
Start converting the cibadmin tool
New library location
Start converting to the new CIB API
Detect the native UUID implementation present in recent Darwin releases
Build: Header-file related cleanup
Use the functions gshi just committed
Use the library functions since they're available
Allow the CRM library(s) to be compiled
Support the other CIB actions
Cleanup up the way we decide who should do what and when (delegated calls,
Implement "erase"
Yeah that looks wrong...
New test for the CIB.
Function proto changed
Allow certain commands to be sent to specific hosts
Some extra fields and error codes
Particularly in the CIB its helpful to have the GCHSource. Everyone
Dont abort if its the lrmd thats respawning
Implement CIB update notifications and a sample client
Use the CIB API for CIB related commands
Formatting
Make sure our CIB is set to master/slave mode as appropriate
Use the CIB notification mechanism to process updates.
Convert the TE to use the CIB notification API.
Dont wait for a callback if we specified cib_discard_reply.
The file that should have been checked in yesterday
Tweak log entries
Tabify and clean out main.c, particularly do_init()
Update the CIB using the API not by sending a message to the DC.
Precautionary meausre
Create a new log level just for inner XML functions
We already have the fragment, no need to go looking for it.
Log to the correct location
Using cib_discard_reply makes the call asyncronous - actually you wont get
Make sure we only set our own CIB service into R/O mode
Make sure to update the CIB with the new node's current LRM status
Fix the shutdown handling
Reverse the previous "fix" after realizes where the true error was...
dont return the fragment but the CIB inside it
Make it legal not to specify somewhere to put the output in... maybe they
Increase some timeouts.
Fix a stupid malloc bug (perhaps using the correct type is a good idea?)
Rename the file handle.
Make sure the_update is assigned at the point we pass it to pre_notify().
Not technically an error, and I'd like to know what it was since it shouldnt happen.
Come back to this...
Make sure we get the latest known state of all crmd processes upon becomming DC
Changes to INCCM and JOINSTATE can also tell us that a node has left.
Fix a logic mug in the detection of failed nodes.
Use a pre-constructed DC heartbeat message
Exit if the CIB goes away.
Make sure CLEAR_* only takes effect once.
Add the ability for the DC to trigger an election from a slave node
Remove irrlevant code
Fix for bugzilla entry #190 - Force a recalc.
Misc fixes
Fix for bugzilla entry #46
Bugzilla entry #62 - Pass incarnation details to the RA as resource args.
Some extra defines
Remove all the shutdown escalation tasks - they're no longer needed
Formatting
Tweak the usage of extra attributes for resources
Bugzilla #116 : Add extra attributes for use in generation counter.
Optionally suppress and pretty-print compiler output - so its more readable
Make sure ccdv.c is part of the distribution
Use a manual rule for ccdv.c so that ccdv is always available and uptodate
Reduce the use of literal strings, use #define's instead so that they may be
Set XML_ATTR_DC_UUID after each election and use it allow constraint
Make calls to find_xml_node() a little quieter when possible
Add test cases for is_dc expression
Spring cleaning
Add support for CIB feature revision checking
Add support for S_HALT : a state to go into when things get messy.
Support CIB feature revision checking
Bring the DTD (more) uptodate
Appologies for the top-level commit, one change necessitated another which
define crm_log_message() which filters based on log level before calling
Convert some macros to real functions
Replace some macros that were causing trouble for newer gcc versions
Prepare for libxml2 to be replaced with HA_Message
Prepare for libxml2 to be replaced with HA_Message structs
Prepare for libxml2 to be replaced with HA_Message structs
Comment out a call to free_xml() that is causing trouble - not that i can
Interpret the result of lrm_rcvmsg() correctly and feed the result into the FSA
Turn back on the freeing of XML data - committing with it turned it off was an oversite
Commit a temporary work-around for a de-allocation bug.
This should not be required now that the DC signs up to itself
Use the more appropriate form of v*sprintf()
Dont invoke the notify callback when there is no message
Log to DEVEL_DIR"/cibmon.(debug|log)"
Include a version of recursive copy function for testing
Check for core files before declaring the test successful
Required for vasprintf()
New logging helper
Add a macro to make switching between libxml2 and FT_STRUCT easier
dump_xml_formatted() now makes sure the whole message is logged
Always print 1 space after a newline - syslog barfs otherwise if it is a '<'
Reduce the log level where it was unnecessarily high
Reduce the log level where it was unnecessarily high
Reduce the log level where it was unnecessarily high
Dont use logfile/debug file directives
Work around for the fact the stonithd wont start properly
I _think_ this is correct, but the tests dont run because of the CCM so I cant check.
Remove the extra logging "features" involving files.
Not needed, the DC signs up just like everyone else now
Use the new set_uuid()
Small test file - currently for testing logging.
*** empty log message ***
Make sure the node's UUID is available to the TE
Specify a log level not a boolean
boolean -> log level
Update the CIB correctly after a successful stonith op
Fix a call to print_state()
get_cib_copy() unwraps the fragment for us and we can then use get_object_root()
Create a common logging init function for CRM pieces
Remove unneeded redirection for fsa_actions. Will also allow do_exit() to
After exit, there's nothing more to do / that we _should_ do.
Send the command if we're _NOT_ testing, instead of if we _ARE_. Doh.
Logging error
Missed a bit
Almost left behind
Update soem timings
Fix BEAM bugs
BEAM Bug fixes
BEAM Bug fixes
BEAM Bugs
Identify XML fragments with the function that asked them to be logged
Test for correct handling of complex messages (as produced by the CRM).
BEAM Fixes
BEAM Fixes
Misc fixes
Use the global definitions
Misc updates
HAVAR_DIR is configurable - dont hard code it
Macro -> function
Fix some parts of the FT_STRUCT implementation
Close memory leak.
Make sure the source is also recorded
Minor fixes after switching libxml2 back on.
Pass the _un_ formatted version to ha_msg_add(). Duh!
Make it easier to do the right thing
Send live.dot entries via syslog where they can be extracted later (instead
Slightly more up to date
BEAM Fixes - printing NULL
Fix the extraction of log messages related to the test
Only process the nodes we have
* revision_check() Make sure revision is re-gotten if we update it.
Pre-check for NULLs
Enable coredumps
Pre-check for NULL XML objects
No longer tollerate NULL XML objects as a general rule - makes it easier to
Check for a NULL generation object (possible if the CIB disconnects)
Check for null XML objects
Use the new get_message_xml() which works for libxml2 and FT_STRUCT
Various macros -> functions
Pre-check for NULL XML objects
Use the generic get_message_xml() call
Unneeded strdup()s
Some extra validation and cleanups
Make our own copy, incase we loose the HA connection. Required? Cant hurt
A correct "revision_check()" fix.
Use generic code (works for libxml2 and FT_STRUCT)
Use more generic code
Remove unused code.
Fix 2 potential memory leaks
Update some timings
Enable core.pid
By default skip F_XML_TAGNAME and F_XML_PARENT as "properties". They're special
Use the generic functions for adding/obtaining XML from/to messages
Differentiate between when things are _really_ bad, and when they are
Switch to time based aborts when searching for strings
Pre-check for NULLs
Formatting
Only some asserts should always be fatal
Not all asserts should be active all the time
Dont free the message until after we're done printing out the fields we
Dont add the entire CIB to every query (its a duplicate anyway since the
BEAM Fixes
Some extra logging
Pre-check for NULLs
Misc updates
Its a const...
Dont enable cl_malloc_forced_for_glib() by default - causes something in the CCM somewhow
No default node names
Enable FT_STRUCT version by default
Add a validation check before free's
Add an extra check to 1node.sh
The script that grabs relevant log entries
Fix resource and constraint creation so that they work with ssh and su
hacluster should be the default crmd user
Just "su" works fine, no need for "su -"
Make sure both timers are stopped when we release the DC
Correction to the transition graph
Fix construction of query fragments
Make activation processing more clear and logical
Use syslog for all logs - append a special identifier so they can be
Use common log init code
Use common log-init code
BEAM fixes
Make sure we're in the right place for core files if the parent enabled them
BEAM fixes and asserts
Rely on the parent to enable core files if appropriate
Remove unused code
If we are shutting down, then signal to HA that we dont want to be started
Timers should be stopped in all 3 cases
These tests now detect if the crmd was started from ha.cf
Theoretical fix for the CIB not being active "enough" when the crmd starts
Disable LRM agent info in the CIB and the checking for it in the PE
_All_ inputs are now raised through register_fsa_input()
Only process 1 HA message at a time.
Clean up shutdown processing. Add R_STAYDOWN and observe it correctly.
Remove the cur_state intermediate variable... no reason for it.
The extra register entry
Make a copy of the XML_ATTR_DC_UUID just in case our caller free's the CIB
Consider unseen nodes as STONITH candidates
Fix "-escaping
Downgrade an ERROR to a WARNING
When stalling the FSA, add the current action back right away so that it
Deprecate the DC heartbeats
Improved asserts
If this is a development build, allow the CRMd to be "killed" rather than
Make sure the timers are set appropriately relative to one another
Missed a case where the other side could loose (no need to keep checking
Clean up the usage of the DC_TIMER so that it is both started and stopped
A new command to simulate failure in the CRM (only works if the CRM was
Simplify the logic a little
Fix the extraction of RA arguments
Definitivly identify two important timers
Use the correct name for name-value pairs as defined in the DTD
Fast track shutdowns that are requested before we have completed startup
Fix the naming of nv pairs
Use stop_node where possible
Add the ability to disable CIB writes to disk
Comment out a debug call that can produce a _lot_ of logs
Add logging to uniquly identify which nodes got which updates
Include CIB generation details in the logs
Improved logging
Add CIB generation details to notify messages
Prevent segfault if we are shutting down but havent connected to the CIB yet
No need for a PID file now... heartbeat looks after us.
Dont think we actually need to make sure everyone else is in slave mode
Clean up some logging
This causes an infinite recovery loop... perhaps its not a good idea? :)
Throw in some extra asserts
Asserts and BEAM fix
BEAM fix?
Sometimes this isnt true... would like to know why.
By default stopping a node involves stopping heartbeat, not just the crmd anymore
Finalize on cib_ok == master, cib_not_master == not master, and everything
Change to the regex and a slight tweak.
By default this really has no need to log anything
Make sure successful updates that need to be sent to the rest of the cluster
Use the new signal handling code.
Just use asserts and be done with it
BEAM {}-fix
BEAM {}-fixes
The add primative never returned failure... fix it.
Remove unused code
BEAM {}-fixes
BEAM {}-fixes
Remove duplicate logs
Keep BEAM happy, might benefit recovery too.
Use the new signal handling code
How on earth did this function correctly?
Break down a few shutdown cases
Use asserts instead
BEAM {}-fixes
How did that get through?
No log message has an inate right to be printed
Start the "where is the DC" timer any time we drop back into S_PENDING
We detected the need to abort... how about actually doing something about it?
Take a tip from the glib guys... include the test that failed in the logs
Actually take notice of the result of extract_event()
BEAM {}-fixes
Downgrade an ERROR to a WARNING
Avoid the use of FALSE as an assert expression now that the assert will
Avoid the use of FALSE as an assert expression now that the assert will
Make sure fsa_our_dc is updated
Variable to avoid calling assert expressions with side effects twice.
Make a BEAM warnign go away
Audit the state of all crmd's in the cluster
Duh! We want to make sure its a "native" resource not a "group" resource :)
These are attributes not operations
Logging tweaks. Rely on the caller to complain.
Treat being evicted as an error as its not expected.
Reduce duplicate logs
Bugzilla #271
Its up to the parent to decide if this is an error
99% of these messages are far to detailed and are really only helpful when debugging
Default to the newly cleaned out LOG_DEBUG (for now, LOG_INFO eventually)
Logging adjustments
Return the top of the XML object created
Remove the parent pointers - instead its a flag to indicate that the blob
Use the new code which removes the blob directly from the parent.
Rewrite to avoid the need to find the blob's parent
Clean up option processing
Return a copy of just the CIB so that we can clean up the whole fragment.
Rejig the way XML is freed
Add an option to clobber the existing CIB
Installing a CIB only has an effect if the CIB process is not running.
Some extra asserts and better logging
Downgrade an ERROR
Do things in a way that we can clean up afterwards
Pre-check for NULL chosen node
There are legitimate cases where this update will fail and does not
Typo
Dont increase the verbosity of child systems
Simplify the logic for starting and stopping timers
Use consistent data for the DC Audit
Require and use a name for connections to the CIB
Fix a election calculation
Reset the color numbers after each calculation
Remove the status entries on startup - we cant trust them.
Ooops
Sometimes there may not be anyone listening to accept the update.
Do not override the default log level
Make sure the DC has reached a stable state too
Use a proper crm assert
We need the container object
Allow the number of nodes being tested to be less than the number in ha.cf
Observe the test preference
Any operation that returns a result other than EXECRA_OK is now treated
Dont accept our own join offer if we have released DC status in the meantime.
This is important information
Prevent being stuck in an infinite loop
Correctly check for _both_ patterns
Dont accept a DC that has since quit.
If we are the DC and get an error, in addition to releasing DC status we also
If we're in recovery mode, we should leave it up to Heartbeat to decide if
Add a tag to identify the source of and update or CIB
Process SIGTERM gracefully.
Use some better terminology as suggested by alanr
Use better terminolofy as suggested by alanr
Do the epoche comparision as an int not a string
alanr's suggested naming
Dont use constant strings
Misc tweakings
Impliment server-side notification filters to cut down on the number of unwanted
Record the last CCM event received
Basically if we ever win an election from now on (given that we didnt vote)
Utility function
Remove unused variables
Start the election only after we released the DC
Do not get into voting matches with nodes that are not in our partition.
Provide better details on why the cluster node is unstable
Remove all blocking-ness from the CRM.
Progressivly throttle back pre and post notifications depending on the queue
Fix logs about logging
Add a new fn() for getting XML from stdin since file2xml() uses tricks that
Ignore any messages from any node not in our CCM.
Even though we are shutting down, we still need to know if there is a DC
Fix reading from stdin
Get ttest working again
Fix a log message
New function
Observe debug and log file directives from heartbeat
Dont try and write to a NULL file pointer
Observe log directives from heartbeat
Doing things in a hurry is always a bad idea
Check how the signon went.
Rename the CCM utility to reflect its new found abilities
Make sure the CIB object is zero'd out if verification fails
If invoked manually and verbosity is asked for, enable stderr logging.
Interpret rc correctly for the case that we specified an async call
The user is unlikely to care - this is what debug is now reserved for.
Catch a mor eimportant error first
The fencing daemon seems to dissappear a lot - dont make this fatal.
Logging when invoked as ttest
Logging improvements
Debug logging to make sure the tests are behaving
Allow the client side to specify a non-default timeout - kinda important :)
Create and use a standard fn() for sending HA messages.
Correct the assert statement
Remove garbage from the Makefile
No need to be so verbose
A bit of a cleanup
Re-emphasis on being in master mode (or not connected at all)
Observe debug level from ha.cf
Change the assert to only check that we are not a CIB slave
The other half of my plan to reduce the number of votes during elections.
Fix potential memory leak in max_generation_from
Pre-check for NULLs
Make sure I_PENDING behaves as I_NOT_DC used to
Record nodes in our CCM membership
The role of DcIPaddrAudit is now performed by PartitionAudit
Check the actual state of the cluster nodes before declaring failure
When running the local version of 1node.sh, support the log format used
Multiple CRM processes logging to the same file as heartbeat? A BadIdea(tm)
Logging tweaks - move messages to a more approriate log level
Logging tweaks
Only call the assert on failure
Only our caller knows if this is a problem
Prevent HA from setting the debug level lower than the compiled in default
These are no longer actually errors - probably not even warnings...
These two blocks were meant to be part of the same.
Use the common send_ip_message() call which also checks the channel.
Indicate when we are asyncronously starting/stopping a node too
SimulStop was actually doing StopOneByOne - use the async version of "stop"
I never did like those function names...
Introduce a new StableTime variable to use instead of DeadTime when waiting
Set deadtime long enough for the CRMd to force-quit
Add a timer to periodically resubmit the shutdown request
Fix the test that checks if the other node is shutting down too
We're already shutting down that this point... no need to redo these commands
Async calls to the CIB can still fail.
Explicitly test NULL variants
Some twaeks to the local version so that it works in BasicSanityCheck
Use alternate methods of determining if the CRMd shut down.
Get rid off stat core files produced with MALLOC_CHECK_=2
Find out what its complaining about
Direct ha_logd output to the test file
BEAM Fixes
Test for 0 - its more reliable.
Observe a bunch of ha.cf options and do it centrally
Misc tweaks that prevent core dirs from being removed
Always use common ipc send code.
Fix a double free - send_via_callback_channel() free's the message
Resolve exacly when we should and should not be voting
Have heartbeat expose all directives as environment variables (with a HA_
Relax the rules for re-announcing nodes - no longer required
s/status/monitor/ for OCF complience
If all nodes are up then the test passed
Better detection of dead channels - check for != IPC_CONNECT instead
Close a rather large memory leak
If the filename is "/dev/null" disable writes
Logging a warning is good enough
Remove excess logging
Close down another memory leak
Use a more flexible test for true/false/on/off/1/0...
Include current options in the DTD
Give the CIB every opportunity to send messages but at the end of the day
Use get_chan_status() to get the very latest channel status
Implement a "symetrical_cluster" option which basically means everything
Stop using imaginary features to process fencing operation results
Prevent matching on our own debug
Make sure all nodes are added to the list when two lists of nodes are
Build the default CIB dymanically
tabs :(
Move the test logic for stop here as it is CM specific (Ie. there is more
Some tests offer an alternative starttime (NOT USED ANYWHERE YET) that
BEAM fix - msg_copy leak (send_ip_message() free's msg_copy for us but only
Logging the message itself as an error achieves nothing
Catch another case where sending a message is pointless but log it as an
Logging tweak specific to the PE
logging cleanup
Reset the to defaults every time
Logging cleanup
Increase the noise when matching actions in the transition graph
If timeout is 0, read until the end of the file
Factor out the post-test audits so they can also be called after the setup
Catch msgfromIPC_noauth() returning NULL with a connected channel
ha_logd should not be started from ha.cf anymore
Dont try sending commands to the CIB if we know its not connected
Consistently use the same #define
Mask out the SIGTERM from heartbeat and only exit when the CRM tells us
Handle IPC_INTR when waiting for synchronous responses from the CIB
Improved ERROR and interaction with the LRM.
No longer needed
This really is an error
Gather info from nodes that pop up
test is "None" the first time around
This is somewhat normal if we're giving up DC status
ssssshhhhh!
Only log the start event if it will actually be runnable
Balance resources throughout the cluster where possible
Differentiate between broadcast requests that dont get answered and
The message itself is not an ERR, just context for it.
The message itself is not an ERR, just context for it.
Required changes for distributing resources evenly in the cluster
Dont abort on new nodes - leave that to the CRMd and the signup process
Allow us to retry sync operations from specific hosts
Prevent a segfault when no node can be found to run the resource on
Make a separate call to obtain the CIB generation
Spelling
Work-around for bugzilla #346 which was causing unncessary elections/
Remove some redundant code and make the log match
Include the return code from the call
Handle the no-op without an error
Define a debug() function that skips printing to the console
Add the RestartOnebyOne test
When shutting down via S_RECOVERY, dont wait for the DC.
Use s/log/debug/ where appropriate
Add the RestartOnebyOne test
The stonithd now accepts the UUID in parsed form
Overzealous lmb
Nothing more to do after a stonith op
Have configure #define HA_LIBDIR and use it
Throw in a -dev assert
I have seen cases where shutdown was delayed due to cibmon... add this so I
Only log CIB mode if it changed
No need for these to be at info now that we know it works.
Be consistent about how these loops work.
3 seconds is a very long time... the lower value should help with 246 too.
Cut down on election logging now that the cause of 346 has been identified.
Remove redundant timestamp
The message itself is supporting data not an error.
Use cl_ macros
Turn off ccm event detail... rely on the ccm to log its own info.
Make sure all calls to obtain messages are not affected by IPC_INTR
Bug 341. Just make sure debug messages turn up on the central node instead.
Dont syncronously wait for child processes to die.
Read all HA messages, _then_ invoke the FSA. This _might_ help reduce
Just log the status section, its the piece most likely to be changing
Multiple partitions is something we want to know about (now that partition
Allow the is_dc check to find DC's that are not idle too
Update the test inputs and outputs to reflect recent changes
Treat successive votes as a single one
Resource attributes are now read from 0+ <instance_attributes> sets.
Define the colocational constraint at a point it will be also used
The total number of incarnations and the number per node is now a resource
Remove superceeded code
Indentation change
Misc logging changes
Add some new #defines
Reflect recent CRM DTD changes
Handle NULL pointers
Separate rule and expression handling into a separate file...
Bug 379.
Actions have their own attributes too
Rule logic is now in rules.c
This is what it _should_ be doing. I think.
Spelling was never my forte... s/symetric/symmetric/
Various logging improvements to make DEBUG less noisey
Get the very latest channel status
Whenever the CCM membership changes, send out a no-op in case its a healing
Twice should be plenty.
Logging changes
Spelling
Make sure the attributes for the resource operation are sent too.
Logging
Spelling
No member called "id"
Small cleanup
When another DC is detected, make sure we start an election NOW!
Use the OCF delay RA as a workaround for #366
BEAM Fixes
Make these proper log messages
BEAM Fixes
Dont loose log messages
Do LRM operation updates to the CIB asyncronously given that the alternative
Use the new manual trigger code for mainloop. It allows a couple of
The crmd really just needs notifications of its own updates (which are now
A more correct BEAM fix
Allow provider to be set for resources
Catch cases where the TE is killed before it can set its signal handler
Dont wait for monitor callbacks... we'll hear soon enough if they fail
LRM uses lowercase
Set the provider
Go back to the IPaddr resource, but this time the OCF version.
Be consistent
Some new defines
Replace the rule "result" with a symbolic +/- INFINITY score.
Remove a bunch of crud from the inputs which had no effect
Admin tools use a different facility
Too noisy
Track how long various actions take and report errors when they exceed
Factorize out the summary so it can be printed when we have too much BadNews
Use the OCF RA
Track down every last blocking call to the CIB and find a way to do it
Cache the results of UUID calls to avoid blocking for HA.
Provider is not manditory
Fix the check for a dissappearing TE
No longer applicable
Logging tweaks
Give IPC clients a chance too
The DTD changed
Misc fixes
Catch unallocated resource actions
Include the node information - not doing so was causing the action to show
When searching for actions, assign actions with no node to the search node.
Differentiate between the different success and failure cases.
This is better done by the TE when the transition completes.
Dont timeout the DC just because _we_ didnt get a speedy reply from our CIB.
Use the built-in Python "stat"
Implement standby and active (un-standby)... no idea if it works yet.
Update the usage info
These logs arent printed anymore
Remove checks for logs that are no longer printed.
Fix unpacking of resources... attributes etc are not manditory
Add Simul{Stop|Start}Lite to speed up the setup for the real tests.
Attach the generation, not the whole CIB!!
Re-enable some logging until gshi can add it to the CCM
Cleanup the calculation of node status
Bad comparision
Nodes due for shutdown should have resources allocatred to them
Only prinbt the warnings on correct occasions
Make sure resources are never scheduled on unclean or shutting down nodes
Add a varient of send_ipc_message() that server's can use so the
Logging fix
Create and use a common weight function that handles "INFINITY"
Remove a loop that only went around once
"Fix" stonith handling
If stonith enabled:
Pseudo ops dont need nodes.
Default log level for ptest
Implement lmb's quorum policies:
No need to update the CIB with cancelled ops
New helper function for obtaining CRM options
Use the quorum_override option when writing status updates and performing
Misc tweaks to catch extra errors
Implement command filtering for functionality that should require quorum
Add the log channel to mainloop
A more appropriate home for the CRM library
Files moved to lib/crm/common
Point to the new home of the common CRM library
All nodes should update their own CIB with CCM and crmd T_STATUS information
Only call cl_log_set_logd_channel_source() if ha_logd usage was enabled
The TE should also override the quorum status
Add a bunch of recovery test cases and make sure they work (with 1 exception)
Enable one more test
s/multi_active/multiple_active/ (its only 3 letters!)
Update the DTD to reflect recent changes
Dont be so verbose at LOG_ERR... the input and output are supporting information
No longer true... the memory is cleaned up in do_update_cib_nodes()
Cut and paste error
Actions are free'd via global_action_list for the moment
Does the same thing really...
This is something we need to know
Send out one last vote, just to be sure.
Ignore successful monitor ops (the first one when it starts)
Heartbeat is the one to decide if we should stay down or not...
Finalize the join (wrt. clients) ASAP and do the rest of the actions after that
The pattern has recently changed
BEAM fix
Make stop the default no_quorum_policy
Make the stdin option work
Logging changes which will also help with 492
consistent logging
Extra function
Make sure the event detail is printed in both cases
Bugfix for 481: timeouts and intervals are passed to RAs in milliseconds
Formatting
Include the current generation, this also changes frequently and is important
There are cases when we shouldnt re-issue the start operation
Reorder action_new() so that we can make sure pre-allocated actions are
Logging
Common functions the CIB and CRMd can use to be consistent
Use the common functions
Use common functions
The function is update _local_ cib... so set the cib_scope_local option
Bring the DTD up-to-date
signedness fixes
Support 'm' and 'min' time units
Some extra defines
New functions
signedness changed
This doesnt look right to me... remove it
Prefix '#' to "automatic" attributes to distinguish them from user defined ones
Simply testing/non-testing logging
Implement a single phase sync operation
Allow a different callback to be invoked for any given CIB invocation
An extra pattern to ignore
This functionality was moved to the CIB client library
Completely overhaul the join protocol
Support the new "i timed out" result from the TE... for now map it to cancel
Factorize the building of CIB updates for operations
Generated attributes are now prefixed by a '#'
Fix a bad merge.
Thankyou BEAM!
Use a common function for creating an identifier for recurring operations
Use the common function for generating idnetifiers for recurring oeprations
Keep BEAM happy
Tweak the filtering of FSa inputs in various states.
Set and unset R_CIB_ASKED and R_HAVE_CIB as appropriate
Various signedness tweaks
TE "pending" updates should be broadcast
Ooops... wrong directory
Oooops... wrong directory
Make sure the join data is initialized before join_offer_all
last_op is the last op sent to the LRM, so set it every time.
Include __FILE__ as a parameter to do_crm_log() so it can be used in grep filters
Use the new allocation macro name
Provide __FILE__ to do_crm_log()
A couple more files using the old crm_malloc macro
Use the new macro name
Remove a cunch of unused code for cases that can no longer happen
Use the new macro name
Use the new macro name
Use the new macro name
Make sure neither the start action (if its ever scheduled) or the stop action
Maybe fix BSC...
Tweak BadNews
Fix possible use before allocation
Avoid re-using the same operation for different stages in the join and
Close a potential memory leak
Rearrange the code so I can print "yes i sent it" messages
Support the new CIB options/flags
Correctly determine node up/down state
Tighter logging
Keep a record of CRMd peer status
Drive the base timer from the value for keepalive
These are used in a number of places...
Clean up R_ flag handing
Remove the reannouce code
Consolidate checking of join phase completion
Use the new macro name
Make it easier to identify some important timers
Fix node up/down calculation
Fix logging
Once we have a DC stop the DC_TIMER... the join will take as long as it takes.
New flags
These weren't supposed to go in yet
Update the pattern
Both timers going off indicate a problem
Implement cib_msg_copy() which strips out various "special" fields that
Make sure the results of any pending CIB operations are discarded once
Use the new uuid code from gshi
Termporary work-around until I get a chance to use the get_last_op(op_type)
Fix for a bug in SimulStop
Logging to track what updates we've seen from the CIB
Only complain after a confirmed stop, rather than a scheduled stop.
No longer required.
Make sure that without quorum and with a no_quorum_policy of stop, that the
In realtive terms, this should ensure that messages from the DC to the TE
Exit if our connection to the CRM dies
Some extra BadNews patterns
So the next condition succeeds
I just wanted to see how often it happened
Pre-check for NULLs
Read and pass start_delay to the CRM
Use a score instead for colocation constraints
Support start_delay and make sure it is normalized into ms like everything else.
Cleanup shutdown sequence
Start the BadNews timer so it catches setup errors
tabs -> 8 spaces
Move the lrm functions next to each other
Remove all uses of op->rsc->*, instead use op->rsc_id and remove any others.
Update I_ERROR with the new shutdown logic
Remove unused code
Clearly we cant shut down resources if the LRM is dead - complain loudly though
Make sure the ordering is ok and unique
Support the new expanded status section from the LRM
Tweak in the sorting.
Missed a header file
Make sure R_SHUTDOWN is always set
New flag
Avoid any possible timing issues, track every command we give to the LRM and
Update most of the test cases:
Use the action's node rather than the supplied one
If the start is pending, then we need to schedule the monitor ops too.
Update a few more test cases
Make sure stops are always marked optional (initially) in this case
Update a couple more testcases
Fix another test input
Logging
Make we are marked runnable when actions depend on us
By default all incarnation actions are optional (and marked manditory later
Test case updates.
Make some macros easier to read
Do not update the CIB with pending STOP operations - they make stopping a
Dont rely on the before and after lists being equivalent... they may be now
Complain if the CIB and CRM disagree on quorum state.
Check we will run the op _before_ we add the resource to the LRM
An extra case
BadNews tweaks
Use the common function
Use rid... its always available in this function
Record the membership instance
Better logging for ignored messages
We want the CCM updates as soon as possible
Record the membership instance in the CIB
Formatting
Check the membership instance seen by the CIB and the CRM match.
These are high priority events
Do this as a proper FSA input
Only OC_EV_MS_NEW_MEMBERSHIP counts as having quorum
Gain a new understanding of what the CCM events mean and update the code
At this point we no longer need the shutdown timer.
Make sure everyone is using the same definition
Two more shortcuts
This is merely noteworthy rather than an error (now that we have a
Formatting
Include the BadNews prefix
We have no need for the rsc field of lrm_op_t.
Never invoke the PE manually, always do it as a result of cancelling the TE
Always confirm cancel ops - the CRMd now uses them to know when it is safe
Make the FSA graph a little more readable
Logging change
Dont wait for the CIB to be activated before setting the new quorum value
Update the number of peers as soon as we get the callback
And the prototype....
Move quorum updating to the callback for flexibility
Extra info for #547
Make sure CCM updates are processed in the order they were supplied
Adjust the order of the audits
Shortcut function
Invert the ordering argument to register_fsa_input_adv()
Only record non-recurring actions in the shutdown list. All others will be
Make sure SimulStart and its variants wait for a stable DC before continuing
Formatting
Duh
No longer required.
English
Fix time profiling of CRM calls
Updated macro - add actions directly to avoid flooding
We dont always need to cancel the TE
Filter out some TE actions
We are in this function as a result of a CIB callback
No longer needed.
Consolidate calls to G_main_set_trigger() by having it automatically set
An extra pattern to look for
Define and use warning levels in addition to error levels
Close two memory leaks.
Reuse the *Lite code.
Two types of stalling, with and without data. With data is possibly a little
Handle the two non-matching cases differently.
Be smarter about how we stall the FSA when invoking the TE.
Eventually use this to find recurring actions
Allow recurring tasks to be sceduled individually of one another.
Fix the creation of recurring actions
Remove irrelevant information from pseudo operations
Process timers in order like everything else.
More appropriate #define names
Cleanup
Add a join counter so we can discard old join responses
Attempt to fix Simul*Lite and have it produce better information when it fails
Only print the PE input if it resulted in an error or warning during processing
New field
Logging changes
Convert to ProcTrack
New key
A new test that (crudely) checks for leaks in memory consumption.
Quiet down logging now that everything is pretty much working
Logging changes
Dispense with the stupid symbolic names for the extra debug levels
BEAM Fixes
Remove support for libxml
An option to prevent exiting so that MallocDebug.app can connect and
Minor memory leaks detected by MallocDebug.app
Macro hell...
Use the redirection macro
We might as well truncate it now since cl_log will do so anyway in due course
Due to the way that the heartbeat process grows until it has sent X messages,
Remove support for libxml2
The free is not required for stack variables... duh.
Print _then_ free
Handle CIB-wide updates a little better
Asked for the wrong section (but at least we know that section of the code works :)
Close a bunch of memory leaks
Shuffle the PE logging down a notch and only use "debug == 1" to log the
More memory leak stuff
Dont loose nodes. Make sure the color always points at the start of the list.
Free the action key if we dont need it (because the action already exists)
Misc tweaks
Fix duplicate filtering
Close a potential memory leak
Use the CIB API directly so that we can inhibit broadcasting the change to
Use the CIB API directly to
If we know the TE isnt in a transition, then we can shutcut the cancelation
What we really want is to just zero out the stats
New action
Helper functions for interacting with the cl_malloc stats
Dont send duplicate aborts for the same unmatched event
Provide proper support for status updates that had a global "section"
Allow differentiation between aborts that the CRM requested and ones we "discovered"
CCM updates do not need to be broadcast
These updates should stay on the local node and not be sent to the master
Misc debugging updates
All the other tests have a capital
Because we never have status entries for nodes the cluster hasnt seen in its
Ask the right object for the StatusCmd key.
Catch this early
CRM logging includes the function name.
Make the logs easier to read
The stop due to quorum case wasnt being logged.
If the last operation we sent to the LRM was a stop, then there is nothing
According to gshi this is no longer correct - though I'm not sure it matters
Use the standard function
This isnt an error
Logging tweak
Grumble dynamic languages grumble
General cleanup of CTS
Fix a rather clumsy use-after-free picked up by debltc on ppc64
Another one...
Streamline detection of failed nodes
General cleanup
This was a DC initiated cancelation
Use a bit in the input register to record if we are in a transition or not
Dont send the LRM status to the CIB twice
There is no need to trigger a transition here - the TE will do that if required
Dont check for leaks if the queue's are not empty - they can produce false positives.
Reconsider a previous change... this is probably a good idea afterall.
We know ahead of time we dont need the timer to start if we are not in
Convience change for when we want to log some specific bits being set/unset
Add a user-defined epoche to the CIB generation
Remove the callback from the hashtable before invoking it. This way
Logging tweak
Try to simplify the FSA mainloop to make it more understandable and easier to
If there are no current actions then there is nothing to store and no reason
Remove a pointless "optimization"
Use a basic FSA so we can wait for pending CIB updates to complete before
Logging
Logging changes
Fix for bugzilla 573/574
Fix CTS message. Stopped nodes dont count.
Logging tweaks
Correction for join requests that arrive at the wrong time.
For for 583: Memory leak in incarnations
Logging reduction
Duh
Code thats not hooked up yet - eventually the TE will use an FSA too.
Fix usage information
Deprecate intermediate level notifications (before and after each object
We changed state to get here - make sure it was necessary.
Use a common util function instead
Additional fix for #583
Misc tweaks
NearQuorumPointTest:
New helper functions
New helper functions (diff'ing XML)
Unused parameter was removed from the function prototype
New util functions for diff'ing XML and printing the results
Remove unused attribues from the stonith incarnation
Logging changes
Reduce logging
No need to dump the CIB because of this.
Logging
Where this function went and how I've been able to compile today is a
Fix for debltc: if the whole update is pruned then there is nothing to add
Instead of using mixture of global variables and passing a million different
Like incarnations, groups should also use "stopped before start" (rather
Remove "schedule_recurring" which was a hack at best
Logging and formatting
BEAM fix
Incarnations...
Complain (and restart the join process via an election) if we try and invoke
New output format for ttest (which now works again)
Changes to make ttest run, and log informative things
Make sure the graph always has a transition timeout
Add a couple more regression tests
Missing element
Fix a memory leak
Test cases for the rsc_location equivalent of rsc_colocation
The CRMd used to assume that clients stay around but the crmadmin tool doesnt.
Formatting
Add a tool for generating and applying XML diffs in the same way the CRM/CIB does
s/CRM_OP_CIB_/CIB_OP_/ and move to cib.h where they belong
Hush up some debug logging
CIB diffs are a more restrictive version of XML diffs - add wrappers for them.
s/CRM_OP_CIB_/CIB_OP_/
Run new tests
Include a complete node_status update - we cant assume there is anything there.
All CIB operations should go to the CIB, not us.
Be consistent with other places
We're never interested in what crmadmin has to say for these tests
Rearrange some logic so that fsa_membership_copy->id is always updated to
There is nothing left for crmadmin to do once the syncronous CIB call has
Portability changes for those without getopt_long()
Make sure diff_of_diff and intermediate are used even if its not a dev build
Include portability.h as soon as possible
Some platforms dont have ctime_r
Logging changes
The TE now uses the diff notifications which makes it impractical to
Move the dc_uuid code to the join finalization piece to avoid setting it twice
Abstract the finalization code-path so that it is performed consistently
Logging
Obtain the transition id that the TE is providing with every operation.
Use diff notifications from the CIB instead of updates since:
Adopt a "cl_msg"-style naming convention for xml.
Adopt a "cl_msg"-style naming convention for xml.
Spelling
Logging
Reduce logging
Use XML_LRM_ATTR_TARGET_UUID instead of XML_LRM_ATTR_TARGET (which may not be
Make sure "sgi1" doesnt match against "sgi11" too - alanr
Implement rsc->on_stopfail more generically as rsc->action->on_fail so that
Logging
New XML helper function for deleting the first object that matches a given
New operation
Add a new delete operation, rename the old one.
Server-side support for the new delete operation
Logging
Add the server hooks for the new delete operation
Support the new delete operation
Logging
Logging - these we can usually do without
Logging - we can usually do without these messages or infer them
If there is nothing to add, dont try.
Add some extra logging to understand the root cause of the NULL diffs that
Doh
Its really not up to the API to log this error.
Logging to make sure we only sync to those that reqiure is
Better names for macro args
Dont overwrite rc when extracting the call type
Remove temporary code
Let cluster_stable() use its default value in most cases
Relax the failure conditions for StopTest - syslog-ng just looses too many
Tweak v2 timeouts
Fix suppression of CIB disk writes: suppress is the inverse of
Fix the version check when processing a replace operation
Remove an unused option
Support complex resource
Make sure all attributes have the id field set
The output from crmadmin has changed - use the new format
Make sure a reply is constructed if the CIB was modified (so we can send a
Fix support for the cib_delete_alt operation
Make sure we dont enter the child_of_child loop if we removed the object
Remove duplicate logging of the input message
s/transition_timeout/transition_idle_timeout/
s/transition_timeout/transition_idle_timeout/
Support the idea of resources that are defined in our configuration but
Force another transition, dont just ignore the new one.
Note to self...
The behaviour of this is now understood.
The _default_ should be managed
Nodes that are shutting down do not count when we are calculating how many
Add a regression test for bugzilla #662
Debug for #670
Add a realtime monitoring tool for the CRM
BEAM fixes
Dont check for "Resource .* was active at shutdown" because it involves
Dont use errorstoignore()
Dont regenerate it every time
Further fix for 662:
Remove unused regression tests
Fix a memory leak exposed by regression test "inc6"
If is_managed == FALSE, then we should neither start nor stop it.
always add the fencing resource so that we test incarnations
Inhibit some warnings in the non-managed case.
Regression tests for is_managed=FALSE
A new regression test showed that implementing resource stickiness
Enable interleaving of incarnations
Make it easier for people to add support for one of the other curses variants
Tweak the curses stuff
Remove unused code
Use #defines for the resource types
Update the tests with the new resource names
Throttle full-sync requests so that we never have more than one active
s/incarnation/clone/
Call printw on the child resource
native_printw(): indicate failed and unmanaged resources
Remove duplicate (and incorrectly placed) logging
Add the ability to daemonize the monitor
s/native/primative/
Dynamically determin the amount of memory we will need.
Update the DTD
Spelling
Better and more reliable way of making sure clones end up in the right place
New CIB names
New CIB names
DEV asserts for zhenh
Use the new CIB names
Bring the readme up-to-date
Updates
Interval must be > 0 for it to be a recurring action - skip anything else.
Logging
Use #defines
Better error logging and structure.
Use mainloop and timers
Fix for bug #695
Logging changes
Up the CRM version
Fixes and regression test for bugzilla 696
Compile fix
More flexible fix for 696
Specifying actions that dont need quorum:
Minor cleanups
Fix the -c option
printing
Actions depending on stonith can also be run without quorum
Stonith resources do not need quorum to start
Make sure the standby attibute has an ID set.
Smarter check for quorum requirements
Why this would make a difference i dont know - but it works now.
Wait for the cluster to stabilize before continuing after the stonith tests
Solaris doesnt like macros within macros
Log HA messages that are too large to send so we can possibly do soemthing
Only display (partially) inactive resources when group-by-node is specified
Prevent the broadcast of diffs that result from applying a broadcasted diff.
Add a --enable-local-ltdl option to configure.
TABs to spaces
Detect (and complain about) tags that require an ID field but dont have one set.
Paranoia
Fix a call to time.sleep()
Fix for bug #723
Fix the reges list - missing comma preventing ERROR: from being found
Fixes for bugs 727 and 728:
Correct spelling of epoch
Fix for 733 where the CRM mistakenly thought it did not have to re-invoke
This is not as critical as I once thought.
These tags should never have an ID assigned even if the admin adds extra fields
Fix the processing of "stop failure"preferences
Dont trigger recomputes for "non-changes" (this was previously possible
If this ever happens we should return and not continue
Update to match reality
Include a unique TE ID in the transition key
Fix creation of recurring actions
Increase the timeout - we want CTS to catch uncompleteable transitions not
RecoverResource: Only ignore it when the resource from the test fails
When cancelling a transition:
Fix and regression test for #726
This isnt cause to print out the whole CIB - reduce logging bandwidth.
BEAM-style fix: remove the possiblity of printing NULL
Handle wrap-around gracefully.
Add some extra required shutdown patterns to NearQuorumPoint
Update two messages to warnings
Set the operation based after the original instruction is restored (if required)
Make sure the call_id is always reset when we pre-update the CIB
Use the correct instruction
Fix and regression test for #726
Log readability
Improved diagnostics for #728
Make sure this error-case is logged
5 minutes is overkill
Back out a recent change (1.117)
Fix for #728 (approved by lmb)
Fix and regression test for blocker 764
Update the DTD
Some extra cases for use with cibadmin
Delete operations should also increase the number of updates
Fix a operation sorting bug that triggered an assert
Transition key needs to be supplied - related to previous sorting assert
Update the version number for the impending 2.0.0 release
Fix and regression test for 797
Increase logging serverity
Some suggested script changes
These fields are filled in for us
Better instrumentation for LogWatcher
Options to run the CRM BSC test via CTS (instead of dodgy shell scripts)
Test required for CRM-BSC
Implementation of the ISO-8601 standard for use with time based constraints
Use the more widespread isdigit() rather than isnumber()
Update the DTD with the time-based constrainst
constructor/destructor for ha_time_t
Linux needs the apphb stuff to link for soem reason
Ooops
Clean up the processing of rules and expressions
Fix the sanity check
Changes to make node instance_attributes and time based constraints work
Factor out test_ruleset()
Allow all constraints to have lifetimes
Cleanup - no functional change
A new function for incarnation notifications
Help text
Implement (unconfirmed) notifications
BEAM Fix
gcc4 was right...
Turn on notifications at the resource level instead
Cleanups and tweaks for ttest
Fix for bugzilla 811, 812
Logging
Updates for notifications
Format changes
Optionally generate a .dot file describing the transition
Make use of the .dot version of the TE graph (a whole lot easier to verify)
Print the summary after stage8 (since thats when some notification actions
Refined support for (non-failing) notifications
Regression tests
headers for lrm wrapper functions
Regression tests
PE: Dont consider "notify" ops when unpacking a resource's state
Updates from wiki
Skip duplicate graph edges (just like the real output does)
Interim fix for BSC failure
Somehow discarding comments and whitespace exposed an off-by-one parsing
The ordering constraint between clone::stopped and clone::start should be
Update test cases
Create the notify attributes - list of starting/stopping resources and nodes
Example usage of date based expressions
Logging tweaks
Add the ability to test with different definitions of when "now" is.
With the adjustment to the array, this is now redundant.
Misc fixes for dates and time
Fix for bugzilla 800
Fix memory leaks matching the pattern spotted by zhenh
Always turn on stderr
s/crm_str_to_boolean/cl_str_to_boolean/
Put all resource parameters in the rsc->parameters hashtable.
Make this impossible to screw up.
duh
Be even more strict - expressions dealing with node attribute also make
gcc4 float comparision warnings
Make sure the option for transition_idle_timeout is always observed
Small optimization
Consistency cleanup
If we're wanting to shut down, dont overwrite the existing expected state.
Make sure an escalated shutdown results in a LOG_ERR
Include node info in the standard location for the TE to use.
Fix and regression test for #829
Asyncronously check the result of CIB updates for Stonith actions
NACK discarded resource ops to resolve 770.
Reliably allow BSC to clean up configs
Fix for handling of +/- INFINITY in default_resource_stickiness
Fix parsing of negative timezone offsets
Use integers instead of floats for constraint and rule scores
Make sure anti-colocation works for groups too
Anti-colocation for clones... cloneA cannot run anywhere cloneB is running.
Tone down the INFINITY logging
Make this a warning until I have time to do some profiling of the CRM
Logging tweaks
We use integers now
New config option
Logging improvments
Fix for cib_sync from a non-DC node during the integration process.
Add a recheck timer to poke the PE every once in a configurable while
Every 5 seconds creates too much noise.
These times can take a number of formats
regression tests
Fix resource stickiness for groups
Internal recovery of group resources needed some work.
Accept "+INFINITY" as well as "INFINITY"
Do not barf on "post-comments" (ones that come at the end of a xml-node).
Hopefully a fix for bug 869
Should be the final fix for 869:
Non-fatal recovery... no need to self-terminate when there is a reliable
Fix a rather severe file descriptor leak.
Remove it completely. It was for the fd leak problem which is now solved.
Correct processing of NACKs
Update OPTARGS to allow -R
An improved ID checker. Now checks for tag+id uniquness which can totally
New files not yet hooked up to be compiled
Handle sets of config options
Logging change
Observe the id-uniquness criteria that we now activly enforce
Link in the new CIB helper functions
This is not an error
Use the new crm_standby command
For 2.0.1 only support putting the node into permanent standby
Logging change - if DEV_ASSERTs are triggered it was really hard to know what
Detect duplicte and complain about lrm_rsc_op entries
Fixes for 878
Future compatibility - eventually we'll be able to use this to detect
Avoid logging an unnecessary ERROR message
Logging changes now that NAK is confirmed to be working.
New function
Remove redundant crap from the updates
New functions
New fields and actions
Remove some obsolete code
Implement master/slave
crm_mon now displays the standby status
Implement parts 2 and 3 of 881:
I really need to think of a goo dname for "foo" but anyway, if its FALSE
Dont overwrit eexisting values (like the way I build the hashtable)
New actions
Dont wait for these actions
Support the new delete and cancel ops
Optionally remove the stopped resource from the LRM
Fix compile for FC4
Clean up memory from LRM operations
Dont complain when things we disconnect from go away
Change the way we do versions/feature sets.
CRM/CIB feature set changes
Resolve the "rsc state" mess
make running_on a common resource property (crmadmin relied on this and
New resource orientated CLI
So thats why i did it that way...
just sketching...
Change the way we store and handle CCM information.
Remove all references to xml-config
Fall back to local node uname
tweak the UUID assignment
Fix up master/slave. Make sure master_node_max is observed.
Teak the range for which we check memory allocations
Merge the various print functions into a single call that takes
Examples
Logging changes - without these we really can't debug the CIB much.
Duh, the CIB wants to know about existing CIB clients not CRM ones.
Support supplying whole CIBs on the cmdline
Audit the contents of the CIB on all nodes
Fix apply/create diff toggle
Fix the detection of partition sets
Fix the CIB audit
Reorder the startup to fix problems detected by the CIB audit.
crm_mon should always use the CIB from the DC
Put election.c back to how it should be after 2.0.2
Support probing new nodes for active resources.
Include op->rc in transition magic
Set XML_ATTR_TE_ALLOWFAIL in the right circumstances
Make sure the CIB feature set is always correct in the future.
Set the section (according to the new rules) when calling create_cib_fragment()
Routing change: allow the CRMd slaves to send messages to the TE
Improvements to the probing of newly started nodes.
Use the monitor operation for probes
Work around a scoping issue for "section"
Change the format so CTS doesnt think its BadNews
s/status_0/monitor_0/
Cut down on CTS logging now that we know it works
Started -> Slave requires no effort and is not an error
Fix a startup ordering issue
Use a slightly more efficient format for adding the action parameters
80-char formatting
Make sure active resources are shut down when we exit.
Update the resource FSA
Be more strict about splitting strings in decodeNVpair()
tone down the paranoia
Make notify for master/slave work
Fix pre-requisits for stonith
Updated usage message
Wel already know what it is
Do not require sequential clone numbers starting at 0.
Fix the ordering of stonith actions for Peter Kruse
Update all test cases
Fix a typo
Support cloned resources which do not require clone numbers to be globally unique.
Create a --one-shot option for crm_mon
Handle permission problems better
Do local queries so that we dont block forever if there is no DC
Find <tag id=/> and update its fields
Various constants
Implement a new modify function for the CIB API which searches for <tag id=/>
Support the new CIB API call
New names for old things
Sometimes its NULL too
Support reprobing for resources, refreshing the CIB from the LRM,
Routing changes for the new LRM commands
Printing nicities and logging tweaks
Process all parts of the update (so we dont wait around for updates that
Always create orphan'd resources - so they show up in crm_mon
Move some functionality from crmadmin to crm_resource
Use the new command
need to specify -r as well
Support migration and un-migration of resources
logging
Handle single failed monitors (ie. from a probe)
Update tests
Handle this - instead of ftell() dumping core.
New tool that attempts to find problems with the configuration
Track how many failed
Log CLI errors to stderr where they will be seen and therefore useful
Better logging for do_id_check()
Differentiate between config and processing errors
Formatting
Calls to free CIB connections
New function header
Save on some memory copies
Add the ability to connect to a running cluster and check its config
Behave exactly as the real PE does
Move CRM CLIs to @bindir@ instead of @libdir@/heartbeat
Actually @sbindir@ is more appropriate
Update is_running fr pseudo parent resources (so crm_resource -M works)
Add call_options to update_attr() so that we can optionally perform the
Add a new notify type - clients may wish to be notified when the
Use the replace notification from the CIB to re-update our copy with our
Consolidate the deletion of resources from nodes into common function
Support crm commands that dont need to be waited for
Add a group to CTS - should have been done before 883 was closed.
The pseduo group actions should only be activated if the real actions
Change the way complex resources are created
Fix some paths now that the CLIs are in @sbindir@
Use action->uuid rather than action->id because its easier to understand
Extra logging
Group ordering and recovery after stonith
Dont delete by default
Cleanup DeleteRsc()
A common function for setting our DC
A better check for active resources at shutdown
When looking for monitor ops, only mark it optional if it was already
Allow standby to be specified in the nodes section and the status section
Use the common function
Mapping function that will get used one day
Revert David's ltdl changes - they break the build for non-developers.
We should never be evicted by the CCM
The migration rule may actually allow it to run
Logging
Always use the memebers and deleted list
Update tests
Support setting standby in the status section.
These fields are generally quite large - tell gshi's code to compress them
Add the ability to silently update the CIB. Ie. without telling the TE and
Adjust the path for crm_diff
Rewrite crm_aoti() as a function and use strtol (as suggested by the atoi
New CLI for displaying the contents of a local node's UUID
Use the wrapper function - which also uses the compressed version
cleanup
If _we_ have the "best" CIB then we still need to send it out to everyone else
Define 2 new return codes so that we can discover master/slave resources
Tweak Dave's fix (thanks!)
This shouldn't do any harm... tests agree.
Remove the crm_is_allocated() check.
Logging
Detect the native UUID implementation present in recent Darwin releases
Logging
Disable ineffectual versioning code until something good can be sorted out
Rework of the stonith callback handling.
Support uname's with capitals
Do checks that the supplied "preferred" node is not the same as the
Small cleanup - no code added/removed, just rearranged for clarity
Logging tweaks
Break each action type out into its own function
Log the call id
Update library versions
Logging
If the resource of a failed action is later found to be stopped then
Always update the CIB with the probe result
Logging
Improved partition audit - extensivly tested
Always use the compress variant
Make parsing XML input to cibadmin less magical
Process all failures _after_ the entire node_state has been read in.
Fix up the ordering of recovery commands
Fix the "we're busy" logic
Fix option passing
BEAM Fixes
Timer fixes prompted by BEAM complaints
Allow BSC to pass on Tiger
Make BEAM happy
Logging
Make sure we never send the same command more than once
Due to the way action parameters are stored in the status section
Remove the LRM from mainloop before disconnecting.
Fix up the help text
Set permissions to the same as the .spec file
Fix and regression test for 994
BEAM Fixes: Use of constant expressions in macros
Allow attributes to be created in sections we dont monitor
BEAM Fixes: Use of NULLs
BEAM: use of NULL
Unparseable config files shouldnt be overwritten.
BEAM Fixes
BEAM Fixes
Fix for a recovery bug (partial reocvery of a group)
Stupidity caught by BEAM
Fix for bug 254
Make sure the correct data is logged.
Put the "unmanaged" indicator on the same line as the resource
Now that we activly enforce tag+id uniqueness there is no need to prepend
Make the code fro reading the on disk configuration more obviously correct.
Be stricter about when we complain about the LRM and HA disconnecting on us.
Differentiate between a group resource that isnt running and one thats
Stop python from doing its 1 letter per line trick
Utility function for debugging stuff that shouldnt happen.
Always calculate the diff
Set a default for all CIB options
Logging
online help
Fix up get_uuid() - not sure how this ever worked.
usual 1 char per line thing
Better logging
Add a \n to the output message
Timerless elections
Remove the special handling for node_state objects and
Process all requests made to the CIB up until the point we were asked to exit.
Since the DC shuts itself down - we need to make sure it first shuts down
Fix for unrunnable actions that appeared in a CTS run
Nothing to do in this block unless we're both shutting down AND fencing
Add the ability to run ptest against a live CIB
Logging tweaks
Log the action *before* we execute it
Logging tweaks
Logging enhancements
Increase logging of failed xml patches
Logging enhancements.
Move the logging to where we have the diff
Send the correct value for shutdown
Borrowed from the CIB - an easy way to determine who we were sending too.
Sort out the code generating shutdown updates.
This should prevent a single channel with a potentially infinite stream of
Increase logging - sorry but too much depends on what is in these updates
Use the same priority as local processes to ensure we make progress locally
Get rid of the hideous macros and use a function instead... I guess I
Increase my ability to remember how the join stuff works
We are the server - if the client goes away it should be a warning only
Tips for new players...
Code formatting
Use the utility function for consistency
Backwards compatibility
Remove temporary logging
The log changed, thus the pattern must also
Reliably detect if the LRM connection is still active.
Cleanups and logging - no functional change
Logging
Fix up buffonary
The DC timer is basically obsolete - so remove all the O_DC_TICKLE stuff
Just use the channel's field
logging tweaks
Set all CIB options
Three new ha.cf options:
Observe the "record_config_changes" setting
Logging
Improvements to IPC usage by CIB clients - might prevent the CPU usage bugs
This preference is no longer used
Turn down some debug logging
Make sure I_NOT_DC is processed stright way.
Logging
Overhaul of IPC and HA channel callbacks
Look for cluster preferences by name not id
Add the ability to get a resource's value for a named parameter or property
Turn off the usage of short resource names - it causes the audits to fail
Read only calls dont need to be logged as prominently
Finally figure out why these couldnt be freed
UUID Cleanup
Fix up the short name stuff
Dont inhibit respawn if heartbeat boots us out
Use the result of the search
Fix memory leak in verification of diff applications
Only turn on verification for development builds
Fix a memory leak
Catch memory leaks if they creep into the code
Observe cib_inhibit_notify correctly
cib_process_change() was broken for everything except the nodes section
The id check was broken
Most transient clients connect, make a blocking call and exit.
Include the config section... duh
Switch a number of CLI tools to use the new syncronous connections
Logging
Dubass fix - peer messages never have the sync option forced
Two new API calls for finding the list of children and correctly obtianing
Improvements to "CIB Stats
Non "dev" builds don't need to call cl_is_allocated() and crm_validate_data().
A_READCONFIG often takes a while... it makes a syncronous call to the CIB.
A small fudge to account for a situation that came up once in about 10,000
Only part of the pervious commit was supposed to go in.
Further spped up the non-dev case by inlining the check for non-NULL.
Disable currently useless code
Time XML copying
Specify a setup and teardown function for each query
Fix a couple of crashes found with CTS
Check for correct permissions on the CIB at startup
Dont shadow the abort() function.
Drive the error level from channel->conntype instead of manually setting it.
Initialize to 0 not -1.
Wrapping the contents of the CIB into "<cib_fragment>"s makes processing
Make sure query channels cant modify the CIB
Dont assume its not NULL.
Tweak
Fix for 1060: The transition wont complete until the node leaves and the
Forgot this file in the last commit
Force one child at a time to exit
Use the group and user names
Stupid...
Logging
Fix a false positive in memory leak detection
With the recent changes to SimulStop processing, there should never be
Fix for a crash bug found by Peter Kruse.
Prevent a crash seen by alanr hen we tried to send a message after
Logging
Log these messages *less* verbosly than the error, not more.
Dont negate the signal passed to kill()
1080: Always wait for the TE to exit before shutting down
Needed the LSB include file
The TE needs the HA connection to stay a little longer
TE Updates should always be sent to the local CIB and *never* forwarded.
Dont add a signal handler for SIGTERM in the PE... just let it exit
Put a node's CIB into slave mode *only* when we loose an election or
SLAVE_ALL doesnt modify the CIB
A common transitioner library
Use the transitioner library
Missing files
Fix for Bugzilla 1086: Take crmd peer status into account when updating the CIB.
Put our CIB into slave mode on exit
Revamp the shutdown code
Use Alan's new asynchronous child process code to write out the CIB
Bug 1084: If an unmanaged resource exists on a node we want to shut down,
Use the #define
Bug 1072: OCFS2 Support - Provide notify data to start/stop actions
Enable the signal handler for SIGTERM now that the crmd puts us in our own
Logging
Bug 1084: Nodes that are offline but have active resources listed need to
Tests transitions for completability
Ignore notify fields when testing for changed parameters
Update test cases
Add new files to makefiles
Remove the fragment creation code - the CIB doesnt require it since 2.0.3
Cleanup
This is always BadNews now
Include class and type in the error message
Updates for rpm builds
Typo
Use temporary filenames and remove them afterwards
Logging
Shortcut the graph execution if its already complete
Logging updates
Some updates to possibly help with bug 1092: Rejoining partitions
Debugging for 1093: lost broadcasts
The CCM transition was not being set correctly
Extra logging for 1093
Potential fix for 1093 : the no-op was being sent to and from the wrong
If we're in an election, triggering another one wont help.
Extra logging for the election loop (bug 1093)
Clean up the join hashtables when a node leaves the membership
Do not build updates for probes any differently to any other update.
Remove a heap of crud from the TE
This could happen but its harmless and doesnt need to be an error.
Fix the id-collision logging
Remove the code that wraps answers up in fragments, all clients since 2.0.3
Downgrade some errors now that we understand them
BSC Regularly fails on lmb's machine
Logging
Add the resource's priority to the synapse so that the TE wont abort
Logging - identify updates with the client and call that made them
Remove unused operations
Logging
Update the test cases, some synapses now have a priority field
Logging changes, particualrly around the end of a transition.
Use the #define
Never invoke the PE for anything other than a I_PE_INVOKE
Make sure the node_state fields are set correctly at all times.
FSA fix: Need to send A_SHUTDOWN_REQ when we're in S_TRANSITION_ENGINE and
Same value, more consistent #define name
Hook up the transition timer again
Update the graph's completion action regardless of abort priority
For now, just warn
Remove disabled code
Clear the shutdown attribute when the crmd process goes offline
Logging of where node status updates originate
We already have this info, no need to ask every time we win an election
Differentiate between a graph pass where we performed an action and one
Only restart the transition timer if a new action was executed during the
Handle updates that do not come wrapped in a section
Fix for 1093: CIB messages go missing.
Logging for IPC_Channel->recv() failures
Dont wrap up the CIB update
Dont wrap the CIB update
Distinguish failed nodes from nodes on their way "up"
Clearly indicate that the Heartbeat connection disconnected because we asked
Remove a bunch of unused timers
The default is to be a memer node
Detect Heartbeat's equivalent of "we couldnt find a UUID"
We dont need to broardcast the nodes section update
Simplify the failure logic
Make it easy to observe the effect of zero'ing out the hostcache
Use the 'type' used by heartbeat (The PE supports it now)
Now that we sign off correctly we dont need this hack
One option when the hostcache stuff is working
The defaults dont need to be that long
Remove unused files
Make it a BadNews item instead
Sort out the action timer code now that we have a common TE library
This should happen at most once *ever* in a cluster
We changed something... trigger the FSA to be invoked
Done automatically - and the timer object is NULL at this point
Update version numbers
Fix for crashes found by debltc
Better quorum check
New "shutdown" message handler, not hooked up yet
Update the CRM version number
Fix for a timing issue where a remote CCM registered the DC exited before
Use a separate group id counter
Increase the prority of cancelling transitions
Remove dead code
Give the logs time to be printed/dispached
Avoid creating -INFINITY constraints where humanly possible
Make the implications of what the admin is doing REALLY obvious
A_TE_HALT was being filtered incorrectly
Dont just detect it, wait for the TE to finish what it was doing before
Stall the FSA *and* exit the function, no point stalling it if we keep going.
Important but not worthy of an ERROR
DTD updates
New fields
Spelling
Forced actions to get lost and made sure the recovery process works.
Updates to crm_master logging - it should not be invoked directly by the
Fix detection of the maximum number of masters
Fix for bug #113 : make the stonith type globally configurable
Update the expected results with the new verison number and with
Bug 197 - Checksum the CIB on disk
Remove the digest file too
Small tweak to reading XML from files
Remove duplicate test
Switch to our homegrown md5 implementation
Some notes to myself
Implement support for "failure stickiness"
Catch empty backup filenames
Only attach the failed tag if the update in fact failed.
We already know how big backup_file is, dont bother with strlen()
Support atomic incrimenting in the CIB.
Change the name of the fail count attr
Fix a typo in the transient set name
Indenting
The error log is sufficient
We cant take shortcuts when adding pieces to the CIB anymore
Enable fail-count tracking - seems to work...
Write the crm_failcount CLI for getting and removing a resource's failcount for a given node.
Change the attribute id for fail counts (matches whats used in crm_failcount)
Avoid creating constraints when we dont have to (failure stickiness)
Deleting an object that doesnt exist shouldn't be an error.
Logging
Logging - make it clear that the value is in ms
There are situations when we need to remove the cached value.
Fix TE detection of cancelled ops
Logging
Make the cancellation code more flexible
Make 100% sure that crm_assert_failed is FALSE when we start processing
When a node joins, remove the cached value for its UUID
Likewise for this delete operation, deleteing an object that doesn't
A script for creating primitive resources
Exit if the connection to the CIB or CRMd cant be established
Code for non-destructivly producing a core file when non-fatal asserts
s/nu/no/ because nu doesnt make sense
Allow the new script to be built
Update the usage
There is no reason to display *stopped* orphaned resources in the cluster
Provide the crm version for crm_resource generated delete events
Route the reply to the delete request to the place that originated it
Replace calls to CRM_DEV_ASSERT with CRM_CHECK which requires a recovery
BadNews for bug 1083
Dont send probes for orphan resources
If probed is set to false we should force all resources to be probed
If a probe detects the resource is running we shouldn't mark the resource
Get all testcases passing again
Re-enable stickiness for failed resources
Use the function's short name
More defines
Use the #define equivalents
Fix up 994
Abstract out the checking of action parameters
Regardless of short_resource_names, which is only needed for the TE graph,
Remove temporary debugging
Get Started -> Promoted working (should fix bugzilla 884)
Always use the calculated uname
Correct interpretation of -n and -r
Simplify
Default stop_orphan_resources and actions to true.
Dont derstoy the params list when checking for changed parameters
Master/slave is now tested and apparently works
Logging
A few tweaks
The last fix for master/slave... 4 node CTS passes start/stop tests
Potential fix for 1151: election bad news
Potential fixes for 1154: Election timeout
Use the node's uname not the unfriendly uuid.
Include role information for master/slave actions.
Just in case, role changes dont require a restart
Fail count only applies to recurring actions
Update the failure patterns to ignore
Add the ability to save the CIB used (so we can repeat the operation
Logging tweak
Remove a bogus check of the reply_id
Downgrade development logging
spacer
Compare the CIB's as CIB's... no just XML blobs.
Fix the example
Simpler method for doing the same thing (and I get a stack trace)
Do the same thing more neatly *and* tell the user what was wrong
Various updates to reflect reality
Allow the numerical equivalent of true|false for all "boolean" fields
wrong case
Code for calculating xml digests is now in the common library
Move the failcount updating code to the TE which already has to figure out
cleanups
Auditing output tweaks
Dont use the action in the callback, the call_id is enough to link us back
In some circumstances (usually when a resource has changed *and* is being
util function
Update the testcases (mostly not to use rc=7 for stop failures)
This mistakenly also applied to start ops
Revert the patch to unpack.c... start should never be called when a resource
Dont display provider for lsb and heartbeat resources
Allow XML to be optionally written to disk compressed in the bzip2 format
Instead of logging the partial inputs to the PE, compress and write them
Cut down on CIB logging...
random rubbish
Allow admin's to specify a role they want a resource to be in
Complain a whole lot more about deprecated options
Updates
Logging
Cleanup
log message
typo
Prevent the CRM from thinking it may be leaking memory
Some compilers dont like #if blocks within macros
Fix the recovery code when changes aren't valid
Limit the scope of deletes and updates when a section name is provided
If we matched the search root, keep searching just in case another copy exists
Set options in a on-deprecated way
Misc logging
Need to set the error code
Tweak the erase operation
Some really verbose logging in case i ever need it
Remove a bunch of semi-broken crud and make sure the supplied input is passed
Make sure all logs are flushed
Factor out the population of the nodes section and call it whenever the
Logging
Remove rsc_state which was only used for logging and not 100% accurate anyway
logging
conform to the dtd
New CIB error code
Dont fake xmlGetNodePath(), just use get_element_name() directly
Fix some asserts when compatibilty mode is turned off
Include the resource's class, type and provider
remove unused #define
Only compile dtd validation if we have libxml2
DTD tweaking
Install the DTD to HA_LIBDIR
debug
Add per-series limits on the number of PE inputs to save and disable those
sanity checking
params wont always be present
Hook the CIB up to discard updates that break DTD validation
Fix creation of options in crm_config
Make sure we can read crm_config options in cluster_property_sets
Actually this is all for 2.0.4 compatability, not .5
typo
Fix for bz #1182
oops, make sure it compiles with compatibility turned on
we do this in the destroy callback now
Allow dtd validation to occur in a 2.0.5-only cluster even if we have
Up the CRM version number
got the types wrong
Disable support for running version 2.0.5 in the same cluster with versions
Require instance_attributes and cluster_property_set to have an ID and
Update the CIB version number
Consistently apply sets of attributes.
If this is really important the CCM can check for it
no point defining it if i dont include the header file its in
Got the condition the wrong way around
Cleanup write_xml_file() a little
Some simplifications brought on by Coverity complaining about dead code
We're not using this code anywhere for the moment... maybe one day
Call abort() less and fill in sensible defaults more
fill in an id
Fix some issues raised by Coverity
Let the CIB fill in defaults (if required)
Revert the CIB if DTD validation fails
Temporary fix for 1183, the nodes section will currently not be populated
Switch to using msgfromIPC_noauth() which uses HA_Message's and also
Force DTD validation
Logging
Daemon for organising attribute updates to the CIB accross all nodes
Reorder the parsing to avoid FILE leak
Do the logging from the parent not the child
New error code
Conform to the DTD
Logging
Do DTD validation at startup too
typo
Allow the section to be defaulted too
There is no need to link against the apphb library
The default needs to be true now.
Do a few things before validation happens
Pay strict attention to when slaves set their DC and its version (and what
Cleanup do_id_check()
Make sure the id is unique
Include an ID for <lrm> tags
The updated (and optionally stricter) update_dc function.
<lrm/> tags need an ID now too
Only log if we changed something
Add the ignore_dtd field to the DTD
Logging conditions
Include an expression for the rule
We can be stricter here - BSC doesn't require it anymore
Correctly identify "failed" monitors only
Mask delete "failures"
Logging to stderr and help text
Make sure the DTD is part of the tarball
Allow update_attr() to set its defaults
help text improvements
Tweak the makefile
Make sure the set name is unique across the whole CIB not just the section
Get rid of the fragment creation piece
Update regression tests with new CRM version number
The ID field isnt interesting either
Logging
Put all cluster options in cluster_propery_set's
The demote of a failed resource is also implied by a stonith
Dont update the failcount for probes
Add options to the delete call
Common function
Untested pingd - replacement for ipfail
attr_delete needs options
Dont block
Remove a bunch of code that used fragments unnecessarily
logging
Only do the read (which is blocking) if we need to match the current value
Include the DTD in the tarball
Hooks to be able to force a full refresh
Extra check
Make the DC timeout dependant on the value of ha.cf/initdead
Wait 5s or initdead/2 (whichever is greater) for a DC to respond
remove dead code
Flush attrd to the CIB during the join process to prevent anything from
More regression tests for master/slave
Logging
Allow -m to work
The extra debugging is not required
Restore the target role *after* we've done the processing... sort of pointless
Better deal with nodes appearing during startup
Fix for CID11 - using the wrong variable in the if-statement
Fix for CID 30 - possible use of NULL
Fix for CID 12 & 13 - possible use of NULL
Allow promotion to work when long resource names are being used
Remove dead code found by Coverity
and the rest...
Make the pattern less restrictive
Only update the CIB if we're connected
Use crm_add_xml_int
Fix detection and handling of resource definition changes and orphan
Update the test cases to include interval and op_digest for lrm_rsc_op entries
Look for both forms of the resource name
Only send updates if we're connected to the CIB
Might come in handy
Compiling on other platforms
Compiling on other platforms
Dont cross quorum boundaries when running stonithd tests
remote_python_call
Add the interval to lrm_rsc_op
Update the AddResource test
more headers
if self.CM.Env["CIBResource"] is false, strangely we shouldn't create any
echo_cp() - copy files from one host to another by cat'ing their contents
Dont test logging in BSC mode
Tweak some log patterns to work regardless of how they're logged (syslog vs. file)
Set some options in BSC mode
Use CTS for checking basic sanity of the CRM and remove the bloody aweful
Logging
Remove some legacy stuff
Formatting
For consistency, put the instance_attributes first
Always set a value for DoBSC
This is way less restrictive than i'd like but i dont appear to have a choice
Need to include the interval
tabs -> spaces
typo
tabs->spaces
tabs->spaces
This slipped in
Not terribly interesting really
Increase some timeouts
A different approach to determining what type of update to send
Logging
This timer can go off in "normal" circumstances
Make sure demotions and promotions happen in the right order
Logging
start/promote ordering
Allow node state discovery
Patch from Dejan Muhamedagic
In the end, this is cheaper than stalling the FSa, setting a timer,
Logging
Remove use of fragments
We should not start with these attributes set
When autojoin is enabled the CCM likes to give us memberships like this:
Logging
Cachefile deletion code *DISABLED BY DEFAULT*
Allow is_managed and resource_stickiness to be "set" to the cluster's global value.
Log a warning if more than one node has the highest score for running a
Perform this action earlier
Logging
Its only a warning if the nodes both "need" the resource with a score of INFINITY
This is a better way to ignore things that happened before a stop
The stop action needs to have completed and succeeded before it counts
Update the test cases now that we're skipping pre-stop actions
Partially revert the last patch and change the approach.
diff options
Probes dont always happen first, we may be reprobing
Logging
Always remove peers from the "voted" hashtable
Code tidy-up
Process LRM events *immediately*
Log the transition once (when we unpack it) rather than potentially many
Dead code
TE Logging improvements
More logging we dont need to see
Check for NULL cluster connections and NULL IPC channels
These commands went to crm_standby
Extra asserts
more logging
Always process our IPC messages immediately
Turn down the logging
Create a LoggingTest but dont turn it on
Improved error message
Fix printing of priority
Actually obtain task first
Set the priority of post notify event to INFINITY so we never skip them
remove dead code
More notify stuff
Fix for 1219: resources with target_role==Stopped should not be assigned
update_xml_child() will now update the complete object
Dont use ssh if running locally
We need the promoted/demoted actions to run so the notifications can
Update the usage and remove --id since it was ignored
Make the CIB file read/writable only by the owner (hacluster)
Always use the correct form of the resource name regardless of what version
Handle a common mistake
Indicate if a resource has failed or not
Handle bad inputs
Now that we use < as the prioirty filter we can set this to 0
Make -U an alias for -M...
A fallback for -C if the resource has been deleted
Allow replace to work on any object in the CIB not only entire sections
The concatinated version of a resource's name
* Default to short resource names
Get the resource's id from the <primitive/> tag
Support an optional version parameter
Bug #1236: Default to stopping deleted resources
Fix for buffer overrun
When logging XML, allow admin-defined sensitive information to be hidden
Look for the failure count in the correct place
If target_role is set to anything other than "default", then any previous
The case used for these strings is not interesting to us, use strcasecmp()
Incomplete synapses are only a problem if we're not trying to abort.
We're rarely interested in this information now
Check for NULL *after* we retrieve the value
Simplify the retreival of cluster options (ie. cib_attribute -n is_managed_default -G)
If we're clearing the status of a resource that exists in <resources>
Only if the resource is orphan'd AND stopped do we ignore it
Keep dead resources out of the crm_resource -L results
Only produce core files for non-fatal asserts if we have debug turned on
Allow crm_mon to be used in a limited capacity even if ncurses wasn't
Support a generic replace operation on HA_Message objects
Update the pingd usage information
Support changes to the attr being updated
Logging
Fix for bug #1248
Correctly observe the "default" option
Parent options take precident over the child
Use the logging macros
It doesnt contain what we wanted... so what *does* it contain?
Function is the third argument
Formatting
Tweak the crm log format
Revert change with unintended sorting consequences
Fix for Novell #174049, OSDL #1247
OSDL #1252: Resources are not probed for when added
Rearrange for maintainability
Allow renames and or new ID's to be assigned silently in do_id_check()
Allow new ID's but no renames
OSDL #1253
Provide a generic way of setting and removing resource attributes
Novell #176014: wait longer before declaring attrd/pingd unable to connect
Allow deletions to be forced in no-quorum cases
Usability improvements for ptest
Coverity: Use of uninitialized data
Coverity: dereference of NULL pointer
Nice log message to keep provo happy
Fix the DTD
Missed a value
A much more useful default:
spelling of symmettric
Revert a patch from lmb - it does not make the id any more unqiue than it was
Allow the interval to be specified with a units eg. 5s
Fix the help text
Allow notifications to work when set as instance_attributes rather than
Tags with no attributes were able to be added to the CIB without having an
Supply the version number that an lrm_rsc_op entry conforms to.
CRM sub-namespace (disabled)
Globally Unique clones need to have all instances probed for
back out the complete patch - only part of it was supposed to go in
Try again...
Differentiate between parameters that the admin supplied for the RA and
Ignore fields with the special meta prefix (for when we turn it on)
New tag for meta attributes
Updated testcases (a result of the meta/regular parameter split)
Need to look for the "meta" version of this attribute name
Include resource and action meta options in the action definition
Support meta_attributes in the DTD
Include meta attributes (for the resource and action) in the action definition
Generically unpack action meta-attributes
Use the new meta namespace for CRM generated "parameters"
Downgrade an error
Logging changes
Fix clone migration in response to shutdowns when clone_max < num active nodes
New test cases
Blindly populate the meta attributes from the regular ones (rather than
Missed a file - tweaked the order_actions() function
Some new defines (ahead of time)
Preemptivly check for NULL
Remove clutter from the action paramters... only a few meta paramters are
Remove redundant code
The interval is already in msec
Allow access to the meta attributes
Notify attributes should also be in the meta namespace
Fix for non-fatal assert
Look for notify attrs in the new namespace
Set the resource parent
We should wait for the stonith action to complete before declaring the stop
Fix for Novell #178488: Notifications not generated for failed nodes
Marking the resource failed was not necessary and may have unexpected
Update the testcases
Slightly more elegant version of the previous patch
The final piece for Novell #178488
Doesnt need to be an ERROR
Look for the properties in the meta namespace (Novell #178488)
Novell #178764 - TE doesn't abort transition
Novell #178764: I_ERROR should always result in a transition to S_RECOVERY
Forced probes can cause an infinite loop... disable until we figure out a
All probes has a priority of INFINITY and will all run even if the
Put the assingment back after the declaration was moved out of the loop.
Use the complete list of members
DoFencing is a non-unique clone
Probes:
Remove redundant use of sizeif(char)... ISO C specifies that sizeof(char) := 1
No point waiting for the hello message... we dont care
Use the standard CRM log level changing mechanism
No connection to stonithd is a hard error (exit).
Logging under and overflow handling
Mark the action and synapse as complete for actions that cant be executed
crm assert only takes an expression
If the DC has the same version of the CIB as someone else, prefer the DC's copy
Turn off a warning
If the buffer cant be parsed, then thats a failure
More flexible placement of attributes and resources
Undo most of the previous change... libxml2 complains about:
Put operations at the start
If we're going to fail, do so with some decent logs
OSDL #1276 - Broken clones cause segfaults
Cleanup the application of node scores so that they are only applied once
Color 0 is never colorable (its the "cant run" color)
Logging changes to make debug level 2 usable.
* Actually return the result of the dtd validation, not TRUE *
Fix for crmadmin -S
Remove an eroneous \n
Accept the other possible status for ping nodes
Matching of hidden fields should be case sensitive
Common code for determining the current cluster state
Some more file that will definitly live in this new location
Output formatting
An extra case
At some points its safe to clear the cache files on all nodes, do so if clear_cache == 1
Novell #180303: Notification ordering
Use new notification rules
Tweak to parse command line options better when use full pathnames
Novell #180799: Multiple probes scheduled for non-unique clone
Look for pending actions on failed/stopped nodes and fail them
use transition_idle_timeout to populate a default action timeout when none
If we're deleting a failcount, then we need to poke the TE into a recalculation
Uh, and the rest
Fix node standby and add a regression test case for it
Logging
More logging
Novell #180699: Probe annoymous clones correctly
Backports from crm/pengine
Output tweak
Downgrade logging
Unpack parameters for new resources correctly
Preparation to send the graph via the filesystem
New defines
Pass the options in the right order.
Corruption in the CIB is to be avoided at all costs...
Always set a value (we used to delete the attribute if the ping count went to 0)
Have the TE look for transient node attribute *deletions* as well
Fix the transition graph (and add a regression test for) when a node wants
Logging for unclean nodes
New test case
Handle action re-creation correctly
Split the PE into three distinct parts:
Turn down some logging, diff failures can happen normally (ie. when a
Use the PE rules library to unpack CRM specific options (not yet enabled)
Coverity: Use of NULL
Shift more non-status functionality out of the lib directory
Message routing should happen immediately
Moved some typedefs around
Fix a memory leak in do_id_check
Cleanup the PE cleanup functions
New header file
New header file
Novell #183221 - Resource migration
The reverse of Novell #183221
Remove logging
The only reliable way to prevent resources from being active when
If no data_set is provided, just update what the resource already has.
Consistently use first_child for group based constraints
Fix clone interleaving that got broken when colocation constraints got fixed.
As a result of the clone interleaving fix, the instance numbers of
Assuming rsc_order(A,B), in addition to ordering B after A, allow B to be
Fix for processing of score_attribute in rsc_location rules.
OSDL #1281: Handle anonymous clone renaming correctly when we have too many
Allow crm_mon and crm_resource to be feed the CIB from an XML file rather
Patch from Bron Gondwana for not using a static array for unpacking XML from stdin.
Group child properties are supposed to take precent over parent options,
modify debug logging now that it works as desired
Serialize all fencing operations because the stonithd cant handle it on its own
Dont unconditionally call ncurses functions
help text additions
Unnecessary logging
Set a value (specifically off)
help text changes
ordered does not default to true
OSDL #1329: Fix for memory leak in CIB_OP_MODIFY
Update the regression tests to:
Cleaner regression test output
"nothing" is an alias for "restart" (which is better name anyway)
Allow DTD errors/warnings to be sent to stderr instead of syslog
Include crm_feature_set for all lrm_rsc_op objects
PE Memory leak
Remove dead code
Remove dead code
Remove dead code
Always include the op_digest
OSDL #1334:
Log a couple of configuration errors
Anonymous clone regression test
Fix a memory leak when overwriting clone_name
-l will print the raw resource hierarchy
More reliably reset the clone_name when the resource wasn't active
Use crm_realloc() to more effectivly create notify strings
Stand-alone version of the CIB for debugging. Start as: cib -s
Logging patch for OSDL #1340
A better label
OSDL #1340
Remove dead code
Stonith constraints need to be calculated against the members of the group,
Only complain about unprotected resources once per resource
Detect invalid prereq's for stonith resources
Update a bunch of tests to DTD-valid state
When the CIB is misconfigured, instead of shutting down (and the CRM respawning
When reconstructing the action key, we need to take into account if its a
*All* actions should get the default timeout, not just the ones that have
Update the test results
Remove unreachable code
Alternate logging setup that doesnt change the working directory
Pass the potentially very large TE graph around as a pointer to a file on
Add some newer options
Cleanup of crm_assert() to be more informative
This is unusual but not fatal
Make startup fencing optional for those brave enough to rely on it
Some extra shutdown logging
log to stderr only when in standalone mode
When sending a stop action, it would seem to make sense that the RA should
TE Logging
OSDL #1356: Group colocation fixes when one group cant run anywhere
Update regression tests: dont send resource parameters for a stop
OSDL #1350: Case sensitive searching for resources
Some extra objects that dont need IDs
Import the annotations back into the on-disk dtd for easier access to documentation
Relax the DTD to avoid common ordering issues.
Support unversioned CIB update diffs via cibadmin
formatting
OSDL #1344 - when an admin replaces the nodes section we should make sure
It appears DTDs cant handle enumerations with # in them
OSDL #1356: reordered the "are any resources active" check to happen after
OSDL #1354: CRM silently ignores trailing characters in XML input
strcmp() doesnt return a boolean like safe_str_eq() does :-/
Drop whitespace before checking for trailing characters
Split some monolithic functions into smaller pieces
Memory statistics helper functions (should have been in the last commit)
OSDL #1360: Clones enforce resource_stickiness == INFINITY
Regression tests
Fix remote replies
CRM version details
include the heartbeat version
Dont delete messages in the send_(ipc|ha)_helpers to make memory leak
More tweaks for trying to pin down the memory leak
More memory usage changes
memory checking for xml diffs
memory allocation tweaks
Free the existing CIB at shutdown
Fix potential memory leak of op disgest result
Implement an option (-f) that instructs the CIB to:
Tweaks so that we behave in read-and-free mode
Dump out unfree'd memory at exit and at regular intervals
Some tweaks to the way that the per-action-cib code works
For now never reset the usage stats
Free ccm_transition_id on exit
Introduced a memory leak in an attempt to fix a memory leak... um....
logging in the case of memory leaks
small cleanup
Dont print the cib contents every time we read it (in per-action mode)
Tweaks to the memory leak detection code
Memory leak
Disable crm_diff_mem_stats for the time being
Only deallocate the CIB at startup when we're in per-action mode
Logging
Broadcast diffs were failing because the input diffs contained version
Now that the memory leak is located:
Fix incrimenting of the update counter, I had missed some cases where it
Reorganise the logging of lrm ops that havent completed at shutdown
Eventually exit even if there are LRM operations that havent returned
Fix memory leak in init_client_ipc_comms_nodispatch()
OSDL #1269: These are processing errors only (not config ones)
Memory leak checking
Regression tests for probes that didnt complete yet
CIB Stats logging
Processing of pending probes.
Detect more cases where the <nodes> section needs to be refreshed
Extra functionality for ptest
OSDL #1369 - Node status
Passing of PEngine output to the TEngine
OSDL #1376 - Prevent graph loops due to STONITH
Always unpack the correct part of a diff operation
Bring test cases up to date
New regression test
Send filtered resource "stops" as successes
STONITH events need to inputs to start events (not stops)
Dont stall the FSA if we try to invoke the TE after we've stopped it.
target_role should not affect the value of is_managed
Utility function for finding all XML nodes matching the supplied criteria
If at all possible, figure out which resource attribute entry the user is
A largely self-documenting shell/meta-CLI for controlling the CRM.
All ops are recorded in the op table
Remove
Disable fencing of unseen nodes in CTS runs because CTS still cant handle it
Separate action timeouts from network delays by introducing the network-delay option
Finish creating generic, self-documenting option processing code
Remove the hideous OCFMSDummy agent that timed out frequently and didnt
Better logging for timed out actions
New tool that tests for OCF compliance
Logging updates - generally make the PE quieter
Tweaks to the OCF tester
Correctly process the "dont fence unseen nodes" option
Fix automatic help
Various tweaks to the cmd completion and lookup code
Applied patch from Matt Stancliff to allow building on platforms that lack
Use new name for macro
Patch for building without bzlib
Remove reference to dead file
Include ocf-tester in the dist tarball
BEAM Fixes: Possible use of NULL and memory leak in error handling
Stop BEAM complaining and just use an int instead of the enum type.
Represent 2.0.2 as it went out using CVS
Merge in fake 2.0.2 release
Reorganise how for the crm shell files are installed
Use _SCRIPTS for 'cluster' because _PROGRAMS assumes it comes from a c-file
Memory leak fixes from Kevin Jamieson
Merge with crm-stable
Filter out updates that arent for cluster members (eg. ping nodes)
Merge with heartbeat - Changes split out the GUI into a new package
merge with ha-dev
Fix find_recurring_actions to only return recurring ones :-/
PE Logging patch
Update expected PE test outputs
Merge with crm-stable
TE: Update resource failcount in all required cases
Remove colors from the status library
Remove extra color related code
PE: Initial conversion from color-based allocator
PE: Re-impliment colocation
PE: Add a (disabled) method for updating node weights
PE: Fix memory leak in native_choose_node() due to inappropriate use of g_list_sort()
PE: Remove dead header
PE: Include a resource's allowed nodes in dev output
PE: Implement weak colocation
PE: First cut at clone allocation without colors
PE: Re-implement clone allocation - conversion from color based algorithm
Re-enable master/slave allocation
PE: All colocation constraints are uni-directional
PE: Fix unpacking of +/-INFINITY in colocation constraints
PE: Fix colocation with resources that cant be allocated
PE: Fix group colocation and regression tests
PE: Fix manditory colocation with resources running where we cannot
PE: Logging improvements
PE: Fix allocation of unmanaged clone resources
PE: Enforce clone_node_max for already running resources. Re-enable clone interleaving.
PE: Make sure unallocatable clone instances are processed
PE: Fix clone allocation when instances cant run on their current nodes.
PE: Dont allocate clones to stopped nodes. Coloring conversion complete!
TE: Implement a more reliable version of parse_op_key() that handles _ in resource names
Merge with the reimplementation of the PE (removes the coloring stage)
Core: Correctly parse the action's interval in parse_op_key()
Add a manpage for crm_resource
Merge with crm-dev (PEngine re-write) and other fixes
TE: Correctly calculate the number of digits in parse_op_key()
Merge with crm-dev
Core: Port the smart attribute matching code from crm_resource to the CIB library
Core: Optionally only return the first match in calls to find_xml_children()
Core: Fixes to CIB attribute handling
PE: Downgrade missing score_attribute log message
CRM: Always include type/class/provider in operation updates
PE: Fix double application of rule scores for rsc_location constraints using boolean_op=and
Admin: Removed dead file
Tools: correct use of 0 in ocf-tester. Patch from Gildas Le Nadan.
Merge with crm-stable
DTD: Include start_delay as a property of operation objects
Admin: give the cluster shell a less generic name
DTD: Reinstate 'hours' as an element of date_spec and duration
Admin: Add help text to the iso8601 date utility
PE: Fix implementation of date_spec when no range is specified.
PE: Print *every* node's score for resources in the debug output.
PE: Correctly handle INFINITY in various numberical preferences
Core: Do not print filenames in CRM log entries
PE: Consistently check node availability in native_assign_node()
PE: Create all internal constraints before we start allocating resources to nodes
PE: Implement a new scoring system for rsc_(co)location constraints
PE: Fix application of rsc_location constraints for groups in non-symmetric clusters
Merge Lars' two heads into one
Tools: The CIB libs now require all attrd entries need a section. Default to status.
Core: Use Hg version details to identify CRM builds
Core: Slightly different build message for crm version.h
Core: Some extra trickery to get crmd to rebuild when version.h changes
Core: Use the heartbeat version instead of the CRM feature set
PE: Only complain the first time we encounter an option's 'old' name
PE: weird actions are only a configuration problem
PE: After calls to native_assign_node(), resources are always considered "done"
PE: Dont shortcut the allocation process in calls to resource_location()
PE: Create resource actions immediately after node allocation
PE: Relax the colocation rules for groups now that weak colocation is implemented
CRM: Time to up the CRM feature set number
PE: Relax colocation rules for clones now that weak colocation is implemented
PE: Improved error message
PE: Implement manditory colocation with clones
PE: Colocation enhancements
PE: Initial implementation of stateful colocation
Tools: Add addtional status checks to the OCF RA tester
PE: Logging updates
PE: Fix colocation with clones
PE: Clones were not being stopped on node shutdown
CIB: Fix for OSDL #1432 - update_attr() causes attrd to hang at shutdown when there is no DC
CIB: Fix for OSDL #1385 - Corrupted config file prevents heartbeat restart
Merge crm-stable back into dev
Core: Use a more efficient set of logging macros for the CRM
Core: Use isspace() instead of a complicated switch statement
Core: Get the debug messages right. Make sure log filtering is working.
Merge dev and crm-stable
Tools: Make it easier to supply configurations to crm_verify
PE: sort regression test DOT output so its order independant
PE: Allow resource colocation based on node attributes other than #id
OSDL #1421: Improved handling when timeout < start_delay
OSDL #1435 - Fix TE regression, never update the CIB with unconfirmed stop actions
Hg: Merge crm-stable back into the development branch
PE: OSDL #1447 - Pre-notifications for promote occur before start is completed
PE: Ammendment to the fix for OSDL #1447
PE: Rewrite clone allocation to reuse more code and be simpler to understand
PE: Remove dead code
Core: Logging cleanups
Core: Remove dead code
PE: Centrally but safely define internal resource structures
Merge crm-stable back into dev
PE: OSDL 1452 - Allow any two pairs of actions to be specified for rsc_order
PE: OSDL 1452 - Asymmetrical resource ordering contraint fixes
CRMd: Log resource actions consistently with the PE and TE
PE: Handle asymmetric clusters where RAs are not installed on all nodes
Ensure variant.h is include in the dist tarball
PE: Update metadata descriptions of some configurable PE options
CTS: Fix an expected pattern to match the new crmd log message
PE: Allow resource stickiness and failure stickiness to change based on
PE: Use the helper fn that supports "infinity" when parsing rsc_order scores
Tools: Include the scope from which the standby attribute is set
PE: Fix a botched commit, char2score only needs one argument
PE: Process group elements in reverse so that the location weights are merged
PE: Reduced logging
CIB: Fix query_standby() - pass the correct section name for nodes
Tools: Fix assert failure in crm_standby - dont copy NULL strings
Common: Logging tweaks
CIB: Startup processing improvements
CRMd: Ensure HA_VARLIBDIR/heartbeat/pengine is writable by the CRM
CRM: Put something in the if-block
Hg: Merge crm-stable into dev
CRM: Reduced logging
PE: Improved magic for rewriting rsc_ordering rules for non-atomic resources
PE: Logging updates
PE: Whitespace
PE: Fix some group allocation properties broken by 63b26c518a7f
Hg: Merge crm-stable branches
Hg: Merge in some more crm-stable changes
CTS: Remove a bunch of delays now that nodes no longer appear from nowhere
CIB: Logging
CRM: Allow RA metadata to be parsed
CRM: Lay the groundwork for reload support
TE: Remove dead file
CRM: Logging macro paranioa
CTS: Support optional gathering of per-test oprofile data for a user-specified node list
PE: Remove dead code
PE: Remove dead code
PE: Correct a careless paste error
CTS: Function correctly when no profiling is requested
TE: Set a reason for new transitions when the TE first starts
TE: Some slightly more efficient and readable logging for the most common abort case
crmd: Allocate enough room to create the "reloadable" key
crmd: Revamp logging of reported LRM events
TE: Include the action's graph id as part of the key sent to the crmd
crmd: fix careless copy/paste
Hg: Merge crm-stable into dev
CRM: Rename the network-delay option to something more appropriate (cluster-delay)
Hg: Refresh crm-stable from dev
CIB: Disable costly memory usage accounting thats mostly unused anyway
cts: Restart the profiling daemon on nodes killed by STONITH
CRM: More efficient use of strlen
CRM: More efficient way of processing atomic XML value updates
CRM: Rewrite crm_strdup to work when CRM_USE_MALLOC is defined
CRM: Put allocation macro into do-while blocks for safety
PE: s/network-delay/cluster-delay/ in regression tests
CRM: Make parse_xml and friends even more efficient by removing more strlen calls
CRM: Cleanup of malloc related macros
CRM: Implement and use add_node_nocopy() for users like parse_xml()
CRM: Remove all references to F_XML_PARENT
PE: Avoid calling safe_str_eq if we can
CRM: Tweak (for speed) and rename safe_str_eq
crmd: clarify error message
CRM: Cleanups and speedups in XML utilities
CRM: Install headers into their correct directory locations
Hg: Merge in some build changes from crm-stable
Core: Stop using the old ha_malloc macros for cl_malloc and friends
CRM: Fix ISO date handling for Jan 01 of any year
CTS: Look for ha.d in the configured location
crmd: Some more speedups for string comparisions and reload support improvements
Util: Fix a botched conversion from CRM_ASSERTs
CTS: Cleanup ResourceRecover
CTS: Logging change
TE: Remove dead code
crmd: Only update the voted hashtable with no-votes for the current election
Backed out changeset c38fb26a9497ab221688bfbfe6639eebcf29b0dd
crmd: add the required extra argument to crm_str_eq
crmd: Fix generation of FSA graphs
CTS: Explicitly test for the string "1" which means we have quorum
crmd: s/crm_str_neq/safe_str_neq/
PE: First cut at migrate support in _very_ limited situations
PE: More migrate code
PE: Final pieces of migration support
crmd: Make sure only current no-votes count towards election completion
DTD: Support the migrate_from value in rsc_op objects
crmd: Prepend CCM inputs to avoid a potential deadlock on startup
Hg: Merge in changes to sanitize malloc macro usage
PE: Update the regression test outputs with new the feature set version
crmd: remove dead code from the lrm wrapper
PE: Tweak log message
CRM: Finalize support for migrate
CTS: option logging improvements
CTS: Tweak v2 resource action interpretations
CTS: Cleanup CIB generation and add a migratable resource
Update comments related to the meaning of CCM callbacks
crmd: Correctly handle out-of-order CCM updates
Hg: Merge crm branches
Hg: Merge in migrate support
Hg: Merge in minor changes from the 2.0.8 release
PE: Add the PE part of reload support
PE: Allow rsc_colocation constraints to be symmetrical
PE: Bug fixes and regression test for reload support
PE: No need to regenerate the param-4 outputs every time
PE: Fix small memory leak in reload support
PE: Make sure lrm_rsc_op IDs in regression tests are correct
PE: Add explanitory comment
CRM: Logging
CTS: Remove unused import
BSC: Gracefully fail when python-xml is not available
PE: Allow migration and reload for children of complex resources
Hg: Fix another botched update from Alan
Hg: Merge back late 2.0.8 changes from Alan
Hg: Merge with upstream
CRM: Add a simple xsl stylesheet for rendering configurations
PE: Rename migration variables for clarity
Hg: Merge crm-dev into dev
crmd: Fix for OSDL #1465 - crmd gets stuck if shutdown while starting up
CIB: Apply patch for OSDL #1472 - Use-after-free in find_attr_details()
crmd: crm_atoi() needs two arguments :-/
PE: Utility functions for finding node's by uuid/uname
CRM: Allow retrieval of default config values without a hashtable
Tools: OSDL #1374 - Check for sane host names when migrating
CRM: Dont skip past the initial '<'
CIB: The decision to log failed connection attempts should be left to the client
Tools: OSDL #1451 - Allow resource meta attributes to be set with crm_admin
CIB: New error code
Tools: Print the duration we parsed
CRM: Refactor date printing to allow date -> string conversion
Tools: OSDL #1462 - crm_migrate should allow creating temporary rules
CIB: OSDL #1430 - Allow automatic assignment of object IDs
Tools: OSDL #1073 - Add clone support to crm_resource
Tools: Set a slightly higher default action timeout
CRM: Ensure the normalized result of (add|subtract)_time is updated before returning
Tools: Place the lifetime rule in the correct location
crmd: If we allocated temporary user_data content, we must free it also
CTS: Indicate debug log messages
PE: Dont waste time checking action definitions on nodes that wont be running resources
PE: Small logging change
PE: Update the dot file after the previous action definition check change
CRM: Changes to fake transition key generation
TE: Minor cleanup
CTS: Support the setting of a few more options for lrmadmin calls
crmd: Remove dead code
CTS: Rewrite the ResourceRecover to use asynchronous failures
CTS: Move a comment to a more appropriate location
CTS: Remove debug message
CTS: Rewrite the LSBDummy RA
CTS: Ignore ResourceRecover testing artifacts
CRM: Correctly handle syslog facility == 0
PE: Fix group recovery when a group it depends on is restarted
PE: Stop regenerating expected test results
PE: Have group-group recovery work for the right reasons
PE: Rename elements of the pe_ordering enum for clarity
PE: Move DeleteRsc to a more appropriate location
PE: Shift code out of the PE library that is only used by the PE application
PE: Change for code clarity
PE: Fix group recovery
PE: Allow order_constraint_t to represent multiple ordering conditions concurrently
PE: Not all pseudo operations should be included in the graph
PE: Correct arguments to CRM_CHECK()
PE: Clean up generation of and contents of dot-files
PE: Updated regression tests with the revised consistant format
PE: Only show elements of the graph in the dot-file by default
Hg: Merge in group-group recovery improvements
CTS: Remove resource tied to the DC - the #is_dc comparision is irrelevant
CRM: Fix replace_xml_child() when neither child nor update have an id
Admin: Auto populate the provider for OCF resources if one is not provided
Admin: Include rsc_location details in the resource listing
Admin: Use the current node uname by default when invoked as crm_standby
Admin: crm_primitive.py: Include all node preferences in the one rsc_location constraint
CIB: ODSL #1480 - Create synchronous r/w connections correctly
PE: Novell #239082 - Order promotions after _all_ demotions have completed
PE: Novell #239086 - Stable Master placement
DTD: The DTD used "integer" instead of "number" (which was used by the code)
CRM: Repair the status_printw macro now that we do logging slightly differently
Tools: Dont do ID checks on cibadmin diff's, duplicates are expected
Tools: Help text and input sanity checks for ocf-tester
PE: Novell #239079 - Ensure action priority is set correctly to ensure notifcations are not lost
Build: Include the exact version being built even when building an archived tree
PE: OSDL #1484 - Repair on_fail=stop. Patch courtesy of YAMAMOTO Takashi.
Admin: Novell #239075 - crm_master uses the same IDs for permanent and transient attributes
cib: Compile when malloc tracking is enabled
CRM: When malloc tracking is enabled, pass through the person that called crm_strdup
Tools: Fix a number of memory leaks in attrd
Tools: Compile when malloc checking is off, fix last mem leak
Hg: Marge in attrd memory leak fixes
Hg: Merge with upstream dev
CRM: Delete dead code, make empty_uuid_cache() safe with NULLs
crmd: Resolve a number of memory leaks
PE: OSDL #1486 - Fix for potential NULL dereference
CRM: Fix memory leak in compare_version()
cib: Clean up the list of clients wishing to be notified
crmd: Clean up leak in crmd_ccm_msg_callback()
crmd: dump newly allocated memory at each state change not only at idle
crmd: clean up more data at exit
CRM: Remove noisy and inaccurate memory logging
cib: Remove more ill-advised memory checking code
CRM: Remove the last references to cl_mem_stats from CRM code
crmd: Memory leak - Clean up the XML result of the LRM queries
crmd: Free the LRM connection at shutdown
cib: Memory leak - Free query matches in update_attr()
cib: Memory leak - Free intermediate results of find_attr_details()
cib: Memory leak - ensure the digest is always cleaned up
cib: Free various bits of memory at shutdown
PE: Logging
TE: Clean up the current graph on exit
CRM: Memory leak - Ensure libxml2 objects are always cleaned up in validate_with_dtd()
cib: Clean up the libxml2 parser at exit
TE: Always exit via mainloop
TE: Delete more items at exit
TE: Free the transition_timer at exit
CIB: Free the channel names at exit
CIB: Memory leak - free the digest file handle
crmd: Free the metadata hashtable and lrm connection at exit
cib: Clean up some hash tables at exit
crmd: Memory leak - ensure 'key' is free'd in get_rsc_metadata(), create 'restart' only when required in append_restart_list()
crmd: Memory leak - free the metadata list after use
CRM: Use unique top-level function names
cib: Allow forked writer processes to clean themselves up before exiting
crmd: Enable Valrind'ing of the PE and TE
crmd: Delete the HA connection only at exit
Hg: Merge branches from two test machines
Backed out changeset c65afe1e2d9006d0c6ec887359ba31d81bca5671
Hg: Backout changeset c65afe1e2d9006d0c6ec887359ba31d81bca5671
CRM: Convenience macro for list deletion
PE: Valgrind was complaining about this for some reason
crmd: Small documentation update
crmd: Memory leak - free the ping reply fragment after use
cib: Exit the writer process correctly so we'll continue to do write outs
crmd: Sort out the whole client deletion business. Also delete the resources hash at shutdown.
cib: Dont examine the group ownership of cib.xml since we set perms to rw-------
Hg: Merge in memory-leak related fixes
Build: Make supplying a Valgrind suppression file simpler
Build: Remove redundant and useless items from the build
crmd: Memory leak - clean up the intermdiate reload data
crmd: Novell #244444 - Free'ing g_alloc() data with cl_free()
Initial implementation of remote CIB connections using TLS and PAM
Switch to G_main_add_fd() instead of using io channels directly
cib: Support processing of commands sent via remote TLS connections
cib: Allow the remote connection code to be used with or without PAM/TLS
cib: Hook up the CIB remote connection listener
PE: Fix gcc warning - dont ignore the result of mktemp()
cib: Start a remote access listener on a port specified by the user
cib: Check for membership of the correct group
cib: Attend to various autobuild warnings
cib: Add but do not install a PAM file for the CIB
Build: Not all platforms need -lpam
cib: Older versions of gnutls used gnutls_session without the _t
CRM: Fixes for Coverty issues: 52, 55-58, 61-66, 68-69, 71-73 in run 240
PE: Make sure things that shouldn't be true are never true (added asserts)
Fixes for Coverity issues in Run 243
Merge in Coverity fixes
cib: valgrind - Only close the File* if we were able to open the file
CRM: Prevent a number of potential use-after-free's by zero'ing out free'd loop variables
PE: Simplify the logic in native_rsc_order_rh
Hg: Merge in some repairs for Coverity issues
CRM: Make the use of valgrind a run-time option
tools: Terminate the attrd process if heartbeat exits from under it
crmd: Clean up the voted hash at exit
PE: Fix a use-after-free of {key} when the monitor op already existed
Hg: Merge in local changes
Core: Better information about valgrind enablement
PE: Remove erroneous if-clause
crmd: ALL completed actions should be removed from the shutdown_list
PE: Prevent a potential NULL deference in clone allocation
PE: Provide useful feedback when a cloned resource is found active somewhere it shouldn't be
PE: Log the transition at a lower level in ptest
PE: Novell #246681 - Teach NoRoleChange() how to migrate resource's in the master role
PE: Re-use StartRsc/StopRsc for consistancy
crmd: Downgrade logging
CRM: Fix a use-after-free in parse_time_duration()
TE: Detect changes to attributes of the cib object
build: Remove dead code as sanctioned by Alan on the Linux-HA Users list Feb 20, 2007
TE: Only a limited number of cib attributes warrent aborting the transition
PE: Add a regression test for OSDL #1484 from Takashi YAMAMOTO
CRM: Return defaults correctly when no hash is used
PE: OSDL #1494 - Add node->score to the clone stability function
Hg: Merge in clone stability change
Admin: map cib_NOTEXISTS to 0 for crm_failcount
PE: OSDL #1499 - Perform #uname comparisions without case
Tools: Centrally define CIB call options so that -f always works in crm_resource
Build: Header-file related cleanup
Build: Header-file related cleanup
Build: Fix recent cleanup. Dispite the names, HALIB != HA_LIBDIR
Build: Fix recent cleanup. Dispite the names, HALIB != HA_LIBDIR
TE: Minor log change
TE: Remove dead code
TE: Small logging improvement
TE: Unpack graph actions correctly
crm: Gracefully fail when the target XML file is not writable
PE: OSDL #1492 - Set common resource attributes before calling the custom unpack function
PE: Move Ncurses related defines to a more optimal location and make sure its included by crm_mon
Build: Misc changes to allow building with crm disabled
PE: Fix anti-colocation when the RHS is not running
Attrd: OSDL #1500 - repeated call to attrd_updater with dampen absorbs attribute changes
tools: Teach crm_uuid to write ascii-form UUIDs to hb_uuid
tools: show help text for crm_uuid
tools: Better help when an invalid uuid is passed to crm_uuid
crm: Check the return code from fflush() when writing xml files - AnÌbal Monsalve Salazar
crmd: No need to create the resource if we're trying to delete it
Tools: Include class/type/provider in resource delete commands
tools: re-default to read if the are no arguments to crm_uuid
CTS: Account for a small timing window in the ResourceRecover test
PE: Make clones semi-sticky by default
Hg: Merge in upstream changes
crmd: dont update the restart list with unset attributes
crmd: Create the restart digest from an object with the same name that the PE uses
crmd: remove dead makefile entry
Build: Dont install crm-related man pages when --disable-crm is used
build: Trial a slightly cleaner way of specifying required libraries
crm: Fix memory leak in error path
Hg: Merge in local changes
build: removed unused library dependancies
CRM: Comprehensive review of error recovery logic for potentially failed calls to fopen()
Hg: Merge in local changes
tools: return the rc we calculated in read_hb_uuid()
PE: Make internal group ordering of bullet-proof
PE: Code optimization. No functional change
PE: Always print the resource's node scores
PE: Avoid a potential NULL dereference in an error path
crmd: Indicate the LRM error code if an operation cannot be cancelled
PE: Novell #251689 - Avoid needlessly restarting changed monitor operations for stopping resources
PE: tweak the dot file handling in ptest
PE: Novell #252693 - Enable migration in more scenarios
Hg: Merge in local changes
PE: Novell #252693 - Teach the PE how to migrate resources in a starting stack
Hg: Merge in local migration fixes
CRM: LF #1514 - Empty values in cib.xml loses other RA parameters
Hg: Merge in bugfizes
CRM: Make sure buffer is set to NULL after we free it
CIB: Allow cluster options to be created when the default set is not present
Admin: Allow cluster config options to be queried/modified/deleted. Broken in cs: 13c450735ba7
CIB: Prevent node_uuid from having any effect for <crm_config> changes
CRM: Dont call fflush() or flclose() if the output stream was NULL
PE: Create an implicit "cant run here" constraint if an action returns "not installed"
LRM: Inform clients when an RA is not installed on the machine
CRM: size_t is always positive... broke error handling in XML parser.
Hg: Merge with upstream
CRM: Use-of-NULL in validate_with_dtd() error path... Only clean up what we created
Admin: Allow crm_verify to be used with a named DTD file
crmd: logging
crmd: Correctly parse the shutdown timer option
crmd: OSDL #1532 - Single node cluster shutdown
CIB: Some extra options for developer testing
Tools: Add two new views to crm_mon. Patch by Christofer Edvardsen.
Hg: Merge in crm_mon patch against 2.0.8
Hg: Merge with upstream
PE: Don't start resources until we can verify they're not still running
Hg: Merge with upstream
Tools: crm_verify - dont complain about missing status section
CRM: handle NULL inputs to xml_has_children() gracefully
cib: Don't increase the version number if nothing changed
cib: Fix stupid compile error
cib: Warn at startup if important version information is missing
cib: Fix memory leak in cib_config_changed()
cib: remove dead code
PE: Memory leak fixes
PE: Memory leak fixes
Tools: LF #1500 - Calls to attrd_updater with dampen=0 not processed correctly
PE: Logging tweaks
PE: Logging
crmd: logging
PE: Logging
PE: Make clones ever so slightly sticky by default
CTS: Explain why we're aborting when nodes aren't found in DNS
crmd: Logging
Hg: Merge with upstream
TE: Tweak logging of unconfirmed actions
Merge with upstream
crmd: Logging
CTS: Logging
PE: LF Bug #1554 - Probes should not have priority INFINITY
crmd: LF Bug #1554 - Wait for the membership to stabilize before starting transitions to prevent unconfirmed actions
Hg: Merge with upstream
Hg: Merge in final changes to the SLES-SP1 version
cib: Potential logging of NULL
cib: Potential printing of NULL
cib: Remove redundant cleanup calls
crmd: Logging for improved debuging of parameter changes
cib: reduce info-level logging
Admin: Novell Bug #270977 - crm_resource ignores --meta in some situations
PE: LF Bug #1572 - Resource ordering not observed under some conditions - pt1
PE: LF Bug #1572 - Resource ordering not observed under some conditions - pt2
PE: LF Bug #1572 - Resource ordering not observed under some conditions - pt3
PE: Remove duplicate function declarations
PE: Code formatting and logging
PE: LF Bug #1572 - Resource ordering not observed under some conditions - pt4
PE: LF Bug #1572 - Resource ordering not observed under some conditions - pt5
Admin: Dont output the transition graph by default when calling ptest
PE: LF Bug #1572 - Resource ordering not observed under some conditions - pt6
PE: LF Bug #1572 - Resource ordering not observed under some conditions - pt7
PE: LF Bug #1572 - Resource ordering not observed under some conditions - final
Admin: crm_master: adjust expected return code from uname() to accomodate Solaris
PE: Stop of a partially stopped group failed when there was only two child resources
Hg: Merge with upstream
TE: Fix potential printing of NULL
cib: More accurate logging
PE: Bug 1580 - Remove logging inserted during developmental of fix for bug 1572
ais: Initial commit of an openais plugin that compiles but does very little else
ais: install into the correct location with a name AIS will accept
ais: fix errors preventing the plugin from loading
ais: Hook up the standard CRM libraries and logging
ais: Send AIS cluster messages
ais: Send and receive messages as well as node and crm membership information
ais: reinstate previous Makefile now that I understand why runtime complained about _cl_malloc
ais: Print out some membership information during events
ais: enable some very useful compile flags
ais: Disable a particularly noisy -W option when compiling AIS related code
ais: More development
ais: get darwin (32-bit) talking to sles (64-bit)
ais: Change the data structures to be packed with non-fixed data length. Use the correct local_nodeid.
ais: support different message targets
ais: plugin name change, tweaks to build on other archs
crmd: Bug #1584 - Always restart the join process when membership changes
cib: improved logging for the remote listener
Hg: Merge with upstream
build: remove unused Make variable CRM_DEBUG_LIBS
crmd: Potentially correct a potential logging of NULL
PE: Downgrade logging
Hg: Merge with upstream
PE: Add a dummy variable to struct native_variant_data_s so that it has a non-zero size
Build: Remove all CRM references to HA_MALLOC_TRACK
PE: Indicate orphan status when displayuing resources
Admin: crm_mon - Indicate the correct number of configured resources (excluding orphans)
CTS: Update log pattern to match new log message output
crmd: make sure fsa_cluster_conn is not NULL before we try to delete it
CRM: Logging enhancements
CRM: Logging enhancements
crmd: note-to-self for future handling of cancelled ops
crmd: remove dead cib-related code
CRM: Remove unnecessary logging
Tools: haresources2cib - Use the correct name for no-quorum-policy
TE: Reduce logging
CIB: Fix cib_config_changed() for when it becomes enabled
CIB: Only write the CIB to disk when the configuration changed (not just the status section)
Hg: Merge with upstream
CIB: #undef shouldnt supply a value
cib: Remove logging we dont care about
cib: More effective mehtod of turning off cib_config_changed() for now
PE: Do not test for changes to dead actions
Hg: Merge with upstream
crmd: Delete cancelled operations from the CIB
Hg: Merge with upstream
crmd: Novell Bug #286393 - Create resources, if necessary, in the lrmd so they can be stopped.
tools: crm_resource - prevent attempts to clean up non-primitive resources
Hg: Merge with upstream
Hg: Merge with upstream
crmd: crmd changes required by lrmd changeset 58b250732bf1
cib: Use a BadNews pattern to catch diff failures instead of ERROR
Hg: Merge with upstream
Admin: crm_resource - Prevent user's from cleaning up non-failed resources (unless --force is supplied).
Admin: Log how various CRM-related CLIs are invoked
Logging
PE: Provide the call id for ops to be cancelled
PE: Ensure stop ops happen before cancelations
crmd: Revise operation tracking
CRM: Re-impliment asynchronous failures without need of trickery from the lrmd
CRM: Additional files part of the async failure reimplimentation
Admin: abort only when a host is _not_ set
CTS: Consistant and configurable STONITH usage
PE: Prevent a potential use of NULL
PE: Pre-emptive change anticipating a change in the LRM
Hg: Merge with upstream
Hg: Merge with upstream
Bug #1601 - PE: Favor correctness over niceties when STONITHing nodes
Bug #1602 - CIB: Try and load backup configuration if loading the main one fails
CIB: read_attr() can safely always talk to the local CIB
Admin: crm_mon: -X has an argument
PE: Dont invoke check_action_definitions() if there are no actions to check
Build: Change occurances of HB_RC_DIR, missed in a recent cleanup, to HA_RC_DIR
crmd: Provide useful feedback when the metadata is unreadable
CRM: Delete resources from the status section when they're deleted from the LRM
CRM: Further optimize removal of resources from the LRM
Admin: crm_attribute - Support --inhibit-policy-engine for delete operations
Hg: Merge with upstream
TE: Log the node's uname instead of UUID for completed actions
Admin: crmadmin - include Hg version in --version output
PE: Dont log stopped orphaned resources
CRM: Allow cd-ing to the the core directory to be optional
CRM: Allow cd-ing to the the core directory to be optional
Admin: Bug #1621 - crm_mon should show failed starts of a resource
Admin: bug #1622 - crm_attribute: Include --inhibit-policy-engine in the help message
Admin: ptest is useful enough to warrent installation to sbindir
CIB: Revise version changes
CIB: Bug 1602 - Archive invalid/unusable configurations so we can continue
crmd: Bug #1575 - crmd cannot exit if lrmd is killed and there are pending ops
PE: Fix two small memory leaks detected by Valgrind
CIB: config_changed always needs to be calculated when processing write commands
cib: Add some comments to the code
PE: Make sure ptest always does a POSIX sort
cib: Clean up diff logging
cib: Fix detection of non-status changes
CTS: Stonith monitoring is not interesting to us - make sure it doesn't fail
RA: Move all remaining autoconf variables into common files that are automatically included
Hg: Merge with upstream
Admin: crm_resource - remove an overly simplistic check for when -C can safely be used
CRM: Bug 1625 - Cannot update an operation's timeout
Admin: Allow crm_diff to operate on compressed inputs
CRM: Only detect timeout changes for recurring ops
pingd: Hard code the name used for logging
CTS: Formatting - tabs->spaces
CTS: Rewrite the LSB script as a proper wrapper for the OCF Dummy RA
Hg: Merge with upstream
PE: NoRoleChange shouldn't always call PromoteRsc for master resources
PE: Use master scores to determine which clones to start where
Hg: Merge with upstream
CRM: Clean up processes left over by crm_abort(). Initial patch by DAIKI MATSUDA
Hg: Merge with upstream
RA: LSBDummy - Make sure the action is passed to ra_execocf
DTD: Add helpful comment regarding rsc_colocation objects
Admin: Bug 1632 - Make sure error messages regarding trailing XML input characters are seen
CRM: Fix matching of objects with an ID when the match-spec has no ID
crmd: Bug 1624 - Fix the delete op for cancelled operations
crmd: No need to log this under normal circumstances
Admin: Bug 1635 - crm_failcount should not complain if the attribute was already deleted
CIB: Bug 1633 - Pass the actual error from activateCibXml back to the user
PE: Bug 1628 - Complain about invalid operation intervals
PE: Fix the handling of remove-after-stop=true - it was causing unnecessary restarts
PE: Correct the CRM_CHECK comparision
PE: STONITH Shutdown ordering
PE: Aggregate the startup of fencing resources into a stonith_up pseudo-action
PE: Streamline STONITH ordering. Make sure 'all_stopped' depends on all STONITH ops.
PE: Potential use of NULL
PE: Include description for the remove-after-stop option
CTS: Give stonith devices penty of time to start
PE: Handle two new actions in text2task
crmd: Only remap monitor operation status to LRM_OP_DONE under the correct conditions
PE: Move the creation of stonith shutdown constraints to native_internal_constraints()
Tools: Bug 1653 - attrd crashes when no attribute is specified
AIS: More progress, build a working IPC client
CRM: Make logging setup consistent and do not log the command-line to stderr.
Tools: Bug 1653 - Further changes to prevent use of NULL when no attribute is specified
Tools: Bug 1653 - Misc attrd/attrd_updater cleanups suggested by Keisuke MORI
Hg: Merge with upstream
crmd: Bug 1655 - crmd can't exit when the PE or TE is killed from underneath it
PE: Update testcases to include all_stopped (added in cs: 800c2fec24ee)
CRM: Fix compare_version() to actually work correctly on a regular basis
PE: Fix up the tests now that compare_version() functions correctly (as of cs: 7d69ef94a258)
PE: Regression: Allow M/S resources to be promoted based solely on rsc_location constraints
PE: Add regression test for previous commit
CRM: Remove debug code that should not have been committed
PE: Update one more regression test output now that it has been verified as still correct
TE: Set failcount to INFINITY for resources that fail to start or stop
Hg: Merge with upstream
PE: Use failcount to handle failed stops and starts
crmd: Simplify the detection of active actions and resources at shutdown
CRM: Update crm/cib feature sets and the set of tags/attributes used for feature set detection
Hg: Merge with upstream
CRM: Repair logging that was broken by Alan in cs: 1f454f857ee8
Tools: ocf-tester - use the default value for OCF_ROOT if it exists
Admin: Low: Bug 1603 - Allow CIB digest files to be verified
CTS: Low: Optimize filtering of BadNews
CTS: Low: Tell NearQuorumPointTest to look for Pat:DC_IDLE before declaring success
CTS: Low: Ignore the correct BadNews message in ResourceRecover
High: crmd: Improvements to shutdown sequence (particularly in response to failures)
Low: CIB: Minor logging improvement
High: CIB: Improved CCM interaction
High: crmd: Simplify node_state maintenance
Low: cts: make cluster_stable() more informative
Low: cts: Resurrect the ComponentFail test
Low: cts: Update the expected log patterns and remove dups
Low: crmd: Ensure our assumption that all resources have a class and type is true
Low: crm: Let the caller decide how to log DTD validation failures
Low: PE: Correct spelling in log message
Low: cib: Rearrange so code to avoid excessive indenting
Low: cib: Enhancements to logging of DTD errors
LOW: cts: Remove unneeded argument for DC_IDLE pattern
Low: cts: repair log message
Low: cib: Fix error message text
Low: cts: ResourceRecover - Match the interval to that used by crm_resource
High: TE: Allow the TE to reconnect if stonithd is killed and respawned
High: cib: Handle master updates that fail DTD validation
High: cib: Allow the CIB to reconnect if the CCM is killed and respawned
Medium: CRM: Ensure fflush() and fsync() are always called when writing XML to disk
Medium: cib: Ensure fflush() and fsync() are always called when writing XML digests to disk
Low: cib: Refactor reading and writing of the CIB to disk
Low: PE: Warn when multiple nodes have the same uname
Low: PE: Update expected outputs with new version number
High: CRM: filter_action_parameters() - Make a copy of interval/timeout in as they're about to be free'd
High: PE: check_action_definition() - Ensure the meta attributes are also set so that the digest is calculated correctly (and consistently with the crmd)
Low: PE: ptest - dont execute the graph by default
Medium: PE: Dont check the parameter digest of orphan operations
Low: PE: Handle the 'fail' task in text2task()
Low: cts: Refine the patterns for the ComponentFail test
High: TE: Allow reconnection to stonithd if it is respawned
Low: PE: Update regression tests
Low: PE: Logging of changed resource definition
High: PE: Only create _one_ pseudo op per name (all_stopped/stonith_up)
Low: PE: Update regression tests for prior commit (58007745aef7)
Low: PE: Verify the changes to regression test bug-1572-2 are indeed correct
High: PE: The all_stopped action should only be part of the graph when the stop actions are runnable
High: PE: all_stopped should only depend on native resources to avoid recursion
Low: cts: Extra stats for the ComponentFail test
Low: cts: Update the expected patterns for ComponentFail
Hg: Merge with upstream
Medium: crm: Bug 1680 - parse_xml() can fail to detect mismatching close tags
Low: PE: Improvements to the graph and .dot creation code
High: PE: Break graph loops involving stonith
Low: PE: Regression test for fcf9cd9bff9b
Medium: admin: Set the exit code to non-zero if crmadmin can't connect to the crmd
Medium: TE: Dont modify the failcount for 'pending' action updates (rc=-1)
Low: TE: Improved logging of unconfirmed actions
Low: cts: 'Updating failcount for' is no longer worth tracking (action failures show up as other ERROR logs)
High: cib: Support the removal of multiple child objects - required for subsequent commit
High: crmd: Sanitize the maintenance of node_state entries
High: PE: Bug 1682 - Prevent use-of-NULL when clones have no active children
High: TE: target_rc should default to 0 (LRM_OP_DONE) to detect monitor actions that fail the first time they are executed
Low: PE: Fix a parse error in the attr8 regression test
Medium: PE: Bug 1685 - Supply the correct default for rsc_order constraints
Low: Admin: crmadmin - exit with rc=0 when the command completed successfully
Hg: Merge with upstream
Low: CRM: DTD - Expand the list of accepted values for boolean fields
Low: CRM: DTD - Explicitly specify a default for rsc_colocation.symmetrical
Low: Admin: Bug 1691 - crm_resource doesn't complain about empty strings as property values
Add the sample openais client
Medium: ais: Change the ais plugin name, disable manuall building now that everything works via automake
Low: ais: Provide missing function prototypes
High: ais: Dont include lha_internal.h in public headers
High: ais: Transplant a functioning crm plugin from the openais tree
High: Build: Add the openais subdirectory to the header path to allow for the way openais headers work
High: ais: Remove ais_fake.h now that the real openais headers are usable
High: Build: ais_fake.h should no longer be built
High: ais: Use the newly created aisutil library and related header
Hg: Merge with upstream
High: DTD: Remove stary character that made the DTD invalid
Hg: Merge with upstream
Low: cib: Start porting the cib to work on ais - compiles and connects
High: AIS: Add the ability to spawn children and notice when they exit
Hg: Merge with upstream
Low: CRM: Make crm_system_name a global variable and initialize it in crm_log_init()
Low: CRM: Make crm_system_name a global variable and initialize it in crm_log_init()
High: ais: More progress towards a functioning AIS port of the CIB
Low: AIS: New function for deciding what host to print
High: AIS: Fix routing bug, enable child process wait thread
High: AIS: Process messages from AIS, ensure uname is initialized
High: cib: Use get_message_xml() in preference to cl_get_struct()
High: CRM: Add children to HA_Messages such that the field name matches F_XML_TAGNAME
High: Build: Allow building with --disable-native-ais
Low: AIS: Minor cleanup
High: crmd: Tell the LRM to refresh the resource's default parameters
Low: CRM: Tweak the values for BZ2_bzWriteOpen() after re-reading the docs
Medium: AIS: Support compressed message payloads
Medium: AIS: Support compression of message payloads for local IPC messages
High: CIB: Make sure quorum is always TRUE for AIS-based clusters
Low: AIS: Clean out dead entries in Makefile.am
Low: AIS: Cast the result of sizeof() when printing
Hg: Merge with upstream
Medium: PE: Expose rsc->children which should mean we can delete all the 'loop' functions for complex objects
Low: PE: Fold child-loop functions back into the ones doing the actual work
High: PE: Implement a smarter algorithm for colocation
High: PE: More accurate method of detecting colocation loops in native_merge_weights
High: PE: Make sure the merge_weights function is called for complex resources
High: PE: Optimize the merge_weights functionfor groups
Medium: PE: Simplify the allocation of group resources
Low: PE: Regression test for group colocation
Hg: Merge with upstream
Medium: ais: Use AIS's native logging system for the crm plugin
Medium: AIS: Avoid pulling in -lcrmcommon and -lclplumbing for the ais plugin
High: PE: Ensure manditory ordering constraints behave correctly
Medium: AIS: Fix up previous commit
Medium: AIS: Ensure glib libraries are present
Low: Tools: Allow pingd to change identity based on the name of its executable - workaround for Bug #1701
Hg: Merge with upstream
Low: PE: Logging cleanup
Medium: PE: Bug #1705 - Don't segfault when allocating empty groups
Low: DTD: Include helpful comment for rsc_order constraints
Low: CTS: Limit the number of PE input files created
High: PE: Include preferences of colocated resources when promoting masters
Low: PE: Stop the master placement score from affecting the synapse's priority
High: PE: Make sure only the correct number of cloned groups are started
Low: PE: Downgrade logging when we fallback to default option values
Low: CIB: Allow notification to remote clients
Low: PE: Prevent users from trying to use on_fail=fence and stonith-enabled=false
Medium: PE: Bug 1710 - A resource's failcount was ignored on nodes with no operations for the resource
Low: PE: More efficient logging of orphan resources
Low: Admin: crmadmin - Logging tweak
Low: CTS: Add BadNews pattern - parameters should never change
Low: CTS: Add extra ignore pattern when the CCM is shot
Low: crmd: Remove redundant code
Medium: cib: Improved transparency when in operating in degraded mode
Low: TE: Pause for a moment before trying to reconnect to stonithd
Low: CTS: Add extra ignore pattern when the CCM is shot - do it properly this time
Low: PE: Logging - Include the transition_magic key when logging changed parameters
Low: cib: Logging - version message should not have been an error
Low: CRM: Downgrade digest logging now that its under control
Low: CRM: Key of the already set debug_level global variable when setting crm_log_level
Low: cts: Fix the log facility map
Low: Build: Build the contrib directory last
Medium: Logging: Sanitize and centrally define the log facility used by various subsystems
High: PE: More generic method of detecting colocation loops in native_merge_weights
Medium: PE: Enact a saner default for rsc_order.score (s/0/INFINITY)
Hg: Merge with upstream
Hg: Merge with upstream
Low: Tools: Resolve previous merge - crm_verify logging initialization
Low: ais: More developement - Added a couple of new queries to the plugin
Low: ais: remove dead code
Low: ais: cleanup
Low: ais: avoid having a single masive file
Low: ais: Headers required on linux
Low: ais: Impliment a generic dispatch handler for clients connecting to the plugin
Low: ais: improvements to signin, make the send_cluster_message smart-define global
Low: ais: Split crmd_ha_msg_callback() into two parts one of which can be called by the ais callback
Low: crmd: new_members from the CCM is never used, so dont bother copying it
Low: ais: Impliement ais callbacks
Low: ais: Create a membership abstraction so that we can plug in ais-based membership
Low: crmd: hook up the ais membership
Low: crmd: remove some dead code, add extra functionality from do_ccm_update_cache() to update_ais_membership()
High: crmd: Membership changes
Low: ais: enable the crmd, send complete membership data for notifications
Low: crmd: impliment populate_cib_nodes() for ais
Low: ais: Get the crmd to mostly start up - missing peer data
Medium: crmd: Reimpliment syncronizing of CIB queries and updates when invoking the PE
Low: CRM: Remove code made redundant by the revised CIB/PE syncing code (previous commit)
Low: CRM: Rename the membership hash ahead of more generic usage
Med: ais: Get the CRM to the point where it mostly functions under openais
Medium: ais: fix a routing issue - make sure messages that come over the cluster are processed
Medium: ais: hook up quorum and make it availble to crmd/cib
Low: crmd: Update the CIB with the node's version data when it becomes DC
Low: crmd: Update the CIB with the node's version data when it becomes DC
Medium: ais: Use the node's uname and send out membership updates
Low: PE: clones - Remove debug logging and simplify asserts in expand_list()
High: PE: Fix manditory ordering with m/s resources
Low: ais: membership - remove the 2-node quorum hack. Use no-quorum-policy=ignore instead
Low: cts: Add support for the crm running on top of openais
High: ais: Don't wait for in-active children during shutdown
High: ais: Only update a member's uname under the correct conditions - avoids it being set to NULL
High: ais: AIS does not always/normally indicate dead nodes. Infer them from membership data.
Medium: DTD: Fix comment regarding rsc_ordering constraints
Low: Tools: crm_resource - send errors to stderr and to regular logging
Low: PE: Update regression tests - some action numbers changed
High: PE: Ensure that resources depending (by order) on a master are not promoted if no master is available
High: PE: Missing header file change from previous commit (cs: 0c14cfe57dd9)
Medium: CRM: Re-evaluate appropriate IPC message queue lengths and throtle IPC _clients_ that hit them
Medium: PE: Fix minor memory leak when stopping orphaned resources
Medium: PE: Make sure all slaves (not just re-allocated ones) have role=Slave
Medium: PE: Master internal ordering enhancements. Added stopped->start, stopped->promote.
Low: PE: Log the current cluster state at LOG_NOTICE instead of LOG_INFO
Low: PE: Avoid pointless copying of XML items (actions) during graph creation
Low: PE: Extra debug information
High: PE: Clone colocation fixes
Low: crmd: Remove dead code
Low: Build: Remove unused LDADD entries (libpils.la and libapphb.la)
PE: High: Remove an errant call to exit() than prevented an assert from being triggered
High: PE: Prevent use of NULL in crm_mon when date_expressions are used by ensuring that data_set->now is always set
Hg: Merge with upstream
Low: cts: crm+openais doesn't (yet) support stonith related tests
High: PE: Prevent an infinit pe/te loop when reprobing with resources in master mode
Low: crm: downgrade a note-to-self log message
Low: cts: Update some seach patterns for AIS
Medium: AIS: Support spawning child processes as non-root users
Low: AIS: Remove the hard-coded debug file
Low: AIS: Downgrade an error log
Medium: AIS: Only update peer data for newly lost nodes
Low: crmd: Use a node's uname not uuid when indicating the election owner
High: crmd: 'born' is only an advisory value when running on openais
Hg: Merge with upstream
Medium: PE: Bug 1722 - By default, exhibit the old start-failures-are-fatal behaviour regardless of how resource-failure-stickines is set
High: ais: Fix the ordering in crm_ais_msg_types and ensure it always matches that of crm_children in the future
High: ais: Prevent the ais plugin from trying to use connections of disconnected clients
High: ais: Fix an assert condition to actually be a condition not an assignment
High: PE: Relax an assumption regarding clones that is not true when they are unmanaged
Hg: Merge with upstream
Low: cts: Remove BadNews pattern - its no longer of interest
High: ais: Define and use process flags consistently
Low: PE: Note-to-self comment for later
High: crmd: Prevent shutdown hangs by allowing the crmd to forget about pending actions for deleted resources
Hg: Merge with upstream
Low: TE: Code cleanup
Hg: Merge with upstream
Low: cts: Run "ps axf" when a node fails to stop
Medium: cts: Port ccm_tool so that we can enable the cts audits
Low: ais: Reduce logging
High: cib: Ensure the peer counter is updated after an AIS membership change
Low: crmd: Minor log message change
Medium: Tools: ocf-tester - Ensure OCF_ROOT is available to the RA subshell
Low: PE: Produce a config error when a clone contains more than one resource/group to clone
Low: cts: NearQuorumPoint - only check for Pat:DC_IDLE if there will be nodes up at the end
Low: cts: Add an extra ignore pattern for stonithd
Low: cts: Wait for the cluster to see the node come back before checking for S_IDLE after shooting the ccm.
Low: crmd: Remove code made redundant by the use of ProcTrack
Low: crmd: Increase the retry interval when stalling the FSA to 2s (used when connecting to the ccm, lrmd, cib)
Low: crmd: Use EXECRA_OK instead of hard-coding rc=0
Low: cts: Add a missing comma to the patterns list
Hg: Merge with upstream
High: PE: Prevent use-of-NULL when the admin creates a colocaiton constraint with an empty group
Low: cts: Fix a number of component patterns that changed recently
Medium: cib: Include digests of the cib a diff was made from and verify it when applying
Low: Build: Make sure make clean works in the cts directory
Low: crmd: Use the unaltered rc for timed out operation events from the lrmd
Hg: Merge with upstream
Low: cts: Remove erroneous search pattern from component fail test
Low: PE: Reduce log priority for some development logging
Medium: Tools: Bug 1738 - the crmd ignores some requests from crm_resource because it exited too quickly
Low: cts: Remove patterns that are no longer of interest
High: PE: Bug 1712 - Ensure manditory ordering constraints can cause complex resources to be shut down
High: PE: Remove an accidental commit that, when called, causes use-of-NULL
Low: cib: Remove development logging
Low: PE: Convert all the resource's boolean flags into a single bit-set
High: crm: Bug 1749 - Make sure the diff-related digests contain the complete CIB, not just the first line
Low: crm: Reduce severity of digest mismatch message in apply_xml_diff
Low: cib: Set callback_source=NULL after call to G_main_del_IPC_Channel()
Hg: Merge with upstream
High: cib: G_main_del_IPC_Channel() doesn't like being called with a NULL pointer and crashes
Hg: Merge with upstream
Medium: crmd: Bug 1737 - Inconsistent join state detected - Possible fix
Hg: Merge with upstream
Low: cts: Remove the backup CIB too if --clobber-cib is specified
Low: cib: Increase the retry interval when connecting to the ccm to 3s
Low: Build: Remove bogus whitespace from Makefile
Medium: crmd: Prevent shutdowns initiated immediately after a node is removed from the cluster with hb_delnode from stalling.
Hg: Merge with upstream
Low: PE: Split the contraint unpacking code into its own file
Low: PE: Reduce logging severity regarding creation of notifications
High: PE: Bug 1765 - Prevent master-master colocation constraints from preventing slaves from starting
Medium: TE: Pending operations shouldn't be processed
Medium: crmd: Don't remap LRM_OP_PENDING when building full lrm updates
High: PE: increment_clone() did not overflow correctly 9->10, 99->100, etc
Low: PE: Actions for stonith agents should never default to requiring fencing or quorum
Low: cts: Only set non-default CIB options. Don't add stonith resources when stonith is disabled
Low: Admin: Decrease verbosity of ccm_tool logging
Medium: crmd: Wait for AIS to send us membership data automatically - instead of requesting it
Low: crmd: Logging enhancements
Medium: Common: Only count peer processes of active nodes
Medium: AIS: Send membership data to children when they connect
High: crmd: Filter election responses based on time not FSA state
Hg: Merge with upstream
High: AIS: Allow messages to the TE to be routed
High: crmd: Resolve renamed variabled re-introduced during previous merge
Low: TE: Disable connecting to stonith since it hasn't been ported yet
Low: TE: Remove commented code
High: cib: Use the return value from call to write() in cib_send_plaintext()
Medium: Build: Rework AIS enablement and plugin installation
Medium: attrd: Bug 1776 - Attrd doesn't exit or reconnect when the CIB is respawned
Medium: PE: All other things being equal, prefer to keep non-failed instances alive
Medium: PE: Disable delete-then-refresh code. The crmd will remove the resource from the CIB itself making the refresh redundant
High: PE: Fix a botched commit (cs: 3de5760b06e0) that incorrectly allowed resources to start without quorum or stonith
Low: Admin: Do a full simulation when we have the live CIB - since we'll always have a status section
Low: crm: Handle a few xml-related error conditions without resorting to CRM_ASSERT
Low: crmd: Minor logging improvement to update_dc()
Low: PE: ptest - Handle malformed inputs more gracefully
Medium: crm: Dont generate core files for non-fatal assert (ie. the CRM_CHECK macro).
Hg: Merge with upstream
High: crmd: Prevent shutdown hangs caused by pending ops that can't be cancelled - because the no longer exist in the lrm
Medium: PE: Dont make changes to location constraints, when applying to groups, persistent
High: PE: Introduce a new API call for determining the location of complex resources
High: PE: Set next_role recursivly so that group promotion will work
Low: crmd: Cleanup - remove all 'return I_NULL' statements from the FSA
High: CRM: In ccm_have_quorum(), an OC_EV_MS_PRIMARY_RESTORED event means we _do_ have quorum
Low: CTS: The Process class requries an integer for triggersreboot
Low: crmd: Provide a more informative error when DCs detect other DCs during a join
Medium: TE: Allow the CRM to limit the number of resource actions the TE can execute in parallel.
Low: PE: Update the expected graphs to include the batch-limit option
Low: Admin: crm_standby - Don't complain abot missing values, print the default value instead
High: CIB: Fix the behavior of update_attr() and delete_attr() when the command is ambiguous
Medium: PE: Create a syntactic shortcut for the common use-case of "{resource} prefers {node} with {score}"
Low: TE: Logging - unconfirmed actions can happen under normal conditions
Hg: Merge with upstream
High: CRM: Future-proof get_message_xml()
Low: CTS: Use a different pattern to detect node shutdown
Hg: Merge with 2.1.2-24
High: CRM: Compressed XML needs to be retrieved as a binary blob, not a string
Low: cts: Turn off logd patterns when using openais
Low: cts: Update for the changes to the Process() class - causes_reboot=0
Low: crmd: Extra logging for quorum updates
High: cib: Fix CIB_OP_UPDATE calls that modify the whole CIB
Medium: crm: Improved logging of errors in the XML parser
High: ais: Fix memory leak when sending cluster messages
Low: crmd: Ignore diff-related errors when updating the cib with completed resource operations
Low: crmd: Avoid copying unused data for join_query_callback()
High: crmd: Fix two memory leaks. One in membership updates, the other in the ais message dispatcher.
Low: cib: Downgrade a log message that occurs naturally during cluster reformation
High: cib: Fix memory leaks caused by the switch to get_message_xml()
High: crm: Adopt a more flexible appraoch to enabling Valgrind
Low: crmd: Better variable name for enabling valgrind
High: cib: HA_VALGRIND_ENABLED needs to be set _and_ set to 1|yes|true
Low: ais: Fix enablment of valgrind by the crm
High: ais: Resolve memory leak in send_client_msg() and send_cluster_msg()
Low: ais: Print a maximum of 90 characters to identify bad messages
Low: ais: Detect and filter bad messages from ais clients
Low: cib: Remove needless log messages when the cib exits with active clients
Medium: ais: Put a limit on the number of times a client process will be repsawned
Low: crm: Fix log message for fatal asserts
Low: crm: Indicate how many bytes were requested when malloc() fails
Low: crmd: Simplify the logic in crm_ais_dispatch()
High: ais: Detect and filter garbled cluster and IPC messages
High: crm: Fix compilation when bzip2 is not installed
Medium: cib: Require --force|-f when performing potentially dangerous commands with cibadmin
High: ais: Use the asynchronous comms channel for unsolicited messages
Low: admin: Add --force as an alias for -f|--force-quorum
High: Build: Stop building Heartbeat
High: Build: First pass at removing non-CRM files
Medium: Build: Remove more heartbeat code from the repository
High: Build: Two extra header files needed for building
Hg: Merge in the code from the AIS port
Hg: Merge with upstream
High: Build: Fix building when OpenAIS isn't installed
High: Build: Allow the choice of cluster stack to be made at run-time
Low: Build: Promote the crm subsystems to be top-level directories
High: crmd: Bug #7 - Connecting to the crmd immediately after startup causes use-of-NULL
Medium: crmd: Delay starting the IPC server until we're fully functional
Low: crmd: Add an extra log message for use in CTS testing
Medium: CTS: Fix the startup patterns
High: cib: Ensure the archived file hits the disk before returning
Medium: Tools: Make sure the autoconf variables in haresources2cib are expanded
Low: Build: Update last release number
Medium: SNMP: Allow the snmp subagent to be built (patch from MATSUDA, Daiki)
Low: Build: Update the .spec file with the one used by build service
Medium: Build: Turn on snmp support in rpm packages (patch from MATSUDA, Daiki)
High: Build: Fix building of packaging directories
Medium: Core: Create generic cluster sign-in method
Medium: Core: General cleanup for supporting two stacks
High: Heartbeat: Accurately record peer client status
Medium: attrd: Port attrd to also run on AIS
High: Tools: Fix compilation of ccm_tool
Low: Alway sbuild man pages
High: Build: Arrange for the binaries to be installed to their tarditional heartbeat location
High: crmd: Revert part of a change that would have made newer builds incompatible with existing clusters
High: Core: ccm nodes can legitimately have a node id of 0
High: Core: Fix use-of-NULL in AIS cluster sign-in
Low: Build: Remove dead makefile code
Low: cib: Shutdown logging
High: crmd: Ensure nodes exist in crm_peer_cache before updating peers in crmd_ha_status_callback()
Medium: cib: Tweak the shutdown code
High: Build: Fix install location for headers
High: Build: Preserve the existing header heirarchy (for now)
Low: cts: Stonith failure doesnt cause node reboots
High: crmd: Relax an assertion regrading ccm membership instances
Low: crmd: Logging improvement
Low: cts: Logging improvement
Low: Build: Remove a bunch of redundant stuff
Medium: ais: Zero out the connection if we find that its failed
Low: Build: Remove build-related crud
Low: Build: Install pacemaker's cts into a different location
Low: Build: Remove more build crud
Low: Build: Even more Makefile maintenance
High: PE: Find headers in the current directory
Medium: Build: Remove dependance on -lnet
High: Build: Allow building without support for Heartbeat
High: cib: Fix compilation when supporting the heartbeat stack
Medium: admin: Prevent hang in crm_mon -1 when there is no cib connection - Patch from Junko IKEDA
High: Build: Compile out fencing when the heartbeat stack isnt supported
Low: crmd: Simplify the error logic in cib_rsc_callback()
High: transitioner: Missing brace - Fix compilation when SUPPORT_HEARTBEAT is true
Medium: core: Isolate more code that's only needed when SUPPORT_HEARTBEAT is enabled
Medium: core: Use a basic type for something thats not terribly important
Medium: core: Isolate more code that's only needed when SUPPORT_HEARTBEAT is enabled
Medium: core: Isolate more code that's only needed when SUPPORT_HEARTBEAT is enabled
Medium: core: Isolate more code that's only needed when SUPPORT_HEARTBEAT is enabled
Medium: core: Fix compilation when Heartbeat support is disabled
Medium: core: Use correct function name
Medium: core: Isolate more code that's only needed when SUPPORT_HEARTBEAT is enabled
Medium: core: Isolate more code that's only needed when SUPPORT_HEARTBEAT is enabled
Medium: core: Isolate more code that's only needed when SUPPORT_HEARTBEAT is enabled
Medium: core: Isolate more code that's only needed when SUPPORT_HEARTBEAT is enabled
Medium: core: Isolate more code that's only needed when SUPPORT_HEARTBEAT is enabled
Medium: core: Isolate more code that's only needed when SUPPORT_HEARTBEAT is enabled
Medium: core: Isolate more code that's only needed when SUPPORT_HEARTBEAT is enabled
Medium: core: Isolate more code that's only needed when SUPPORT_HEARTBEAT is enabled
Medium: core: Isolate more code that's only needed when SUPPORT_HEARTBEAT is enabled
Medium: core: Isolate more code that's only needed when SUPPORT_HEARTBEAT is enabled
Medium: core: Isolate more code that's only needed when SUPPORT_HEARTBEAT is enabled
Medium: core: Isolate more code that's only needed when SUPPORT_HEARTBEAT is enabled
Medium: core: Isolate more code that's only needed when SUPPORT_HEARTBEAT is enabled
Low: PE: Remove dead code
Low: PE: Remove dead code
Medium: core: Isolate more code that's only needed when SUPPORT_HEARTBEAT is enabled
Medium: core: Isolate more code that's only needed when SUPPORT_HEARTBEAT is enabled
Medium: core: Isolate more code that's only needed when SUPPORT_HEARTBEAT is enabled
Low: Remove dead code
Medium: core: Isolate more code that's only needed when SUPPORT_HEARTBEAT is enabled
Medium: core: Isolate more code that's only needed when SUPPORT_HEARTBEAT is enabled
Medium: core: Isolate more code that's only needed when SUPPORT_HEARTBEAT is enabled
Medium: core: Isolate more code that's only needed when SUPPORT_HEARTBEAT is enabled
Medium: core: Isolate more code that's only needed when SUPPORT_HEARTBEAT is enabled
High: Build: Move memberships.c to -lcrmcluster so that everything compiles
Low: resources: Get resources from the heartbeat project
High: Tools: crm_attribute needs time.h to compile
Low: Build: Fix the doc directory location
High: crmd: Pass in the dispatch and destroy callbacks correctly
Medium: AIS: Add a version field to the 'node-id' message
Low: cts: Support respawn or fastfail mode
High: TE: Update failcount for oeprations that time out
Low: Admin: Tweak handling of extra parameters in crm_resource
Low: Core: Remove redundant if-clause
Low: CTS: Make stonith start failures fatal
High: Core: Fix peer-process tracking for the Heartbeat stack
Medium: PE: Prevent a potential use-of-NULL in cron_range_satisfied()
Low: PE: Force anonymous clones to use clone_node_max=1
High: crmd: Handle all possible peer states in crmd_ha_status_callback()
High: Core: Heartbeat doesn't send status notifications for nodes that were already part of the cluster. Fake them instead
High: crmd: Use (node->processes&crm_proc_ais) to accurately update the CIB after replace operations
High: PE: Parse fail-count correctly (it may be set to ININFITY)
High: PE: Fix compile error - s/flag/flags/
Medium: crm_mon: Indicate when a node is both in standby mode and offline
High: crmd: Make sure the current date/time is set - prevents use-of-NULL when evaluating rules
High: TE: Ensure failcount is set correctly for failed stops/starts
Medium: crmd: Fix potential use-of-NULL in string comparison
Medium: crmd: Check we are present in the membership table _and_ active
Low: cib: Remove extra whitespace
Low: cts: Noticing the action fail is enough for CTS to indicate a problem
High: admin: Exit and indicate a problem if the crmd exits while crmadmin is performing a query
High: PE: Prevent graph-loop when stonith agents need to be moved around before a STONITH op
High: attrd: Don't sign off from heartbeat when using the openais stack
Low: PE: Simplify the start/stop failure handling logic
Medium: Build: Clean up configure.in
Low: cib: Add regression test
High: te: Cancel active timers when destroying graphs
Low: cts: Dont ignore crmadmin errors
High: Admin: crm_uuid - Look in the right place for Heartbeat UUID files
Low: Admin: crmadmin - Exit directly after finding the required response
Low: cib: Add more tests to the regression suite
Medium: Core: iso6601 - Support parsing of time-only strings
Low: Build: A cleaner way to rebuild python files
High: AIS: Messaging fixes and enhancements
High: PE: Prevent graph-loops when two operations have the same name+interval
Medium: AIS: Read a number of configuration variables from openais.conf
High: PE: Implement three recovery policies based on op_status and op_rc
Low: PE: Adjust the testcases not to use rc_code's that use the hard/fatal recovery introduced in cs: 6e625c514422
Medium: PE: on_fail=nothing is an alias for 'ignore' not 'restart'
Low: crmd: Add support for timing data when the LRM starts setting it
High: PE: Bug 1777 - Allow colocation with a resource in the Stopped state
High: PE: Bug 1822 - Prevent use-of-NULL in PromoteRsc()
Medium: AIS: Logging improvments
Hg: Merge with upstream
Low: Build: Add some helpful make targets for creating releases
Low: Build: Fix the number of changesets in the 'changes' target
Medium: Build: Fix comilation of utils.c when only heartbeat-common is installed
Low: Build: Add the start and end fields for changelog entries
Low: Build: Fix the changes make target
Low: Core: Explicitly cast variables in slist_iter and friends
Medium: Build: Finally implement a sane way to consistantly install pacemaker.lcrso
Low: PE: Update version number in regression test output
Low: PE: Downgrade some developement logging
Medium: PE: Bug 1822 - Don't try an promote groups if not all of it is active
Low: PE: Add a regression test for bug 1822
Low: Build: Include the .spec file with the project for convenience
Low: Build: Fix the postun specfile target
Medium: Build: Change the rpm targets to build from the current specfile not the in-tarball-specfile
Medium: CIB: Mask cib_diff_resync results from the caller - they don't need to know
Low: XML: Impliment a Relax-NG schema for validation and start on an xslt conversion from the old config format
Medium: XML: Fill in missing IDs during conversion
Medium: XML: Make the _ to - character translation in attribute names generic
Medium: XML: Sanitize the value of score during conversion
Low: XML: Make boolean-op optional
Low: XML: Allow multiple rules per rsc_location constraint
Low: XML: Make description optional for groups
Low: XML: Make description optional for clones and master/slave resources
Medium: XML: Mark various parts of resources as optional
High: UUIDs cannot be XML IDs
Medium: XML: Convert various resource attributes into nvpairs
Low: PE: Make some ID fields truely unique
Low: PE: Make sure the regression tests have valid XML
Medium: A whole swag of improvements to the conversion process
Medium: XML: score_attribute isnt an IDREF, sanitize the date_expression input combinations
Medium: XML: Allow multiple (meta|instance)_attributes for resources
Low: XML: Dont bail on the first failed conversion
Low: PE: Remove duplicate node from regression test
Medium: XML: A whole swag of improvements to the conversion process
Medium: XML: Make rsc_order.score optional
High: XML: Make rsc_colocation from/to text fields instead of IDREFs (for now)
Low: PE: Change the node uname's not to start with a digit to conform with XML ID requirements
Low: PE: Removed unused location constraint
Low: PE: Resolve duplicate ID values
Low: PE: Use the correct action in regression tests
Low: PE: Repair the probe-0 expected output
Low: PE: Only call parse_xml_duration() if one was supplied
Low: PE: duration/end is not relevant for operation='gt'
High: XML: Rename constraint attributes for greater clarity
Low: Build: Copy the tarball to the correct location
Low: PE: handle CRMD_ACTION_MIGRATE in text2task()
Medium: PE: Bug 1820 - Allow the first resource in a group to be migrated
Medium: PE: Bug 1820 - Check the colocation dependancies of resources to be migrated
Hg: Merge with upstream
High: CIB: Ensure changes to top-level attributes (like admin_epoch) cause a disk write
Hg: Merge with upstream
High: CIB: Repair the ability to do 'atomic incriment' updates (value="value++")
Hg: Merge with upstream
Low: Core: stdin2xml() - Skip XML processing if nothing was read from stdin
Medium: PE: Implement cibpipe, a tool for performing/simulating config changes "offline"
Medium: CIB: Implement a client-side operation timeout - Requires LHA update
Medium: Admin: crm_mon - include timing data for failed actions
Medium: Admin: Bug 1438 - Allow the list of all/active resource operations to be queried by crm_resource
Medium: CIB: Only invoke g_source_remove if we started a timer this time
High: Core: Adopt libxml2 as the XML processing library - all external clients need to be recompiled
High: PE: Move the call to free_xml() to the right place after the libxml conversion
Medium: crmd: Cleanup up the converted HA_Message (libxml conversion)
Medium: cib: Remove testing artifact - dont set a default timeout
Medium: Core: Fix memory leak in subsystem_msg_dispatch() - libxml conversion
Medium: Core: Use crm_element_value() in xml_prop_iter() and xmlGetProp() creates memory to be free'd - libxml conversion
Medium: Core: Post libxml conversion cleanup and fixes
Medium: crmd: Cleanup more memory at exit
Low: Build: Clean out dead files before invoking global
Low: Admin: Add space to node status message
Medium: Core: Avoid pointless copying of XML data - particularly after calls to get_message_xml()
Low: TE: Logging - display the op's magic field for unexpected and foreign events
Low: PE: Logging - make it easier to highlight filter_colocation_constraint logs
High: PE: Anti-colocation with slave shouldn't prevent master colocation
Hg: Merge with upstream
Medium: Core: Bug 1849 - Invalid conversion of ordinal leap year to gregorian date
Medium: PE: Ensure implies_left ordering constraints are always effective
High: PE: Bug PM7 - Ensure masters are always demoted before they are stopped
High: PE: Bug PM12, LF1648 - Extensive revision of group ordering
High: PE: Bug LF1836 - Allow is-managed-default=false to be overridden by individual resources
Low: PE: Turn off debug code
High: PE: Bug LF 1768 - Wait more often for STONITH ops to complete before starting resources
High: crmd: Bug LF1837 - Unmanaged resources prevent crmd from shutting down
Medium: PE: Don't claim to be stopping an already stopped orphan
High: PE: Create probes after allocation to allow smarter handling of anonymous clones
Low: PE: Print class::provider:type not provider::class:type
High: PE: Bug LF185 - Prevent pointless master/slave instance shuffling by ignoring the master-pref of stopped instances
Medium: PE: Reprocess the current action if it changed (so that any prior dependancies are updated)
Medium: PE: Bug PM1 - Failcount applies to all instances of non-unique clone
Low: PE: Removed developmental code
Low: PE: Use the common uber_parent() function
High: PE: Bug PM1 - Fix setting failcounts when applied to complex resources
High: PE: Don't prioritize clone instances that must be moved
High: PE: Bug N-191176 - Implement interleaved ordering for clone-to-clone scenarios
High: PE: Bug N-347004 - Ensure clone notifications are always sent when an instance is stopped/started
Low: PE: Bug N-347004 - New regression tests relevant to HASI
High: PE: Bug N-347004 - Include notification ordering is correct for interleaved clones
Low: PE: pe_order_implies_* for notification ordering is now redundant due to cs: cab8f49ff961
Low: PE: Some extra master/slave debugging
Medium: PE: Correctly set the state of partially active master/slave groups
High: PE: Bug PM-11 - Directly link probe_complete to starting clone instances
Low: PE: Bug PM-11 - New testcases
Low: PE: Print the complex resource's scores too - Used by the showscores script
Medium: CIB: Bug LF-1861 - Filter irrelvant error status from synchronous CIB clients
Low: Admin: Output formatting
Low: crmd: Logging
Low: PE: Downgrade dev logging
Low: PE: Clearly indicate which actions are being cancelled in the dot graph
Medium: crmd: Bug LF-1860 - Automatically cancel recurring ops before demote and promote operations (not only stops)
Low: Admin: Bug N-369914 - Print usage information consistently to stderr/stdout based on exit code
Medium: TE: Bug LF-1859 - Wait for fail-count updates to complete before terminating the transition
Low: PE: Cleanup of how scores are displayed
Low: PE: Make the pengine less verbose at log debug
Low: CIB: Add some debug logging
High: Admin: Bug LF:1848 - crm_resource - Pass set name and id to delete_resource_attr() in the correct order
Hg: Merge with upstream
Low: TE: Resolve compilation issues from previous stable-0.6 merge
Low: PE: Resolve regression test changes resulting from previous stable-0.6 merge
Low: Tools: Start of a new, stack-independant, pingd implementation
Low: Tools: Initial ipv6 implementation
Low: Tools: pingd v2 - Make ipv4 and ipv6 share most codepaths
Low: Tools: pingd v2 - Remove unused code
Low: Tools: pingd v2 - Cleanup and remove unused code
Low: Tools: pingd v2 - Consolidate ping_open() code-path for ipv4 and ipv6
Low: Tools: pingd v2 - More cleanups
Low: Tools: pingd v2 - More cleanups
Medium: Tools: pingd - Implement a stack-independant version of pingd
Low: Tools: Remove development code
Low: TE: Simplify the logic in te_graph_trigger
Low: Core: Reduce logging
Medium: PE: Bug LF:1866 - Restore the ability to have start failures not be fatal
Medium: TE: Bug LF:1859 - Prevent the TE from interupting itself
Medium: TE: Bug LF:1859 - Don't abort graphs due to our own failcount updates
Low: CIB: Remove unused code
High: TE: Repair failcount updates
High: Build: SNMP has been moved to the management/pygui project
High: Core: XML memory corruption - catch, and handle, cases where we're overwriting an attribute value with itself
Medium: cib: General cleanup and implementation of a client-side, file-based CIB
Low: cib: Simplify the creation of cib API objects
Low: cib: Cleanup prototypes
High: TE: Use the correct offset when updating failcount
Low: Tools: Bug LF:1868 - Don't supply parameters to the metadata call
Medium: PE: Bug LF:1866 - Fix version check when applying compatability handling for failed start operations
Low: Core: Add comment on the operation of compare_version()
Low: TE: Repair failcount log message to match what CTS had been looking for
Low: CTS: Use the correct init script for openais
Medium: Core: Drop compataibility code for 2.0.4 and 2.0.5 clusters
Low: CTS: Include a netmask so that findif works with the default route
Medium: crmd: Save the current CIB contents if we detect the PE crashed
Low: PE: Reduce debug logging
Low: PE: Remove whitespace
Medium: PE: Prevent a possible use-of-NULL
High: PE: Allow the cluster to make progress by not retrying failed demote actions
Low: PE: Regression tests - Fix op_digests so that the definitions only desired test actions are detected as changed
Low: PE: Downgrade a sanity check to a warning
High: PE: Implement promote/demote ordering for cloned groups
High: PE: Group start ordering fixes
Low: PE: Use pe_order_implies_*_printed for clone meta actions
Low: PE: Use pe_order_implies_*_printed for master/slave meta actions
Low: PE: Reduce developmental logging back to normal levels
Low: cts: Make disk-space and logging checks into audits
Low: cts: use remote_py when making lrmadmin calls
Low: cts: Re-implement primitve and group audits in a way that doesn't require parsing the CIB
Low: cts: Run the disk and logging audits before the scenario setup
Low: crmd: Extra logging when nodes are removed from join consideration
Low: PE: Downgrade developmental logging
Low: cts: Add some colocation constraints to the generated CIB
Low: cts: Implement colocation constraint auditing
Low: cts: Simplify the Standby test
Low: cts: Make active_resources() and ResourceLocation() detect resources in master/slave mode
High: crmd: Prevent use-after-free in lrm interface code (Patch based on work by Keisuke MORI)
Low: cts: Account for some strange error codes comming from remote_py
Low: cts: 1 clone instance per node is enough for the master resource
Medium: Admin: Add a summary output that can be easily parsed by CTS for audit purposes
Low: cts: Handle resources that aren't installed in the local lrm
Medium: PE: Indicate each resource's 'promotion' score
Low: PE: Allow some log levels to be tuned together
Low: cts: You can't assume resources will move back after a node is taken out of standby
Low: Increase the 'last' version number
Low: PE: Turn down logging
Low: PE: Logging improvment
Low: cts: Only have one master
Low: Build: Change name of Makefile helper target
Low: PE: Indicate which node the promotion score applies to
High: PE: Fix error in previous commit that allowed more than the required number of masters to be promoted
Use a consistent logging format for allocation/promotion scores
Low: cts: Fix log message
Low: cts: Be more informative about cluster partitions
Low: Build: Fix the date format for changelogs
Low: Build: Update .changes in the 'home' Makefile target
Low: Build: Fix alignment of changelog stats
Medium: Admin: crm_mon - Only complain if the output file could not be opened
High: PE: Make sure 'slave' monitoring happens after the resource has been demoted
Low: TE: Slight logging improvement
Low: cts: slight change for clarity
High: SNMP: Remove snmp from the project - it's maintained in pygui/management now
Hg: Merge with upstream
Low: Resolve previous merge
Medium: Core: Support value='value+=N' XML updates (in addtion to value='value++')
Low: PE: Extra regression tests
Low: Hg: Merge from stable
High: PE: Replace failure-stickiness with a simple 'migration-threshold'
Low: Core: xml - Only complain when erasing a set attribute with crm_xml_add()
High: cib: Fix the server's replace operation after the previous code re-org
Medium: Core: Catch empty replies from msgfromIPC() in xmlfromIPC()
Low: cib: Remove some debug logging
Low: cib: Remove some debug logging
Medium: PE: Implement a global 'stop-all-resources' option
Low: crmd: Remove testing code
High: cib: Repair receiving of cluster messages after conversion to xmlNode
Low: Core: Some better logging when msgfromIPC() returns NULL
High: PE: Prevent failure stickiness underflows (where too many failures become a _positive_ preference)
Medium: Core: ipc - Better interpretation of msgfromIPC_timeout() failures
Medium: ipc - Consistent handling of xmlfromIPC() 'failures'
High: crmd: Switch to DC-only updating of node names and status
Low: Build: Helper target to see whats new for the next stable _series_
High: PE: Ensure 'master' monitor actions are cancelled _before_ we demote the resource
Hg: Merge with upstream
Low: PE: Remove debugging
Medium: PE: Create orphan cancelation actions with the operation_key of the action to be cancelled
Low: cts: Some extra debugging information
High: crmd: Bug LF:1881 LF:1882 - Overhaul the logic for operation cancelation and deletion
Hg: Merge with upstream
Low: PE: Some minor logging changes
Low: Core: Getting a NULL message from msgfromIPC_noauth() is quite normal on disconnect
High: PE: Bug LF:1459 - Allow failures to expire
High: PE: Bug LF:1459 - Hook up the option to configure how long to wait before considering a failure to have expired
Medium: PE: Bug LF:1461 - Give transient node attributes (in <status/>) preference over persistent ones (in <nodes/>)
Low: PE: Reduce development logging
Medium: PE: Fix the standby when it's set as a transient attribute
Medium: PE: Bug LF:1328 - Don't fencing nodes in clusters without managed resources
Medium: Admin: Bug N:371785 - crm_resource -C doesn't also clean up fail-count attributes
Low: crmd: Bug N:239085 - probe_complete causes transition abort
Low: PE: Move stickiness and failure migration calculation to the allocation 'section'
Medium: PE: Use a consistent date for expiring failed operations
Medium: PE: Also allow fail-counts to expire
Medium: TE: Bug LF:1875 - Correctly find actions to cancel when their node leaves the cluster
Low: PE: Make the retrieval of fail-count data into a common function
Medium: Admin: Bug LF:1883 - crm_mon should display fail-count and operation history
Medium: Admin: Bug LF:1883 - crm_mon should display operation timing data
Medium: TE: Include target rc in the transition key
Low: Build: Fix creation of LCRSODIR
Medium: admin: Repair the ability to query failcount
High: ais: Have the plugin create needed /var/lib directories
High: Admin: Repair the ability to delete failcounts
Low: PE: Reduce logging
Low: PE: Make sure that we don't break asymmetric clusters - always require explicit location constraints
Medium: PE: Prevent a possible use of NULL in sort_clone_instance()
Medium: Bug LF:1900 - Resource stickiness should not allow placement in asynchronous clusters
Medium: PE: Bug LF:1906 - Remove an optimization in native_merge_weights() causing group scores to behave eratically
Low: PE: Reduce logging
High: PE: Ensure groups are migrated instead of remaining partially active on the current node
Dev: PE: Fix group migration when the last child in the group fails
Dev: PE: Fix group migration yet again
High: PE: Apply (failure-)stickiness values to the whole group, not the individual resources
High: PE: Apply only stickiness values (not failure-stickiness) to the whole group, not the individual resources
High: PE: Bug N:396293 - Enforce manditory group restarts due to ordering constraints
Dev: PE: Reduce developmental logging
High: ais: Make sure the sync and async connections are assigned correctly (not swapped)
High: PE: Correctly recover master instances found active on more than one node
Medium: TE: Include target rc in the transition key
Low: cts: Allow the warnings about inactive nodes to be turned off
Hg: Merge with upstream
High: TE: Parse the transition key correctly
High: TE: Parse the transition key correctly
High: PE: Use the correct validation function for default-failure-timeout
Low: Admin: crm_mon - Fix the formatting of operations
High: PE: Fix assert failure leading to core dump - make sure variable is properly initialized
Hg: Merge with stable-0.6
Medium: cib: Start implementing the ability for clients to connect to remote servers
Medium: cib: The ability for clients to connect to remote servers is now functional
High: cib: Allow the remote password to be read from stdin (no local echo)
Low: cib: Update the PAM file for suse
High: Core: Allow sending TLS messages larger than the MTU
Medium: cib: cib_remote - Delay asking for password until signon()
Medium: Admin: crm_mon - Make displaying of operation timing data optional
Medium: Admin: crm_mon - Use synchronous queries to allow use with remote CIB connections
Low: Admin: crm_mon - cleanup the curses printing
Low: Admin: crm_mon - remote connections are in the 'connected command' state
Medium: cib: Simplify signon and registration code paths
Low: cib: Remove developmental logging
Medium: cib: Implement callbacks and async notifications for remote connections
Medium: cib: Remove some useless parts of the API
Medium: Common: filter_action_parameters - enable legacy handling only for older versions
Low: Build: Calculate the diffstat for new features
Low: crmd: Use the standard define
Low: crmd: Reduce the number of files that are always rebuilt by 1
Medium: PE: Bug LF:1884, Bug LF:1885 - Implement N:M ordering and colocation constraints
High: Admin: All tools now use --node (-N) for specifying node unames
High: Admin: All tools now use --xml-file (-x) and --xml-text (-X) for specifying where to find XML blobs
Medium: Admin: Fix ptest invocation
Medium: PE: Support zero-or-more 'resource_set's in colocation and ordering constraints
Medium: PE: Reduce the complexity of N:M ordering by adding pseduo ops
Low: PE: Use action id's that wont cause conflicts when we invert the ordering constraint
Medium: PE: Allow the new 'resource_set' ordering constraints to be inverted automatically
Low: TE: Increase log level of the graph being used
High: TE: Update the last-failure with the correct value
Low: Build: Update the last release
Low: PE: Update the testcases so that the correct digest is created (now that filter_action_parameters() was fixed in cs: 174b23b04970)
Medium: cib: cib_file - don't allow connection to missing, empty or invalid configurations
High: cib: Implement CIB_shadow - a facility for making and testing changes before uploading them to the cluster
Low: PE: Better logging of what resources are actually doing
Medium: PE: Fix the set_role() logic for fixing resource states
High: PE: Bug N:387749 - Colocation with clones causes unnecessary clone instance shuffling
Low: PE: Bug N:387749 - Add regression test
Low: PE: Update regression tests after last couple of commits
Medium: PE: Use the correct node when logging resource stop events
High: crmd: Bug LF:1894 - Make sure cancelled recurring operations are cleaned out from the CIB
Medium: Core: Add support for validating xml against RelaxNG schemas
High: Core: Implement a more flexible set of functions for validating configurations
High: Core: Don't free xmlDoc after validation as its now attached to the xmlNode
High: cib: Try and upgrade to the latest schema when the CIB is promoted to master mode
Low: Core: Switch back to the only functional validator
Medium: xml: Install versioned schemas
Low: xml: Add N:M constraints to the schema
Low: xml: Fix closing tag
Low: xml: Use a common #define for 'validate-with'
Low: Build: Include xml in the list of subdirs
Medium: xml: Ship an xslt for upgrading from 0.6 to 0.7
High: Core: Add a facility for automatically upgrading old configurations
Low: Build: Include header changes for previous commit
Low: xml: Reduce copying when validating xml
Low: xml: Reduce copying when transforming xml
Low: TE: Remove dead file
Low: Build: Mark new series as unstable
Medium: crmd: Reduce the amount of xml copying occuring
Low: PE: Avoid pointless xml copying
Low: cib: Avoid pointless xml copying
Low: cib: Avoid more pointless xml copying
Low: dtd: Bug LF:1895 - Remove erroneous comments
High: crmd: Always send a direct ack as the TE doesn't see delete's
High: PE: Have the PE do non-persistent configuration upgrades before performing calculations
Low: PE: Simplfy transition_id logic
Low: admin: Fix segfault introduced in recent commits
Low: cib: master now needs the config so that it can set the correct schema version
Medium: cib: Only send registration messages if the connections are active
Medium: crmd: Cleanups needed now that we don't needlessly copy message inputs
Medium: crmd: Cleanups needed now that we don't needlessly copy message inputs
Low: Admin: Repair crm_resource - use a valid connection type
Low: TE: Remove dead file
Low: cib: Remove a redundant hack for the TE
Low: Core: Add some extra log information when crm_update_peer_proc fails
High: cib: Remove all trace of the now default *_synchronous connection types
Low: crmd: Restore the ability to use valgrind with the PE/TE (broken last friday)
Medium: ais: Do plugin initialization in .exec_init_fn instead of .config_init_fn
High: cib: Remove all trace of the now default *_synchronous connection types
Medium: xml: Add validate-with as a top-level CIB attribute
Low: xml: Revert transition_magic back to being implied for compatability with pacemaker-0.6
High: A new tool for setting up and invoking CTS
Low: xml: Add the crm dtd back to the project
Medium: Build: Install new cib headers
Low: cts: Help users create a key pair
Medium: Build: Remove reference to an internal header by a public one
Medium: Build: Remove reference to an internal header by a public one
High: PE: crmd: Have the crmd sign in to the PE instead of the other way around
High: TE: Simplify the design by folding the TE process into the crmd
Low: TE: Remove the last traces of the TE
Low: Core: Disable the new schema for now
Medium: crmd: There are some states we should not run the TE graph in
Low: crmd: Remove dead code
Low: cts: Remove references to the TE
Low: cts: Fix interpretation of lrmadmin return codes
Low: cts: Fix comment
Low: PE: Enable stderr if ptest is invoked with -s
Medium: PE: Bug N:385265 - The failure stickiness of group children is ignored until it reaches -INFINITY
High: cib: Make registering per-op callbacks an API call and renamed (for clarity) the API call for requesting notifications
Low: PE: Bug N:385265 - Add a new regression test
Low: Build: Always create LCRSODIR to make building on debian easier
Low: admin: crm_failcount - Detect invalid scores (patch from Junko IKEDA)
Low: crmd: Remove dead code
Low: crmd: Simplify the starting of transitions now that the TE has been merged
Low: cts: relax a search pattern
Low: ais: Disable config reading until the API is sorted out
Hg: Merge with stable-0.6
Medium: PE: Implement master and clone colocation by exlcuding nodes rather than setting one's score to INFINITY (similar to cs: 756afc42dc51)
Low: Build: Helper target to beautify the code
Low: Build: Fix typo
Low: Build: More indent options
Low: Build: Fix typo
Low: Build: More indent options
Low: crmd: Inhibit notifications for 'node probed' CIB updates as they needlessly cause extra transitions
Low: cts: By default, do _not_ warn about inactive resources
Low: Core: Remove note-to-self logging
Low: cts: Reduce debug logging
Low: cib: Reduce callback logging
Low: cib: Reduce logging of operation timeing data
Low: crmd: Remove dead code
Low: crmd: Create a new default graph completion action 'tg_done'
Medium: crmd: There is no need to wait for CIB updates to complete before signaling a transition is complete
Hg: Merge from stable-0.6
Low: PE: Merge two log messages
Low: PE: Clean up a log message
Low: PE: Use the standard #defines for resource options
Low: cib: Reduce logging
Low: Core: Reduce logging in do_id_check()
Low: crmd: Add whitespace to log message
Low: cts: Don't generate resource IDs that start with a number - its not legal XML
Low: Build: Fix installation of the lcrso dir
Medium: Remove old compatability code for resource ops
Medium: xml: Continuing evolution of the new schema and improvements to the XSLT upgrade capabilities
Low: PE: Use standard #defines for various attributes
PE: Low: Fix the automatic upgrading of configurations
PE: Low: ptest - Do automatic upgrading of configurations
Low: PE: Make the variables used in rsc_order constraints more logical
Low: PE: Cleanup of the standard #defines
Low: PE: Repair ordering constraints
Low: PE: failure stickiness has been replaced by migration-threshold
Low: PE: Remove invalid constraint
Medium: PE: We don't allow colocation with specific clone instances
Medium: PE: We don't allow colocation with specific clone instances
Medium: PE: We don't allow colocation with specific clone instances
Medium: PE: The xml ID type doesn't allow resources that start with a number
Low: Core: Expose the sorted_xml() function
Medium: PE: Create action attributes in sorted order
Medium: PE: Create action attributes in sorted order
Low: xml: Fix the normalization of rsc_order constraints
Low: PE: Remove the operation's ID from the meta data
Low: PE: Remove the operation's ID from the meta data and don't set intervals for non-recurring ops
Low: Core: Ensure 'meta' attributes always use underscores
Low: PE: Use a single code-path when adding meta attributes to the action
High: Core: Activate the new configuration syntax and RelaxNG validation/conversions
Medium: crmd: Ensure we process CIB updates until the transition graph is complete
Hg: Merge in new configuration syntax
Low: Core: Simplify logic
Medium: cib: Implement get_object_root() using Xpath
Medium: cib: Remove dead code - cib_pluralSection()
Medium: cib: Add sections for op_defaults and rsc_defaults
Low: xml: Add the two new cib sections to the schema
Medium: xml: Never modify the id of a primitive resource - even if it's invalid
Low: PE: Update the expected graphs for the new syntax
Low: cib: Do all validation with DTDs and schemas
Medium: cib: Remove compatability code for Heartbeat 2.0.4
Medium: cib: Make cib->cmds->update() an alias for modify at the API level (also implemented in cibadmin)
Low: Simplify the implementation of create and delete_absolute
Low: Core: Abstract the xpath searching into its own function
High: cib: Cleanup the API - remove redundant input fields
Low: xml: List the valid schemas
Low: xml: Set a valid schemas after upgrade
High: Implement XPath based versions of query/delete/replace/modify
Low: crmd: Implement two deletions as being xpath based
Low: Core: Dont abort if the user specifies an invaid XPath expression
Low: Core: Don't tollerate XML objects without names
Medium: cib: Callers of xpath_search() need to check if it returned xmlDocPtr or xmlNodePtr :-/
High: cib: file - Give callers a copy of the output and don't free query results - its the real CIB
Low: crmd: Make the compiler happy by passing string literals to snprintf
Low: Admin: XPath based commands aren't inherently dangerous, don't require --force
Low: Core: Fix some sanity checks in dump_data_element()
Medium: Core: Allow more than one item to be modified/deleted in XPath based operations
Low: crmd: Remove more 2.0.4 code
Medium: crmd: Also watch link status for node leaving events
Low: cib: Remove dead function reference
Low: cib: Make validation messages more generic
Low: xml: Smarter conversion of top-level CIB attrs
Low: Core: Test that the xml to dump _isnt_ NULL
Low: cib: Cosmetic change
Medium: xml: Create an intermediate DTD that will allow the new syntax in key places
Low: cts: Specify how to validate the configuration
Low: Remove ccm_transition, num_peers and cib_feature_revision from the CIB
Low: cib: Remove section checks now that they can contain XPath exressions
Low: cib: Update the validation logic when a new configuration is read from disk
Low: crmd: Stop using cheap hacks to remove the lrm and transitent_attrs sections
Low: xml: Tweak the transitional DTD
Low: Core: Remove unneeded compatability workaround for validation
Low: cib: Adjust the allowed call types now that 'update' has been removed from the table
Low: xml: Simplify the xslt conversion
Low: cib: Adjust the section paths to avoid getting an xmlDocPtr back
Low: cib: Add a comment indicating when we get an xmlDocPtr from xpath_search()
Medium: cib: Disable the no-quorum filter for cib changes
Low: Remove whitespace
Low: Core: Validate input to search_xpath()
Medium: crmd: Handle invalid transition graphs from the PE
Low: xml: Use a valid value as the default for remote-tls-port
Low: xml: Tweak the top-level CIB attributes
Low: PE: Call set_working_set_defaults() before we potentially set data_set.graph
Low: PE: s/last_run/last-run/
Low: PE: Provide more details when assertions in sort_op_by_callid() fail
Low: Core: Fix up the free_xml macro
Medium: cib: New call option 'cib_can_create' which can be passed to modify actions - allows the object to be created if it doesn't exist yet
Low: crmd: Use 'cib_can_create' to allow node_state updates to create the entry if needed
Low: cts: Log inactive resources
Low: crmd: Implement erase_status_tag() with a delete op
Low: cib: Remove the useless 'generated' attribute
Low: cib: Return the whole CIB if the section is invalid - makes calls to cib_pre_notify() easier
Dev: Core: Fix update_validation() when transform=TRUE
Dev: cib: Implement the XPath 'create', fix XPath 'replace' and allow 'modify' to create the object if it doesn't exist yet
Dev: cib: Fix compiler warning
Dev: cib: Fix compiler warning
Dev: crmd: Defer local shutdowns until the rest of the graph has completed
Low: crmd: Remove dead code
Low: crmd: Remove dead code
Low: cts: Remove a stray character in BadNews pattern
Low: crmd: Improve logging - log shutdown actions as text not integers
Low: crmd: Fix log message
Dev: crmd: Fix the filtering of cib updates when there is an active transition
Dev: crmd: Make sure we don't 'start' the transitioner if its already active
Low: cib: Indicate the ID of the client changing their notifications setup
Low: Build: Link crmcommon against -lgnutls so everyone else doesn't have to
High: PE: Ensure groups are migrated instead of remaining partially active on the current node
Dev: Build: Fix the building of STONITH-related items
Dev: PE: Fix group migration when the last child in the group fails
Dev: PE: Fix group migration yet again
High: PE: Apply (failure-)stickiness values to the whole group, not the individual resources
Hg: Merge with upstream
High: PE: Apply only stickiness values (not failure-stickiness) to the whole group, not the individual resources
High: PE: Bug N:396293 - Enforce manditory group restarts due to ordering constraints
Medium: cib: Mark the CIB as disconnected if the IPC connection is terminated
Medium: cib: Better matching notification callbacks (for detecting duplicates and removal)
High: Overhaul the detection of nodes that are starting vs. failed
Hg: Merge with upstream
Low: Build: Fix location of stonithd library
Dev: crmd: Use the correct destination variable 'crm_msg_lrmd' not 'crm_proc_lrmd'
High: Tools: Use the correct destination variable when sending attrd messages - 'crm_msg_attrd' not 'crm_proc_attrd'
Hg: Merge with upstream
Dev: Tools: Use the correct function for free-ing messages
Low: Build: Include ais_common.h
Dev: PE: Reduce developmental logging
Low: Admin: Use a less convoluted message for pending nodes
Medium: PE: Disable the group_merge_weights() hack from stable-0.6
High: Core: Smarter handling of XML values containing quotes
Dev: Core: Fix smarter handling of XML values containing quotes
Dev: Core: Add some extra comments
Low: cts: Updates for newer openais builds
Medium: crmd: Add some extra paranoia checks to do_election_count_vote()
Low: cts: Turn on stonith testing for crm-ais
Medium: crmd: Ignore num-updates when deciding which node has the 'best' copy of the config
Medium: Core: Update the CRM feature set to 3.0
Low: ais: Remove duplicate header imports
Low: ais: Remove duplicate header imports
Medium: ais: Make the relevant parts of openais headers available when building for whitetank
Medium: ais: Make the relevant parts of openais headers available when building for whitetank
High: ais: Make sure the sync and async connections are assigned correctly (not swapped)
High: ais: Make sure the sync and async connections are assigned correctly (not swapped)
Low: ais: Check the IPC ack we get is of the size we expect
Low: ais: Check the IPC ack we get is of the size we expect
Low: ais: Check the IPC ack we get is of the size we expect
Medium: ais: Make the relevant parts of openais headers available when building the plugin for whitetank
Medium: ais: Make the relevant parts of openais headers available when building the plugin for whitetank
Medium: ais: Changes to allow building against whitetank - the stable series of openais
Medium: ais: Changes to allow building against whitetank - the stable series of openais
Medium: crmd: If the CIB was replaced as part of the join - there is no need to start another one
Low: cts: Allow the list of nodes to be specified on the command line
Low: cts: Allow the list of nodes to be specified on the command line
Dev: ais: Start the stonith subsystem before the crmd so that it also stopped _after_
Dev: ais: Start the stonith subsystem before the crmd so that it also stopped _after_
Low: ais: Enum whitspace
Low: ais: Enum whitspace
Medium: ais: Add attrd to text2msg_type()
Medium: ais: Add attrd to text2msg_type()
Low: Build: Fix install permissions for stonithd
Low: Build: Fix install permissions for stonithd
Dev: ais: Make the process start order adjustable without creating incompatibilities between versions
Dev: ais: Make the process start order adjustable without creating incompatibilities between versions
Low: Core: Reduce logging
Low: Core: Reduce logging
Dev: ais: Re-enable support for transient clients
Dev: ais: Re-enable support for transient clients
Hg: Merge with upstream
Dev: ais: Adjust the child start order
Low: cib: Tweak the filtering of cib_diff_resync return values
Hg: Merge with stable-0.6
Low: cts: Be less verbose for failed CIB audits
Low: cts: Only perform stonith (no 'd') tests for non-crm clusters
Low: cib: Log the operation that caused the disk write
Medium: cib: Detect updates that decrease the version tuple
Medium: cib: Don't set version details in blank CIBs - they cause the version to go backwards
Low: crmd: Only send the top-level cib object - no need to create a whole blank CIB
Low: PE: Go back to using '_' for crm-feature-set, increase it to 3.0
Low: PE: Update test case
Dev: Core: Fix detection of invalid transition keys
Medium: PE: Make use of target_rc data to correctly process resource operations
Low: PE: Testcases - we don't trust the op_status, set rc_code to indicate failure
Medium: PE: Make effective use of target_rc information when processing resource operations
Hg: Merge from stable-0.6.5
Hg: Merge from stable-0.6.5
Medium: Core: Improved decoding of the different transition_key formats
Low: PE: Reduce logging
Medium: PE: Avoid clone instance shuffling in more cases
Low: PE: Logging tweaks
Medium: Core: Fix the sort_pairs function for creating sorted xml objects
Low: Logging
Dev: crmd: For now revert a previous commit that stopped considering num_updates when deciding who had the "best" CIB
Low: PE: Update the valgrind exceptions file for ptest
High: PE: Fix memory leaks reported by Valgrind
Hg: Merge with stable 0.6.5
Dev: ais: Interacting with objdb from whitetank during crm_config_init_fn prevents openais from fully starting
Medium: Admin: crm_mon - Misc improvements from Satomi Taniguchi
Medium: Admin: crm_mon - Cleanup on exit patch based on the work of Satomi Taniguchi
Medium: ais: Allow clients to retry sign-in up to 30 times when openais returns SA_AIS_ERR_TRY_AGAIN
Dev: ais: Correctly terminate the connection attempt when we exceed the retry threshold
High: ais: Dampen membership changes when we loose more than half the known nodes
Low: crmd: Remove redundant code - this is now handled in our ais glue
Hg: Merge woth stable-0.6.5
Low: Highlight bad partition details
High: cib: Correctly detect configuration changes - num_updates does not count
Medium: cib: crmd: Logging reduction and consolidation
Hg: Merge from stable-0.6.5
Medium: cib: core: Logging reduction and consolidation
High: ais: Audit IPC handling between the AIS plugin and CRM processes
Dev: PE: Correctly handle probes that find active resources
Low: PE: Add regression test for active resources at startup
Low: crmd: More logging reductions
Hg: Merge with stable-0.6.5
Dev: PE: Bug 1920 - Correctly handle probes that find active resources (including masters)
Hg: Merge with stable-0.6.5
Low: cts: Fix the They_stopped pattern
High: PE: Bug N:385265 - Ensure _all_ groups (even ones no colocated with anything) are migrated instead of remaining partially active on the current node
Low: PE: Bug N:385265 - Add regression test
Medium: ais: Don't try to uncompress empty messages
Hg: Merge from stable-0.6.5
Dev: PE: Disable hack from stable-0.6
Low: Admin: crm_mon - Code consolidation
Low: Build: Update versions
Dev: Admin: Teach crm_mon to upgrade old configs
High: cib: Don't upgrade configurations automatically - add a new API call for the admin to use
Medium: Tools: cibadmin - Add support for the upgrade API command
Medium: cib: Reimplement get|set|delete attributes using XPath
Low: cib: cib_file - Log the change diffs
Dev: Core: List new tags that don't require IDs
Low: cib: avoid querying the validation type again in validate_xml
Low: cib: Use the standard define for deciding what type of validation to use
Medium: cib: Support setting options in the new rsc_defaults and op_defaults sections
Medium: cib: BUG LF:1918 - By default, all cib calls now timeout after 30s
Low: Admin: Make find_resource_attr able to look in for any type of resource attribute (meta or otherwise)
Medium: Admin: Bug LF:1708 - crm_resource should print a warning if an attribute is already set as a meta attribute
Medium: Core: Create a common function for CLIs using the PE libraries to upgrade the configuration
Medium: PE: Bug LF:1886 - Create a resource and operation 'defaults' config section
Medium: cib: Remove the 'attributes' scaffolding from the new format
Dev: PE: Use the action's existing hashtable
Dev: admin: crm_resource - query the CIB when looking for IDs to avoid problems with renaming
Medium: crmd: Ensure joins are completed promptly when a node taking part dies
Merge in the stable-0.6.5 branch
Merge with dejan
Low: cib: Allow non-recurring querys (return just the matched element, no children)
Dev: cib: Figure out what kind of attribute update the CIB will accept before we send it
Low: Admin: Add support for new cib options
Low: cib: Dont log query results
Low: cts: Better formatting of 'running test X' messages
Medium: crmd: Add support for lrm_ops->fail_rsc if its available
Dev: PE: Correctly process attribute sets now that they no longer have the 'attributes' scaffolding
Dev: Admin: Repair the ability to set master/slave attributes when in compatability mode
Dev: Admin: crm_resource - repair the ability to modify existing resource attributes
Dev: Admin: crm_resource - Find master/slave attributes in either the new or old syntax
Low: cib: Don't log failed queries
Dev: cib: Only set a default timeout if msgfromIPC_timeout() is available
Low: cts: Tweak the 'Running test' log message some more
Low: admin: ccm_tool - Allow the cluster type to be specified
Low: cib: Reduce logging
Low: cts: Tell ccm_tool which cluster type to use
Low: crmd: Cleanup log message
Medium: crmd: Bug LF:1924 - Improved handling of lrmd disconnects and shutdowns
Low: crmd: Improved logging of resource actions
Medium: crmd: Remove some HA2.0.3 compatability code
Low: cts: Tell ResourceRecover the new log patterns to look for
Medium: admin: Teach crm_verify about schema versions
Medium: Core: Teach validate_xml how to validate 0.6 based configurations (which dont set XML_ATTR_VALIDATION)
Medium: PE: Bug LF:1926 - Recurring actions should be started for active (but unmanaged) resources
Hg: Merge with upstream
Dev: Clean up after previous merge
Low: cts: right justify the test number
Low: cib: Let validate_xml chose the validation method - it has built-in smarts for older versions
Low: cts: Wait for 60s before timing out the status command
Low: PE: Remove deprecated code - this is handled by the DTD/schema anyway
Low: ais: Downgrade logging now that I know it works
Low: cib: Remove dead code
Dev: Build: Remove unused variable
Low: PE: Update testcases after merge with 0.6
Dev: xml: Upgrade attribute sets correctly (include rules and scores)
Low: PE: Update the testcase s/failure-stickiness/migration-threshold/
Low: PE: Update the expected outputs (sorting and crm version)
Medium: cib: Bug LF:1348 - Allow rules and attribute sets to be referenced for use in other objects
Medium: PE: Bug LF:1926 - Probes should complete before stop actions are invoked
High: Core: Fix parsing of time-only ISO dates
Low: Core: Some new ISO date helpers
Medium: crmd: Do not wait for actions with a start_delay over 5 minutes. Confirm them immediately
Medium: Core: iso8601 - Implement subtract_duration and fix subtract_time
Medium: PE: Bug LF:1892 - Allow recurring actions to be triggered at known times
Low: ais: Reduce logging
Low: ais: Default to 'debug 0'
Medium: ais: Read options from the environment since objdb isn't completely usable yet
Low: cts: Make sure we test the use of ISO8601 durations for operation intervals
Low: cts: Fix extracttests for the new log format
Low: ais: Only log nodes in crm_update_peer() when they change
Low: Admin: crm_mon - Include each node's pingd score
Low: PE: Downgrade dev logging
Medium: Build: Fix building when libltdl is not pre-installed
Low: Build: Remove unused headers
Medium: PE: Prevent use-of-NULL if anyone ever passes NULL to native_assign_node()
Medium: PE: Print allocation scores after target role has been taken into account
Low: PE: Fix testcase - colocating a clone with a primitive makes no sense
High: PE: Enforce mandatory colocation for clones
Hg: Merge in fixes for mandatory colocation constraints involving clones
Hg: Merge with Dejan
High: PE: Apply stickiness after location constraints are applied so that they work in asymetrical clusters
Low: crmd: Remove pointless logging
Low: PE: New testcase for recent commit
Low: PE: New testcase for recent commit
Medium: PE: crmd: Allow pending operations to be optionally recorded in the CIB
High: PE: Bug LF:1932 - Probes that find resources misconfigured or not installed should not mark them as running
Low: cts: Start a remote access port by default
Medium: PE: Tweak the rsc_order syntax s/(first|then)-rsc/(first|then)/
Hg: Merge from stable-0.6
Dev: xml: Repair conversion of rsc_order constraints
Low: cts: Improve oprofile support
Low: PE: Downgrade error message to a warning
Low: Core: Various changes to aimed at making key functions more efficient
Low: admin: Remove dependancy on cl_malloc
Low: cib: Remove reference to cl_malloc header
Low: Core: Cleanup alloc/free macros
Low: ais: Remove dead code
Dev: crmd: Make sure the right function checks for the XML_OP_ATTR_PENDING option
Low: cts: Tweaks
Medium: crmd: Attempt to streamline some key functions to increase performance
Medium: cib: Attempt to streamline some key functions to increase performance
Medium: Core: Attempt to streamline some key functions to increase performance
Medium: cib: Skip construction of pre-notify messages if no-one wants one
Medium: Core: Some performance enhancements to subtract_xml_object
Dev: Core: Do not write past the end of the digest array
High: ais: Log terminated processes as an error
Dev: Core: expand_plus_plus - Don't read past the end of the value array
Dev: crmd: Sanity check the list of parameters in append_restart_list
Low: crmd: Make an error message more helpful
High: crmd: Enable the do_dc_join_offer_one code which results in significant performance improvement
Low: cts: Make sure we always catch terminated processes in BadNews
High: cib: Reorganize things to avoid calculating the XML diff twice
Dev: cib: Remove development logging
Hg: Merge in a new testcase from 0.6.5+
Medium: PE: Prefer the current location if it is one of a group of nodes with the same (highest) score
Hg: Merge in stable-0.6
Medium: admin: crm_resource - create more compact constraints that don't use lifetime (which is deprecated)
Low: admin: Whitespace
Low: cts: Increase batch limit to 10
Dev: cib: Calculate the diff before it is free'd by activateCibXml
Medium: cib: Allow cib_config_changed to return a full diff
Low: Core: Streamline crm_xml_add and expand_plus_plus
Low: cib: Streamline the cib code by removing the per-action CIB code
Low: cib: Speed up notifications by checking if they're required
Medium: cib: Avoid recalculating the change diff
Low: Core: Optimize the logging macros
Low: cib: Optimize activateCibXml
Low: cib: Optimize cib_get_operation_id
Dev: cib: Use the result of strcmp correctly
Dev: cib: Check for the right global update operation types (the list was rearranged)
Low: cib: Optimize key functions for performance
Low: PE: Don't save input files if max == 0
Low: More performance motivated changes
Low: crmd: Optimize some key functions
Dev: crmd: Fix detection of complex actions
Low: crmd: Only wait 3s for the PE to come up
Low: crmd: Throttle election wins to one per second
Medium: crmd: Detect and optimize the crm's behavior when processing diffs of an LRM refresh
Dev: crmd: append_restart_list - Don't free restart_list, it's cached. Fix check for params.
Hg: Merge in performace sub-branch
Hg: Merge in stable-0.6 changes
Medium: cib: Teach cib-file to read/write compressed files
Medium: crmd: Disable do_dc_join_offer_one, it's not quite ready for prime time and the performance impact isnt as great anymore
Dev: cib: cib_perform_op - Make sure the diff has the versions set in all the right places
Low: Core: crm_element_value - Remove a layer of indirection
Low: cib: Never massage the diff versions of global updates
Low: crmd: Ensure the cause of transition aborts is always seen
Low: crmd: Better logging of the cause of transition aborts
Dev: crmd: Fix a NULL that became CamelCased
Dev: cib: Only set the epoch on the cib-object if it changed
High: PE: Make OCF_ERR_ARGS a node-level error rather than a cluster-level one
Low: Build: Update version numbers
Low: ais: Give some context as to which message was not sent
Low: crmd: Simplify detection and processing of non-status changes
Hg: Merge from stable-0.6
Low: Core: Add a function for printing XML diffs to stderr/stdout
Medium: admin: cib_shadow - Implement a --diff option
Medium: admin: cib_shadow - Implement a --switch option
Dev: crmd: Disable election optimization with unintended side effects
Low: Core: Allow crm_int_helper to parse long longs
Dev: Core: crm_int_helper - Only print errors if there was one
Medium: ais: Approximate born_on for OpenAIS based clusters
Medium: crmd: Enable born_on as an election criteria for OpenAIS based clusters
Medium: admin: cib_shadow - Implement --locate to display the underlying filename
Low: cib: Indicate why an upgrade failed
Low: cib: Indicate why an upgrade failed
Low: xml: Move the upgrade XSLT to its final name
Medium: admin: cibadmin - Show any validation errors if the upgrade failed
High: ais: Change the tag for the ais plugin
Merge from stable-0.6
Low: admin: Remove development logging
Dev: Build: The upgrade script was renamed
Low: ais: Fix compiler warning about printing uint64_t
Low: admin: Fix spelling in help message
Low: ais: Remove unneeded and incorrect definition
Low: ais: Fix compiler warning about printing uint64_t
Low: PE: Clones and their children cant have the same ID
Low: PE: Reduce logging
Low: PE: Free input when validation fails
Medium: Core: Clean up XML parser after validation
Low: PE: Update log message
Dev: Core: Cleanup memory in crm_get_interval()
Medium: Fix memory leaks when resetting the name of an XML object
Low: PE: Ensure the XML code cleans itself up after every run
Low: PE: do_calculations already calls set_working_set_defaults. Doing so again causes a memory leak.
High: PE: Fix memory leak in action2xml
Low: PE: Make sure the test inputs validate without needing to be massaged by do_id_check()
Medium: cib: Remove do_id_check, it is a poor substitute for ID validation by a schema
Low: core: use our own glib log handler
Low: core: use our own glib log handler
Low: PE: Make unpacking attributes a little more obvious
High: PE: Properly handle clones that are not installed on all nodes
Low: PE: Glib makes it impossible to check for leaks with --show-reachable=yes
Hg: Merge from stable-0.6
Medium: PE: Prevent multiple calls to set_working_set_defaults from leaking memory
Low: PE: Update the valgrind suppression file
Low: Core: Another cast when printing uint64_t
Low: PE: Fix memory leak when set_working_set_defaults was called too many times
Dev: PE: Revert prior change that involved reading potentially uninitialized memory
High: PE: Bug LF:1941 - Handle failed clone instance probes when clone-max < #nodes
Low: cib: Explicitly include GSource.h for DEFAULT_MAXDISPATCHTIME
High: PE: Bug LF:1941 - Handle failed clone instance probes when clone-max < #nodes
High: admin: Fix use-after-free in crm_mon
Medium: PE: Bug LF:1941 - Mark extra clone instances as orphans and don't show inactive ones
Medium: PE: Bug LF:1941 - Mark extra clone instances as orphans and don't show inactive ones
Low: Core: Add a cast to keep the compiler happy
Low: Core: Record where G_LOG_LEVEL_CRITICAL messages were logged from
Medium: crmd: Avoid calling glib functions with NULL
Medium: cib: Clean up the CIB if it was already disconnected
Low: Build: Add back the stable series varirable
Low: cib: Make sure standalone mode runs as the correct user
Low: Build: Tweak soem helper targets
Dev: cib: Clean up the passwd struct after use
Medium: Don't use the cluster UID and GID directly. Look them up based on the configured value of HA_CCMUSER
Dev: Apparently the result of getpwnam() doesn't need to be free'd afterall
Low: Core: Tweak the free_xml() macro
Medium: Core: Switch to the libxml2 parser - its significantly faster
Low: Build: Link libcib.so against required libraries
Low: cib: Add a section for dlm configuration
Low: ais: New function for looking up node by their id
Low: Build: Link libcrmcluster.so against the cluster libs
Low: Core: Indicate to clients which openais version we're using
Medium: ais: Avoid double-free in error path
Merge in dejan's latest
Dev: Build: Specify libcrmcommon correctly
Medium: ais: Allow transient clients to receive membership updates
Hg: Merge from stable-0.6
Medium: PE: Allow groups to be involved in optional ordering constraints
Medium: PE: Implement no-quorum-policy=suicide (FATE #303619)
Low: PE: Fix up regression tests
Remove dead code
Fix compilation when GNUTLS isnt found
Low: Build: Fix compilation on s390
Medium: Fix compilation when heartbeat is not supported
Low: ais: Include a separator in node interface details
Low: Core: Include internal header in crm.h
Low: Core: Do not update last-seen for nodes that are inactive
Low: cib: Fix build warning
Low: cts: Use the correct init script
Medium: cts: CIB's can no longer contain comments
Low: xml: Add new attribute to transitional DTD
Low: ais: Set gid before uid when spawning child processes
Low: Core: Simplifiy free_xml_from_parent()
Medium: Core: Fix memory leak in xpath searches
Dev: Core: Fix problems with the switch to the libxml2 parser
Low: cib: Remove dead code
Dev: Remove logging
Medium: Core: Get more details regarding parser errors
Low: cib: Create common defines for xml diff scafolding
Low: cib: Use the new library function instead
Dev: crmd: Don't use quotes in xpath expressions, the libxml2 parser can't handle it
Medium: crmd: Fix te_update_diff() now that get_object_root() functions differently
Medium: crmd: Repair the input validation logic in do_te_invoke
Medium: crmd: Fix the lrmd xpath expressions to not contain quotes
Hg: Merge in fixes that make the cluster functional again after switching to the libxml2 parser
Medium: cts: Enable a bunch of tests that were incorrectly disabled
Dev: ais: Don't call setgid() for child processes, openais expects gid=root for clients
Medium: cts: The libxml2 parser wont allow v1 resources to use integers as parameter names
Medium: crmd: No further processing is needed when using the LRM's API call for failing resources
Medium: crmd: Asynchronous failure actions have no parameters
Low: cts: Update the ResourceRecover test for the LRM's API call
Dev: Core: Add a better entry point for parsing xml in files
Low: cts: Update the BadNews suppresion patterns for ResourceRecover
Medium: Core: Repair expand_plus_plus - don't call char2score on unexpanded values
High: crmd: Make sure async-failures cause the failcount to be incrimented
Low: crmd: Use a fake UUID that at least has the right format
Medium: cib: Only incriment num_updates if something actually changed
Low: ais: Enable per-child enablement of valgrind
Low: ais: Warn if valgrind is requested but not available
Medium: ais: Spawn the PE from the ais plugin instead of the crmd
Medium: pengine: Teach the PE how to stop old copies of itself
Medium: pengine: Clean up resources after stopping old copies of the PE
Low: cts: Re-enable the ComponentFail test
Dev: PE: Implement a smarter way of dealing with parameters starting with a digit
Low: cts: Remove the tengine component
Medium: PE: Determin the correct migration-threshold during resource expansion
Dev: Core: Remove dev logging
Medium: PE: Allow sets of operations to be reused by multiple resources
Low: Core: Simplify expand_idref()
Low: cts: Fix the component fail test
Low: ais: Smarter logging of child termination
Low: cts: More log pattern fixes for ComponentFail
Dev: Core: Sanitize input to expand_idref()
Low: cts: More log pattern fixes for ComponentFail
Medium: cib: Prevent use-after-free in client after abnormal termination of the CIB
High: crmd: Correctly handle a dead PE process
Low: cts: More log pattern fixes for ComponentFail
High: crmd: Always connect to stonithd when the TE starts and ensure we notice if it dies
Medium: crmd: Do not allow an election to promote a node from S_STARTING
Medium: PE: Backport a20017bed8cf - Apply stickiness after location constraints are applied so that they work in asymetrical clusters
Hg: Merge with dejan
Low: cts: Log the component killed as debug
Low: Fiddle with a test tool
Low: Build: Link crmcluster with -lcrmcommon so that all required symbols are present
Medium: Core: Use a libxml2 library function for xml -> text conversion
Hg: Merge with stable-0.6
Low: PE: Update raw testcase output to include xml header
Dev: Core: Strip of the xml header when calculating an xml digest
Hg: Merge in some extra fixes for the switch to xmlDocDumpFormatMemory()
Dev: Core: Resolve memory leaks when calling dump_xml() with a non-root xmlNode
Low: Tweak a test util
Low: cts: Ignore a log message the has since been downgraded to debug in heartbeat
Dev: Core: Add missing prototype
Dev: cib: Fix special handling when valgrind is enabled
Low: crmd: Remove valgrind support from deprecated code
Dev: Core: Fix the implementation of dump_xml()
Hg: Merge in fix for dump_xml()
Low: Tools: haresource2cib.py - Convert complex IP addresses correctly. Patch from Yauheni Labko
Dev: crmd: Fix a recent change to use the correct state comparision in do_election_count_vote()
Low: Build: Update the last release version
Low: Core: Clean out a bunch of dead code
Medium: crmd: Do not vote if we've not completed the local startup
Dev: Core: Reimplement get_xpath_object_relative() to not use the magic that caused so much trouble in dump_xml()
Low: cib: Any missing sections will cause validation failures, just warn here
Low: cib: Tweak define
Medium: crmd: If we get a join offer during an election, better restart the election
Dev: crmd: Do a relative search as there may be other objects containing a diff too
Medium: crmd: Only update have-quorum if the value changed
Low: ais: Be less verbose when logging which membership event we're up to
Backed out changeset 490a84446ae7
Hg: Merge out a previous change
Dev: Core: Produce the same xml text as before to ensure the digests come out the same
Dev: Core: Only enable special handling in dump_xml() when being used for digests
Low: Admin: crm_mon - Exit immediately after SIGINT. Patch from Satomi Taniguchi
Low: Resources: Add some Pacemaker specific resources from Heartbeat
Low: cib: Reduce log message for unfound section to debug. True errors will be picked up in validation
Medium: ais: Include in the mebership nodes for which we haven't determined their hostname
Hg: Merge with stable-0.6
Low: cts: Switch to external/ssh
Hg: Merge from stable-0.6
Dev: Core: Use the correct xpath in get_xpath_object_relative()
Low: Rearrange the lib directory and move the ais plugin to its new home
Dev: ais: Fix undefined symbol
Dev: Build: Fix some broken links created during recent reorg
Low: Build: Move the contents of crm/admin into tools
Low: Extra cleanup after previous commit
Low: Build: Move the remaining parts of the crm directory to more appropriate locations
Clean out some crud in the doc directory
Low: Build: Merge the bootstrap and autogen.sh scripts
Low: Build: Don't automatically run configure from autogen.sh
High: PE: Correctly handle unmanaged resources to prevent them from being started elsewhere
Low: PE: Track the allocation scores for regression tests
Low: PE: Create a better implementation of show scores and use it to track changes to the scoring algorithm
Low: Core: A better name for crm_find_peer()
Low: RA: controld - Load required modules and mount configfs when starting dlm_controld
Low: Tools: Make sure there is a newline after the cibadmin output
Low: Tools: crm_attribute - Fix help text
High: cib: Use single quotes so that the xpath queries for nvpairs will succeed
Low: Logging
High: PE: Parse resource ordering sets correctly
High: PE: Prevent use-of-NULL - order->rsc_rh will not always be non-NULL
Low: Logging
Low: RA: controld - More changes to support ocfs2_controld.pcmk
Medium: Tools: cib_shadow - On create, spawn a new shell with CIB_shadow and PS1 set accordingly
High: Tools: crm_mon - Prevent use-of-NULL for orphaned resources
High: PE: Unpack colocation sets correctly
Low: Tools: crm_mon - Print the call-id of operations and skip any offline nodes
Medium: Tools: Rename cib_shadow to crm_shadow
Medium: cib: By default, new configurations use the latest schema
Low: Tools: crm_shadow - Modify the prompt
Medium: PE: Bug N:417585 - Don't restart group children who's individual score drops below zero
High: PE: Make sure orphaned clone children are created correctly
Medium: ais: Use the agreed service number
Hg: Merge with dejan
Medium: Tools: Backport hb_report updates
Low: RA: Remove support for ocfs2_controld - its started by o2cb
Low: RA: controld - By default, turn off quorum
Low: Build: Determin the correct format for printing uint64_t so that it need not be cast
Low: Build: Make sure needed CRM directories are created
Low: Build: spaces->tabs in top-level makefile
Dev: Build: Fix the creation of crm runtime dirs
Medium: ais: Add support for a synchronous call to retrieve the node's nodeid
Low: Build: Fix debian builds now that bootstrap works differently
Low: extra: Add pingd RA form heartbeat, use new dlm_controld name
Low: cts: Update the search patterns for ais
High: ais: Move the service id definition to a common location and make sure it is always used
Low: ais: Rename the pacemaker plugin
Low: ais: Remove development logging
Low: Build: Fix OBS helpers
Medium: Build: Fix building with builtin ltdl libs
Medium: Build: Fix building when heartbeat support is disabled
Low: Build: Fix installation of pacemaker.lcrso
Hg: Merge with dejan
Medium: Build: Provide more hints to automake about build order in lib/common
Medium: Build: Provide more hints to automake about build order in lib/common
High: PE: MAke stonith-enabled default to true and warn if no STONITH resources are found
Low: PE: Combine a number of PE dataset flags into a bit set
Low: PE: Explicitly turn off stonith for those tests that require it off
High: Core: cl_malloc and friends were removed from Heartbeat
Medium: PE: Bug LF:1950 - Notifications do not contain all documented resource state fields
Low: Core: Inline bitset helper functions
Hg: Merge from stable-0.6
Dev: Core: Fix lmb's last patch
Low: PE: Better logging for unmanaged resources
High: PE: Bug LF:1955 - Umanaged mode prevents master resources from being allocated correctly
High: Tools: crm_verify - clean up the upgrade logic to prevent crash on invalid configurations
High: PE: Monitors for unmanaged resources don't need to wait for start/promote/demote actions to complete
High: PE: Bug LF:1955 - Fix another case of filtering causing unmanaged master failures
High: PE: Bug LF:1955 - Ensure unmanaged masters are unconditionally repromoted to ensure they are monitored correctly.
Hg: Merge with stable-0.6
Hg: Resolve previous merge
Low: PE: Update test scores now that we don't filter location rules for unmanaged resources
Low: PE: Remove dead code from stable-0.6
Medium: PE: Print the correct message when stonith is disabled
Low; PE: Make a function static
High: PE: crm_resource - Fix the --migrate command
Low: PE: Remove dead code from stable-0.6
High: PE: Bug N:420538 - Anit-colocation caused a positive node preference
Low: Tools: crm_mon - remove debug message
Low: Build: Fix compilation with AIS support disabled
Low: Build: Remove dead makefile comments
Medium: Build: Make sure crm_config.h is only ever included once
High: PE: Bug LF:1950 - Ensure the current values for all notification variables are always set (even if empty)
Medium: PE: Implement a true maintenance mode
Hg: Merge with upstream
High: Update version numbers for 1.0 candidates
Low: PE: Simplify the notification creation logic
Low: Core: Use free() for destroying simple hash entries
Dev: PE: Make sure stop notifications are correctly marked manditory when necessary
Low: PE: Update test cases to include blank notifcation values
Low: Core: Clean up crm_get_msec. Original patch from Satomi Taniguchi
Merge with Dejan
Low: Build: Target for emailing patches to the mailing list
Low: Build: Simplify the bomb target
Low: Build: Tweak the bomb variables
Low: Build: Don't send patches if local changes exist
Low: Build: Clean up the tools makefile
Dev: Core: Suppress errors about trailing characters in crm_get_msec. Patch from Satomi TANIGUCHI
Low: RA: Add a catch-all field for passing options to pingd
Medium: PE: Allow external entities to ask for a node to be shot by creating a terminate=true transient node attribute
Dev: Core: Fix compile warning in previous patch from NTT
Hg: Merge with dejan
Medium: xml: pacemaker-0.7 is now an alias for the 1.0 schema
Hg: Merge with upstream
High: Core: Only call xmlCleanupParser() if we parsed anything. Doing so unconditionally seems to cause a segfault
Dev: Build: crm_mon - Fix header location
Update version
Low: cts: Repair cts paths
Low: cts: AIS nodes don't have a UUID
Medium: PE: Add a stonith_timeout option and remove new options that are better set in rsc_defaults
Dev: PE: Fix filter_action_parameters now that crm_get_msec can return negative values on error
Hg: Merge in regression fixes
Low: PE: Only ask for (op|rsc)_defaults once
Low: Build: A new helper target
Low: Build: Cleanup up OBS targets
Hg: Merge with dejan
Low: Core: Reduce logging on XML_ERR_DOCUMENT_END since it is almost always bogus
Low: Core: We don't need a stack trace for evey parsing error
Low: PE: Migrate the testcases to the new syntax
Low: xml: Add new tests
Low: xml: Test that the expected translation didnt change
Low: xml: Only set a crm_feature_set value if one was present in the configuration being upgraded
Medium: PE: ptest - check the input is valid before proceeding
Low: xml: Allow io_dir to be overwritten
Low: PE: Update tests - set default-migration-threshold via rsc_defaults section instead
Low: PE: Remove testing artifacts
Low: Core: Reduce logging on parsing errors - we get detailed logs from libxml2 anyway
Medium: crmd: Prevent potential use-of-NULL in global_timer_callback
Medium: PE: Implement on-fail=standby for NTT. Derived from a patch by Satomi TANIGUCHI
Hg: Merge with dejan
Hg: Merge with dejan
Medium: PE: Revert group stickiness to the 'old way'
Low: Tools: pingd - clarify the Bad Echo log message
Low: PE: Fix spelling mistake
Medium: PE: Detect clients that disconnect before receiving their reply
Low: cts: Only dump the test-run summery before the extracted test
Medium: crmd: Rationalize the logging of graph aborts
Low: Core: Update the minimum schema version to 1.0
Low: PE: Remove dead code
Dev: PE: Provide stonith-timeout in the transition graph
Low: PE: Remove dead code
Low: Core: Make crm_str_eq a non-inline function again as some compilers complain it causes the code to grow :/
Hg: Merge with dejan
Low: crmd: Reduce logging
Medium: crmd: Hook up the stonith-timeout option to stonithd
Low: Core: Make crm_element_value a non-inline function again as some compilers complain it causes the code to grow :/
Dev: crmd: Avoid looking for values in NULL XML objects
Hg: Merge with dejan
Low: Build: Another helper target - this time for weekly reports
High: Tools: Make pingd functional on Linux
Dev: Core: Add the crm_perror() macro used by pingd
Medium: PE: Use the correct attribute for action 'requires' (was prereq)
Low: cts: Fix ComponentFail test for Heartbeat clusters
Medium: crmd: Avoid calling GHashTable functions with NULL
Hg: Merge with Dejan
High: Core: Allow the expected number of quorum votes to be configrable
Low: PE: Fix developmental logging
High: cib: Implement --create for crm_shadow
High: cib: Remove dead files
Medium: cts: Wait for any STONITHd nodes to come back up after a ComponentFail test
Medium: cts: Use no-quorum-policy=ignore for clusters with less than three nodes
Low: cts: Remove two log patterns that are no longer used for ComponentFail tests
Low: crmd: Remove invalid comparision - timer is not an integer
Low: Core: Allow INFO logging to be turned off
Low: crmd: Reduce logging
Medium: cts: Improve the ComponentFail test
Low: crmd: Improved transition logging
Low: crmd: Improved transition logging
Low: crmd: Improved logging for failed action
Low: Build: Tweak the version header creation
Low: crmd: Centralize detection of failed PE processes
Hg: Merge with dejan
Medium: tools: Add a new regression test for CLI tools
Low: tools: Add expected outputs for regression CLI tests
Medium: tools: Expand regression tests
Low: tools: crm_standby - dont complain about missing values
Medium: tools: crm_resource - return with non-zero when a resource migration command is invalid
Medium: tools: crm_shadow - Allow the admin to start with an empty CIB (and no cluster connection)
Low: cib: The cib regression test has been superceeded by the one in tools/
Medium: crmd: Delay raising I_ERROR when the PE exits until we've had a chance to save the current CIB
Low: Build: Refresh the spec file from OBS
Medium: Build: Supply prototypes for libreplace functions when needed
Medium: Build: Reliably detect heartbeat libraries during configure
Medium: Build: Fix makefile whitespace
Hg: Merge with dejan
Medium: Core: Provide better feedback if Pacemaker is started by a stack it doesn't support
Medium: Build: Fix compilation against CoroSync
Low: Build: Conditionally use g_log_set_default_handler which isn't available on RHEL4
Dev: Core: Fix compilation when Heartbeat support is not configured
Hg: Merge with dejan
Low: Build: Conditionally use another instance of g_log_set_default_handler()
Low: Build: Conditionally use another instance of g_log_set_default_handler()
Medium: Tools: crm_mon - Fix the display of timing data
Low: Tools: pingd - Allow the update interval to be modified
Low: Core: Revert unintended logging change
Low: cib: update_attr - Complain on any return value != cib_ok
High: Tools: LF:1969 - Attrd never sets any properties in the cib
Low: Tools: cibadmin - Don't print the usage when the --force flag is required
High: Tools: Make attrd functional on OpenAIS
Dev: Tools: Fix attrd compilation
Low: Build: Send patches to the mailing list
Medium: PE: Allow group and clone level resource attributes
Dev: PE: Add missing header files
Low: PE: Add stonith-timeout to regression tests
High: Core: Don't redefine __FUNCTION__ unnecessarily
Dev: PE: Group parameters take precedence over those of any parent resource
Dev: PE: Primitive parameters take precedence over those of any parent resource
Low: PE: Fix the 1.0 testcases after f9992f741472
Medium: Ensure meta attributes are used consistantly
Low: PE: Remove useless instance_attribute duplicates of meta attributes
Dev: PE: Re-enable code that was accidentally commented out
Medium: Tools: crm_verify - check that we're being asked to validate a complete config
Low: PE: Remove reference to failure stickiness in 1.0 regression test
Low: xml: Regression tests - Remove the random portion of automatically generated IDs so that the diffs are useful
Low: xml: Use the most recent schema in regression tests
Low: xml: Tweak the upgrade xslt to prepare for an upcomming change
Low: xml: Do not bring across clone attributes from 0.6 configurations
PE: Medium: Make a best-effort guess at a migration threshold for people with 0.6 configs
Low: PE: Update the inc9 regression test
Low: PE: Remove reference to failure stickiness in 1.0 regression test
Low: Tools: Update regression test output
Hg: Merge with dejan
Hg: Merge with dev
High: Tools: crm_shadow - Show more line number information after validation failures
Low: Tools: crm_mon - clean up extra colon in summary title
Low: crmd: Remove inline function that was causing compile issues
Medium: Tools: attrd - Automatically refresh all attributes after a CIB replace operation
Low: PE: Improve log message
High: ais: Ensure HA_RSCTMP exists for use by resource agents
High: PE: Fix optional colocation
High: crmd: Bug:1975 - crmd should wait indefinitely for stonith operations to complete
Low: crmd: Fix logging
Dev: crmd: Make sure the transition timeout is restored if stonithd terminates
Medium: crmd: Fix the recording of pending operations in the CIB
Low: PE: Remove code not relevant to the 1.0 series
Medium: PE: Bug N:437719 - Ensure scores from colocated resources count when allocating groups
Dev: crmd: Fix segfault caused by use-of-NULL - transition_graph isn't always non-NULL
Dev: cib: update_attr - rc values greater than 0 occur normally for a-synchronous updates
Dev: crmd: Fix comparision in start_global_timer to supress error message
Low: PE: Disconnected clients are not the PE's problem, downgrade log message
Medium: cib: Ensure remote access ports are cleaned up on shutdown
Low: crmd: Stop the transition timer when 'disconnecting' the TE
Medium: Tools: Bug LF:1987 - crm_verify incorrectly warns of configuration upgrades for the most recent version
High: PE: Prevent assert failures in master_color() by ensuring unmanaged masters are always reallocated to their current location
Medium: Tools: Bug LF:1985 - crm_mon - Correctly process failed cib queries to allow reconnection after cluster restarts
High: PE: Bug LF:1988 - For unmanaged resources we need to skip the usual can_run_resources() checks
Hg: Merge with dejan
High: PE: Bug LF:1988 - Ensure recurring operations always have the correct target-rc set
Low: cts: Add a pingd resource to cts and some constraints which use it
Low: PE: Clean up some confusing log messages
Low: Tools: crm_mon - Wait a moment before exiting so the user can see that the config can't be upgraded
Medium: xml: Relax the restriction on the contents of rsc_locaiton.node
Low: PE: Bug BNC:440242 - Do not restart the entire unordered, uncolcoated group if one child is migrating
Medium: crmd: Detect deleted resource operations automatically
Medium: crmd: Erase a node's resource operations and transient attributes after a successful STONITH
Low: crmd: Minor logging change
Low: cts: Rewrite the Stonithd test to exercise the complete STONITH pathway
Low: ais: Tweak the subsystem start/stop order
High: ais: Allow the crmd to get callbacks whenever a node's state changes
High: PE: Ensure the terminate node attribute is handled correctly
Low: cts: Ignore failed pingd updates during component fail
Low: crmd: dc-deadtime doesn't require two levels of defaults
Dev: crmd: Check for the right state value when erasing node attributes
Low: Core: Log the use of configuration defaults at debug
Low: Core: Create a common function for getting xpath results
Low: crmd: Simplify the processing of CIB update diffs
Low: cts: Don't barf if the remote action result was empty
Medium: cib: Always store cib contents on disk with num_updates=0
Medium: crmd: Initiate an attrd refresh _after_ the status section has been fully repopulated
Medium: crmd: Only the DC should update quorum in an openais cluster
Low: cib: Let the caller decide if DTD/schema validation issues are an error
Low: Tools: attrd - Mask some known/acceptable update errors
Low: crmd: Removed duplicate code - this is now handled before the update is sent
Low: Core: Provide compilers with optimization hints for our logging code
Low: Core: use the hinted logging function that indicates to the compiler that these logs are unlikely to be printed
Low: Core: use the hinted logging function that indicates to the compiler that these logs are unlikely to be printed
Low: cib: Correctly log the origin of replace operations
Low: crmd: Simplify the resource operation updates
Low: Build: Tweak the hg version output
Low: Core: Tweak the logging in get_xpath_object()
High: Core: Add an extra snaity check to getXpathResults() to prevent segfaults
Dev: Core: Fix the comparision in getXpathResult()
Dev: crmd: Fix the xpath expressions for processing cib update diffs
Low: cib: It is up to the cib client to report operation errors
Dev: crmd: Repair the logic that maintains and updates the CIB with our quorum status
Dev: crmd: Repair the logic that maintains and updates the CIB with our quorum status
Medium: crmd: Find a more appropriate place to update quorum and refresh attrd attributes
Medium: ais: Provide better feedback when nodes change nodeid's (in openais.conf)
Low: crm: Add TODO note
Low: Build: Add a new helper target
High: PE: Improve up the detection of 'new' nodes joining the cluster
Dev: crmd: Erase the correct node's status section when it returns
Low: cts: Fix up the stonithd test
Dev: crmd: Detect and ignore lrm_rsc_op deletions caused by cancelation graph events
Low: cib: Only log failed actions in cib_client_register_callback if only_success=TRUE
Low: crmd: Log failed cib queries when trying to invoke the PE
Low: crmd: Fix comment
Dev: crmd: Timers are also allowed in the S_POLICY_ENGINE state
High: ais: Hook up the openais.conf config logging options
Low: cts: Fix the stonithd test
Low: ais: Tweak a log message
Medium: crmd: Fix the handling of unexpected PE exits to ensure the current CIB is stored
Medium: ais: Hook up the options for specifying the expected number of nodes and total quorum votes
Low: crmd: Add A_DC_JOIN_FINAL to fsa_action2string
Low: cib: Simplify the cib_config_changed() logic with xpath
High: cib: Ensure global updates cause a disk write when appropriate
Low: Core: Use the correct str_to_boolean function in crm_is_true()
Dev: ais: Correctly interpret the admin's configuration intentions
Medium: Tools: crm_mon - display the meaning of the operation's rc code instead of the status
Low: crmd: Improve the log message when a node is NACK'd from joining the cluster
Low: PE: Improve startup logging
Medium: PE: Prevent lsb scripts from being used in globally unique clones
Medium: Resources: controld - ensure we are part of a clone with globally_unique=false
Hg: automated merge
Low: Tools: Don't print usage when the --force option is required
High: Core: Repair the ability to have comments in the configuration
Low: Build: Unset GREP_OPTIONS as any coloring can mess up libtool's AC_CONFIG_AUX_DIR matching patterns
Low: crmd: Reduce transition logging
Low: crmd: Improved logging of token identifying a PE computation
High: crmd: Ensure PE processing does not occur for all error cases in do_pe_invoke_callback
High: crmd: Requests to the CIB should cause any prior PE calculations to be ignored
High: ais: Zero out the PID of disconnecting clients
Hg: Automated merge
Low: ais: Better shutdown logging
Dev: ais: Only reset child PIDs if we're shutting down
Hg: Automated merge
Low: Build: Fiddle with reporting helper targets
Dev: crmd: Remove the correct node's status section when a node returns
Low: cts: Fix up heartbeat patterns for ComponentFail test
Hg: Automated merge
Low: cts: Remove pointless key lookup: s/self.CM[up]/up/
Medium: ais: Look for pacemaker options inside the service block with 'name: pacemaker' instead of creating an addtional configuration block
Low: Tools: Add the purpose of common tools to their help text
Low: Tools: Add the purpose of common tools to their help text
High: ais: Create an option for starting the mgmtd daemon automatically
High: heartbeat: Wait for membership 'up' events before removing stale node status data
Low: Build: Update version numbers
Hg: Automated merge
Low: cts: Initial cut at implementing a versioned CIB
Low: cts: Remove legacy crud
Low: PE: Tweak a debug message
Low: PE: Default clone-max to max(1, num_nodes)
Low: cib: Remove developmental logging
High: crmd: Prevent node re-probe loops by ensuring manditory actions always complete
Hg: Automated merge
Low: Tools: Dont complain abotu existing shadow configs when creating a new one
Low: PE: Allow random failing of actions during ptest runs
Low: cts: Create 1.0 based CIBs using the crm shell
Low: cts: Clean up the startup/initialization code and help text
Low: cts: Remove dead checks
Low: cts: Defer CIB creation until its needed
Low: crmd: Increase the usefulness of error message
Dev: cts: Fix the 0.6 config generatior
Use the 0.6 syntax for now
Hg: Automated merge
Dev: cts: Fix spelling mistake
High: cib: Ensure diff's contain the correct values of admin_epoch
Hg: Automated merge
Low: cts: Create a valid heartbeat IP resource
Low: Build: Update the crm_config.h template
Low: PE: Indicate the failed op's return value not status
High: PE: Bug LF:1996 - Correctly process disabled monitor operations
Low: cts: Fix copyright - look at the changelog if you want to argue
High: PE: Correctly calculate the number of available nodes for the clone to use
Medium: crmd: do_dc_join_filter_offer - Check the 'join' message is for the current instance before deciding to NACK peers
High: crmd: Bug LF:1962 - Don't NACK peers because they're not (yet) in our membership. Just ignore them.
High: ais: Repair and streamline service load now that whitetank startup functions correctly
Hg: Automated merge
Low: cts: Add components for attrd and openais
Low: crmd: Remove redunandnt check
Low: crmd: Expand comment
Low: Tools: Unset GREP_OPTIONS during regression tests
Medium: Tools: Allow attrd clients to specify the ID the attribute should be created with
High: Tools: Bug LF:2001 - Transient node attributes should be set via attrd
Medium: Tools: Make pingd resilient to attrd failures
Low: Tools: Improved attrd logging
Medium: Tools: Bug LF:1994 - Clean up crm_verify return codes
Dev: Tools: Fix node calculation and crm_master updates
Low: Tools: crm_attribute - Clean up logging
Dev: Tools: Fix compile error
Dev: Tools: crm_attribute - Fix a braindead comparison in set_via_attrd()
Low: Tools: attrd - Improve logging
Dev: Tools: attrd - Make sure we only delete own entries from the CIB by only setting entry->uuid from local messages
Low: Tools: attrd_updater - Allow a full refresh to be triggered
Medium: Extra: Resources - Use HA_VARRUN instead of HA_RSCTMP for state files as Heartbeat removes HA_RSCTMP at startup
Medium: crmd: Update the crmd options to use dashes instead of underscores
Medium: cts: Add 'cluster reattach' to the suite of automated regression tests
Low: cts: Add comment
Medium: crmd: Implement shutdown using a transient node attribute
Low: crmd: Extra logging
Medium: Tools: attrd - Allow attribute updates to be performed from a hosts' peer
Low: PE: Remove redundant code
Dev: crmd: Revert to centralized shutdowns now that attrd supports non-local option updates
Medium: crmd: Find option values without having to do a config upgrade
Dev: crmd: Fix compile error
Low: cts: Remove dead tests
Low: cts: Remove more dead tests
Low: cts: StonithdTest doesn't need to derive from StonithTest anymore
Low: cts: Remove redundant class method redefinitions
Low: cts: True 0.6 configurations did not have a value for validate-with
Low: cts: Remove dead audit
Medium: cts: Implement an automated SplitBrain test
Hg: Automated merge
Low: cts: Run unisolate_node() in a loop in case multiple iptables rules existed somehow
Low: cts: Make the SplitBrain test a little quieter
Medium: PE: Remove the symmetrical attribute of colocation constraints
Low: cts: Remove reference to deleted variable
Low: cts: Always generate more than one partition for SplitBrainTest
Low: cts: Get the process of the openais component correct
Low: cts: Make the pingd resource non-unique and add missing ordering constraints
Low: cts: Add note for later
Low: cts: Look for promotion and demotion during ReattachTest
Low: Resources: Stateful - Restore master preference if found active during a probe
Low: Resources: pingd - Use the attribute name instead of resource id for the pid filename for non-unique clones
Low: PE: Update 1.0 regression tests for cd100e3f2610
Low: Core: Update the current feature set due to shutdown changes
Low: PE: Update PE tests for new feature set
Low: Build: Build pengine earlier
High: PE: Prevent on-fail settings being ignored after a resource is successfully stopped
Low: PE: Update testcases for cs:9919f48d3313
Hg: Automated merge
Hg: Automated merge
Low: Tools: crm_resource - Include class and unique information for cts
Low: cts: Fix the ordering constraints
Low: cts: Ignore 'multiple active' if the resource is non-unique
Low: cts: Ignore stonith resources that stop/start during ReattachTest
Medium: Resources: pingd - fix the meta defaults
Medium: Resources: Stateful - Add missing meta defaults
Dev: Resources: pingd - Fix spelling mistake s/gloablly_unique/globally_unique/
Low: Resources: Stateful - Use a common state file for non-unique clones
Medium: PE: Defer logging the actions performed on a resource until we've processed ordering constraints
Low: PE: Update regression test outputs for 658e8daa4653
Medium: Tools: crm_shadow - Replace the --locate option with the saner --edit
Low: cib: Allow the user to specify where CIB shadow files live
Low: Build: Exclude pengine regression tests from the tarball
Dev: PE: Fix use-of-NULL
Low: Core: Remove development logging - it can happen when pruning diffs and is normal/allowed
Low: cts: Wait for the single partition to contain all members before continuing the SplitBrain test
Low: cts: Try to heal clusters as quickly as possible
Low: crmd: Streamline the final phase of the DC's join logic
Medium: crmd: Don't fast-track the processing of LRM refreshes when there are pending actions.
Hg: Automated merge
Low: PE: Fix resource action logging
High: PE: Bug 2005 - Fix startup ordering of cloned stonith groups
Hg: Automated merge
High: PE: Bug 2006 - Correctly reprobe cloned groups
High: Tools: Bug 2003 - pingd doesn't correctly detect failures when the interface is down
Low: Tools: crm_mon - Distinguish node standby caused be resource failures (Patch from Satomi TANIGUCHI)
Low: PE: Bug LF:2011 - Fix ptest help text
Low: Core: Ensure crm_peer_init() only takes effect once
High: crmd: Bug LF:2010 - Ensure fencing cib updates create the node_state entry if needed to preent re-fencing during cluster startup
High: ais: Bug BNC#456243 - Ensure the membership cache always contains an entry for the local node
Low: Tools: Sort the list of partition members to avoid confusing CTS
Low: cts: Dont add the automatic patterns for aisexec - it wont respawn itself
Low: cts: Add expected patterns for aisexec component
Low: Use crm_perror which includes the function name in the logs
Medium: HB: Bug 1933 - Fake crmd_client_status_callback() calls because HB doesn't provide them for already running processes
High: Tools: crm_master shouldn't define a set id now that attrd is used. Defining one can break lookups
Low: cts: Further refine the aisexec ComponentFail patterns
Low: Core: Include the errno when logging the result of strerror()
Medium: Tools: Change the pingd defaults to ping hosts once every second (instead of 5 times every 10 seconds)
Low: Tools: pingd - Log to stderr in stand-alone mode
Low: Tools: pingd - Reduce info-level debug for active hosts
Low: attrd: pingd - Set the default dampening interval to 5s
Low: Tools: crm_mon - Tweak the display of failed actions
High: Tools: Bug 2003 - pingd doesn't correctly handle node-down events on multi-NIC systems
Low: Core: Error logging is sufficient in op_status2text()
Low: Tools: pingd - extra help text
Low: Build: Dont fail if we can't build the CLI docs
Low: Build: Disable creating the html version of the CLI docs - asciidoc isn't available in enough places
Low: cts: Increase the threshold for stonith resources
Low: cts: Catch some errors we dont care about
Low: Build: Build doc last
High: Tools: attrd - Correctly determine when an attribute has stopped changing and should be committed to the CIB
High: crmd: Correctly handle reconnections to attrd
Low: Resources: pingd - The value for dampen should be greater than the re-ping interval
High: Tools: Bug 2021 - pingd does not detect sequence wrapping correctly, incorrectly reports nodes offline
Medium: Tools: pingd - fix the command line switches
Dev: Tools: Fix pingd compile errors
Low: PE: New helper function
Low: Build: Support latest CoroSync changes
Hg: Merge with tip
Hg: Automated merge
Hg: Automated merge
Low: cib: CID:2 - Remove dead code
High: Core: CID:10 - Prevent indexing into an array of schemas with a negative value
Medium: PE: CID:7,8 - Prevent hypothetical use-of-NULL in LogActions
Low: cib: CID:1 - Remove dead code
Medium: cib: CID:20 - Fix potential use-after-free in cib_native_signon
High: Core: CID:6 - Fix use-of-NULL in copy_ha_msg_input
High: cib: CID:18 - Fix use-of-NULL in cib_perform_op
High: PE: CID:19 - Fix use-of-NULL in determine_online_status
Medium: PE: CID:17 - Fix memory leak in find_actions_by_task error path
Medium: Core: CID:16 - Fix memory leak in date_to_string error path
High: Core: CID:15 - Fix memory leak in crm_get_peer
Medium: Core: CID:14 - Fix memory leak in calculate_xml_digest error path
High: Core: CID:13 - Fix memory leak in log_data_element
Medium: Core: CID:12 - Fix memory leak in decode_transition_magic error path
High: stonith: CID:3 - Fix another case of exceptionally poor error handling by the original stonith developers
Low: cib: Check the lower bound of internal_update_feature_set results before indexing the arrary
High: cib: CID:18 - Repair errors introduced in b5a18704477b - Fix use-of-NULL in cib_perform_op
Medium: ais: Allow dead peers to be removed from membership calculations
Medium: Tools: LF:2029 - Display an error if crm_resource is used to reset the operation history of non-primitive resources
Low: Tools: Clean up the code for ccm_tool
Low: DevL Fix compile error in ccm_tool
Medium: Tools: Rename ccm_tool to crm_node
Medium: ais: Pass node deletion events on to clients
Low: Tools: crm_node - Add long getopt option processing
Low: Tools: crm_node - Add long options to the help text
Medium: cts: cluster_test - Make some usability enhancements
Medium: Build: Install cluster_test
Dev: cts: cluster_test - Use HOSTNAME instead of HOST
Dev: cts: cluster_test - Test for logs being saved by IP rather than uname
Dev: cts: cluster_test - A better uname to IP lookup
Medium: Tools: crm_mon - Include the DC's version in status summary
Medium: CTS: cluster_test - suggest a valid port number
Medium: CTS: Remove references to deleted classes
Dev: Build: Not all cts scripts are generated
Medium: CTS: Remove references to deleted classes
Medium: CTS: Fix python import order
High: Tools: cluster_test - Allow the user to select a stack and fix CTS invocation
Low: Tools: cibmon - Remove dead code and teach it to maintain a current copy of the CIB
Medium: Tools: cibmin - Detect resource operations with a view to providing email/snmp/cim notification
Low: Tools: cibmon - remove experimental code
Low: Core: Ensure perrors are always seen by the user
Medium: Tools: crm_mon - switch to event-driven updates and add support for sending snmp traps
Medium: Tools: crm_mon - Add support for sending mail notifications of resource events
Medium: Tools: crm_mon - Sanitize startup and option processing
Dev: Tools: crm_mon - Make sure as_console is set correctly
Dev: Tools: crm_mon - Fix reconnection to stopped cluster
High: Core: Fix crash in the membership code preventing node shutdown
Dev: Tools: crm_mon - Defer refreshing the display until all updates have been processed or -i seconds have passed
Dev: Core: Fix a crash when a peer comes up with a new uname
Dev: Tools: crm_mon - Make sure updates are delivered to the application after a reconnect
High: crmd: If there are no nodes to finalize, start an election.
High: crmd: If there are no nodes to welcome, start an election.
Low: cts: Ignore crm_mon errors
Low: cts: Repair SplitBrainTest test
Medium: ais: Supply the node's uname in addtion to the id
High: ais: Detect and handle situations where ais and the crm disagree on the node name
High: crmd: Prevent node attribute loss by detecting attrd disconnections immediately
Low: cts: Where possible, capture and display stderr from remote commands with our logging
Low: cts: Have the Stonithd class use RemoteExec instead of its own custom code
Dev: ais: Fix compilation erorr
Medium: Build: Use more restrictive CFLAGS and fix the resulting errors
Medium: Build: Use more restrictive CFLAGS and fix the resulting errors
High: ais: Bug BNC:456208 - Prevent deadlocks by not logging in the child process before exec()
Dev: ais: Ensure the uname is sent nul-terminated
Low: cts: Fix the calls to popen()
Dev: Tools: crm_mon - Fix compilation when ncurses is disabled
High: Tools: crm_mon Use the OID assigned to the project by IANA for SNMP traps
High: crmd: Bug BNC:467995 - Delay marking STONITH operations complete until STONITH tells us so
Low: cts: Support clusters that don't start Pacemaker at boot time
High: PE: Bug BNC:465484 - Fix the no-quorum-policy=suicide option
Dev: PE: Fix compilation
High: PE: Only shoot online nodes with no-quorum-policy=suicide
Dev: Tools: crm_mon - Fix the status summary when supplied a file
Low: Build: strtoll is not defined by ASNI, fall back to strtol if ANSI_ONLY is defined
High: PE: Repar the ability to colocate based on node attributes other than uname
Low: cts: Consolidate all remote commands to use the RemoteExec class
Low: Tools: cluster_test - Support the new stonith syntax and --at-boot option
Low: ais: Tweak the subsystem start order
Low: Build: Use the HG macro from configure
Low: cts: Fix stonith resource creation
Low: cts: Wait for user input if the cluster does not reform during SplitBrain test
Low: cts: Minor tweaks
Dev: crmd: Raise the correct input when a join fails
Dev: crmd: Prevent use-of-NULL if we can't connect to attrd
High: Tools: Bug BNC:468066 - Don't use the result of uname() when its no longer in scope
High: ais: Add support for the new AIS IPC layer
High: cib: Allow clients without mainloop to connect to the cib
Low: cts: Add flags for enabling tests that don't fully work yet
Low: Tools: crm_verify - remove ignored -D option
Low: Build: Install attrd.h
Low: ais: Optionally return the local node id during signon
Dev: ais: Supply the correct values to openais_msg_send_reply_receive()
Low: crmd: Indicate why cib sign-on failed
Dev: cib: Fix native signon broken in cs: 6e769a5fe034
High: ais: By default, disable supprt for the WIP openais IPC patch
Dev: ais: Only obtain the ipc fd if the connection was successful
Medium: ais: Sanitize ipc usage
Dev: ais: Disable flow control - it prevents IPC communications while membership is in flux
Dev: ais: Enhanced debugging
Dev: ais: Enhanced error handling for openais IPC code
Dev: ais: Fix compile error
Dev: ais: More debugging code for IPC anomolies
Low: cts: Add support for generating a HASI configuration
Dev: ais: Fix developmental log message
High: ais: Ensure crm_peer_seq is updated after a membership update
Low: cts: No need to ping dead nodes every second until they come up
Low: cts: Fix generation of 1.0 configurations
High: PE: Remove an optimization for the terminate node attribute that caused the cluster to block indefinitly
High: ais: Make sure all IPC header fields are set to sane defaults
Dev: ais: Check for the correct value of header.error - SA_AIS_OK
Dev: ais: Don't use the new ipc code by default
Dev: ais: Sort out the AS_AIS_OK != 0 mess for IPC
Dev: ais: Fix compile error
Low: cts: fix SplitBrainTest.is_applicable()
Low: cts: Prompt for continue on a BadNews storm
Low: PE: fix cut/paste error in log message
Medium: Core: Try to track down the cause of XML parsing errors
High: Tools: Bug LF:2036 - crm_resource cannot set/get parameters for cloned resources
Low: Tools: crm_attribute - Better error message when the connection to attrd fails
Low: Tools: crm_attribute - Minor change to option processing
Dev: Tools: Don't do a resource lookup for -c
High: ais: Always set header.error to the correct default: SA_AIS_OK
High: cib: Fix four moderately sized memory leaks detected by Valgrind
Low: Core: Indicate which dtd/rng file we couldn't find/parse
High: Core: Fix more memory leaks foudn by valgrind
High: Tools: Bug LF:2046 - Node attribute updates are lost because attrd can take too long to start
Low: Core: Reduce debug logging
Low: cts: Prevent false negatives in ResourceRecover test
Low: crmd: Connect to attrd and flush our attributes when we get to S_NOT_DC
Low: uild: Detect which IPC code OpenAIS is using during configure
High: Tools: Bug BNC:473265 - crm_resource -L dumps core
Dev: ais: Fix compilation errors
Low: cts: Modify a search pattern
Dev: ais: fix compilation error
Dev: ais: Fix compilation
Low: crmd: There is no need to start a timer after unpacking the graph or to default to a non-zero value
Low: crmd: It is up to the application to manage transition timers
Low: crmd: Minor code cleanup
Medium: crmd: Bug BNC:472473 - Reduce the transition timeout to action_timeout+network_delay
Low: ais: Reduce plugin logging
Medium: crmd: Bug BNC:472473 - Don't wait excessive amounts of time for lost actions
Low: cts: Fix the ResourceRecover test
Low: Remove pointless abstraction function
Low: ais: Add some more messaging asserts
Dev: PE: Never wait for shutdowns to complete when executing graphs
Dev: crmd: Remove developmental logging
Low: ais: Don't validate non-AIS_Message messages
Low: crmd: Dont log when there is nothing to print
Low: crmd: Combine logs
Low: PE: Remove dead code
Dev: crmd: Fix use-if-NULL in fail_incompletable_stonith()
High: Core: Prevent unterminated strings after decompression
Medium: Core: Bug BNC:474727 - Set a maximum time to wait for IPC messages
Low: cts: LogAudit - Check the hosts are logging with the expected hostname
Low: cts: NearQuorumPointTest - Wait for any nodes left up to notice the nodes we're stopping leave
Low: Build: Cleanup bootstrapping code
Low: cts: Look for unexpected stonith ops
Low: Build: Cleanup
High: PE: Prevent use-of-NULL for failed migrate actions in process_rsc_state()
High: crmd: Ensure updates for lost migrate operations indicate which node it tried to migrating to
Low: Build: Clean up version macros and output
Low: cts: Ensure all nodes are up before testing logging
Low: cts: Ensure all nodes are up before testing disk space
High: Tools: Cause the correct clone instance to be failed with crm_resource -F
High: PE: Start the correct monitor operation for unmanaged masters
Low: PE: Fix the action logging for unmanaged resources
High: PE: Clones now default to globally-unique=false
Low: cts: Make logging restarts quieter
Low: Tools: Update crm_attribute help text
Low: cib: Low message doesn't need to be sent to stderr
Low: ais: expected_nodes should defaut to 1
Dev: PE: Update test cases for and resolve issues created by cs: 656ea620eca7
Dev: PE: More test updates for cs: 656ea620eca7
Dev: PE: Update the expected outputs for cs: 656ea620eca7
Low: Build: Replace the deb target with something sane
Low: cts: Teach ResourceRecover how to deal with anonymous clones
Low: cts: Rename the CM_LinuxHAv2 class/file
Low: Build: Minor tweaks to OBS targets
Dev: cts: Fix previous LinuxHAv2 rename
Low: Tools: crm_mon - Minor output enhancements
Low: Tools: crm_mon - Indicate node unclean status in output
Low: PE: Bug LF:2064 - Display active slaves as 'slave' not 'started'
Low: crmd: Force PE inputs to use our local UUID for dc-uuid
Low: cib: Message doesn't need to be sent to stderr
Low: Build: Only calculate the build version during configure
Low: cts: Add Lars' standy loop test
Low: cts: Name the standby test properly
Low: Build: Refresh supplied spec file
Low: Build: Refresh supplied spec file
Low: cts: Fix the standby loop test
Low: Core: Define an enviroment variable naming the current process
Medium: crmd: Automatically detect and use new values for crm_config options
Low: crmd: Change the default recheck interval to 15 minutes (was disabled)
Low: cts: Enhance the valgrind tests
Low: cib: Remove developmental debugging
Low: cts: Remove tests made redundant by the new valgrind test(s)
High: crmd: Fix memory leak in cib notication processing code
Low: Tools: attrd - fix memory leak in attribute cleanup during shutdown
Low: cts: More enhancements to the valgrind tests
High: crmd: Fix memory leak in transition graph processing
High: ais: Centralize all quorum calculations in the ais plugin and allow expected votes to be configured int he cib
Low: cts: Add missing file
Medium: crmd: Add new expected-quorum-votes option
Dev: Core: Aid xml debugging
Low: crmd: Cleanup and extra validation of messages to/from the cluster
Low: cib: Reduce logging
Dev: ais: Sanitize header.id of cluster messages relayed over ipc, simplify the new quorum set/get messages
Dev: crmd: Make ais-specific code depend on SUPPORT_AIS
Dev: crmd: Membership updates don't contain quorum details
Dev: ais: Supply quorum details with membership updates
Low: Build: Fiddle with helper targets
Low: Build: Fiddle with helper targets
Dev: ais: Also update quorum values during membership updates
Medium: ais: Remove the ugly hack for dampening AIS membership changes
Medium: PE: Ensure clones always have a value for globally-unique
Dev: PE: Ensure clone _children_ always have a value for globally-unique
High: PE: Bug LF:2075 - Fix large pingd memory leaks
Low: ais: Remove unneeded log message
High: PE: Bug BNC#478687 - Don't crash when clone-max is 0
High: Resources: pingd - Correctly tell the ping daemon to shut down
High: PE: Look up the correct promotion preference for anonymous masters
Low: crmd: Rationalize logging
Medium: Tools: pingd - Avoid using various clplumbing functions as they seem to leak
Low: Core: Use the regular Glib functions for timers, there's no need for the clplumbing wrappers
Low: Core: Don't get mainloop involved in signal handling
Low: Core: Replace the manual trigger from clplumbing with a simplified version
Low: ais: Rename functions for clarity
Dev: Build: Add new header
Low: cts: Adjust ais shutdown patterns
Dev: Build: Add missing C file
Low: cts: Populate the expected quorum votes
High: crmd: Bug BSC#479543 - Correctly find the migration source for timed out migrate_from actions
Dev: cts: Fix expected-quorum-votes setting
High: crmd: Avoid a race condition by waiting for the attrd update to trigger a transition automatically
Medium: crmd: Allow up to 5 retires when an attrd update fails
Low: crmd: Improve logging
High: crmd: Fix memory leaks found by valgrind
Dev: core: Remove redundant call to realloc
High: crmd: More memory leaks fixes found by valgrind
High: ais: Fix memory leak in error path
High: ais: Another memory fix leak in error path
Medium: crmd: Fix memory leak in shutdown
Low: cts: Allow valgrind to be turned on any time
Low: ais: Rename some functions
High: All signal handlers used with CL_SIGNAL() need to be as minimal as possible
High: cib: crmd: Revert part of 9782ab035003. Complex shutdown routines need G_main_add_SignalHandler to avoid race coditions
Hg: Automated merge
Low: ais: Increase log level for invalid cluster messages
Low: ais: We should never get an pcmk_ipc() or pcmk_cluster_callback() callback for a message smaller than AIS_Message
High: PE: Ensure orphaned clone members are hooked up to the parent when clone-max=0
Low: PE: Remove dead code
Low: crmd: Extra error-path logging
High: PE: Bug bnc#466788 - Make colocation based on node attributes work
High: PE: Bug bnc#466788 - Exclude nodes that can't run resources
Low: PE: As a consequence of cs: 06ad44a8c468, resource score for standby nodes are -INFINITY
Medium: crmd: Prevent the crmd from being respawned if its told to shut down when it didn't ask to be
High: PE: Fix memory leak in LogActions
Low: cib: Bump the ipc timeout to 120s
Low: ais: Record the process state for invalid messages
Low: Build: Some users report pingd needs glib.h to compile
Medium: crmd: Send resource updates via the local CIB during elections
Medium: crmd: Send resource updates via the local CIB during elections
High: ais: IPC messages are only in scope until a response is sent
Low: crmd: Downgrade a log message to warning
Low: PE: Indicate unique/non-unique clone status in output
Low: crmd: Also send resource updates to the local CIB when in S_PENDING
High: PE: Correctly attach orphaned clone children to their parent
Low: PE: Add new regression test
Medium: PE: Determin a promotion score for complex resources
Dev: Core: Overhaul signal handling
Low: Core: Remove the other references to CL_SIGNAL
Dev: Core: Used the supplied hander in crm_signal()
Dev: Build: Add missing header
Low: Build: Use g_timeout_add() which is more portable. Patch from Junko IKEDA
Low: Build: Remove crud
Low: Build: Remove more crud
Low: Build: Remove more crud
High: ais: Correctly handle a return value of zero from openais_dispatch_recv()
High: PE: Simplify handling of start failures by changing the default migration-threshold to INFINITY
Low: PE: Don't display allocation scores for orphans
High: PE: Fix the determination of whether a group is active
High: PE: Bug LF#2087 - Correctly parse the state of anonymous clones that have multiple instances on a given node
Low: PE: Update regression tests
Dev: PE: Fix regressions in cs: 71a5506f058b
Low: PE: Update regression tests
Low: PE: Tweak display for orphaned resources
Low: PE: Add regression test for LF#2084
Low: Tools: Remove reference to deprecated/pointless PE API call
Medium: PE: Prevent orphan clones from being allocated
PE: Low: Update regression tests
Medium: PE: Clean up the API - removed ->children() and renamed ->find_child() to fine_rsc()
Low: cts: Make the valgrind error pattern stricter
Low: Code cleanup
Low: Code cleanup
Low: More cleanups
Dev: Build: Fix installtion of xml schemas
Low: crmd: More logging
Dev: Build: Fix building of versioned schemas
Low: Build: Header file cleanups
Low: PE: cib: Sanity check the directories we need at startup
Medium: Tools: Bug bnc#482558 - Fix logging test in cluster_test
Medium: Tools: Bug bnc#482828 - Fix quoting in cluster_test logging setup
Medium: Tools: Bug bnc#482840 - Include directory path to CTSlab.py
High: crmd: Call crm_peer_init() before we start the FSA - prevents a race condition when used with Heartbeat
High: Tools: pingd - prevent double free of getaddrinfo() output in error path
High: Tools: Bug bnc#483365 - Ensure the command from cluster_test includes a value for --log-facility
High: ais: Disable logging to a file
High: cib: Bug bnc#482885 - Simplify CIB disk-writes to prevent data loss. Required a change to the backup filename format
Medium: Tools: Reduce pingd logging
Dev: Build: Fix compilation on x64 - We dont need longclock_t intervals/precision for debug timers
Medium: Tools: Make attrd quieter
Low: Build: Misc cleanups
Low: cts: Simplify CIB cleanup
Low: cts: Bug bnc#483355 - Replace the deprecated use of popen2 with subsystem
High: Tools: The default re-ping interval for pingd should be 1s not 1ms
Low: Tools: Simplify crm_uuid now that we no longer ship it for ais-only clusters
Low: ais: We dont need millisecond resolution for resending messages
Low: Build: Cast the result of getpid() when printing
Medium: crmd: Ingore join messages from hosts other than our DC
Low: cts: Add the ability to run every test once and exit
Low: cts: Fix ResourceRecover - Anonymous clones are often restarted with a different clone number
Dev: crmd: Fix compilation error
Low: ais: Clean up the code for spawning children and remove out-of-date comment
High: ais: Bug bnc#482847, bnc#482905 - Force a clean exit of OpenAIS once Pacemaker has finished unloading
High: PE: The ordered option for clones no longer causes extra start/stop operations
Low: PE: Update regression tests for cs: 07c74f08cafe
Low: crmd: Downgrade an error message - by itself its not a problem
Medium: Core: Add library functions for instructing the cluster to terminate nodes
Dev: Core: Allow clients to force a new connection if they know the old one failed
Low: Build: Remove dead file
Dev: Build: Fix autoconf directory substitutions
Dev: Build: Remove dead file reference in makefile
High: Tools: crm_mon - Prevent use-of-NULL when summarizing an orphan
Low: Tools: crm_resource - include full class/type/provider details in -c output
Low: cts: Implement two tests for the HA extension
Low: cts: Tweak the new HAE tests
Dev: Tools: Provide the correct path to the pengine
Low: Build: Use standard autofoo constructs
Low: cts: Fix syntax typos
Dev: cts: Fix bad news handling
Dev: cts: Increment the test counter
Low: cts: Update valgrind suppressions file
Low: cts: Define the length of loops centrally
Low: Build: Update the example spec file
Low: cts: Define the length of loops centrally
Medium: cib: Fix memory leaks by using mainloop_add_signal
Medium: crmd: Fix memory leaks spotted by Valgrind
Low: cts: Make things easier to maintain by avoiding generation of CTS files
Low: cts: Fix condition in PrimitiveAudit
Dev: cts: Add missing file
Low: cts: Fix ResourceRecover test for anonymous clones
Low: cib: Remove developmental logging
Low: attrd: Log updates that fail with cib_diff_failed as warnings
High: PE: Bug lf#2091 - Correctly restart modified resources that were found active by a probe
High: PE: Bug lf#2089 - Meta attributes are not inherited by clone children
Low: ais: Fix compilation with corosync
Low: Tools: Add help text
Low: ais: dlm_controld needs the peer id cache
High: ais: Bug bnc#486858 - Fix update_member() to prevent spamming clients with membership events containing no changes
Low: cts: Detect when lrmadmin can't be found
Low: Tools: crm_mon - Don't log the node id if its the same as the uname
Medium: PE: Compress the display of healthy anonymous clones
Low: cts: A bunch of cleanups so that we dont peform inappropriate test when ocfs2 is active
Low: Point the factory target to the internal build service
High: PE: Correctly handle terminate node attributes that are set to the output from time()
Dev: PE: Print the expanded details of failed or unmanaged clone instances
Dev: PE: Fix compilation on linux
Dev: PE: Indicate stopped clones also
Low: PE: Minor code cleanup
Low: PE: Add comment
Low: cts: Fix Reattach test
Low: cts: lrmadmin is now in the path
Low: cts: Fix CM.active_resources used by the Recover and Standby tests
Low: cts: Handle skipped tests
Low: PE: Fix compile error
Low: cts: Fix CM.active_resources properly
Medium: Tools: crm_mon - Compress the node status for healthy and offline nodes
Dev: Tools: Fix compilation of crm_mon
High: PE: Bug bnc#488721 - Fix id-ref expansion for clones, the doc-root for clone children is not the cib root
High: crmd: Bug bnc#489063 - Ensure the DC is always unset after we 'loose' an election
Medium: PE: Bug lf#2101 - For location constraints, role=Slave is equivalent to role=Started
Low: PE: Remove extra newlines from clone summary
Low: PE: Don't log the handling of stopped orphans
Low: PE: Remove noisy and unhelpful logging
High: PE: Bug lf#2094 - Fix probe ordering for cloned groups
Low: PE: Make the same change for clones as in cs: 3cf340443465. Lower priority since clones are currently at the top of the resource heirarchy
Low: PE: Downgrade developmental logging
Low: crmd: Clean up election logging
Low: crmd: Reduce election logging further
Medium: crmd: Clean up and optimize the DC election algorithm
Dev: crmd: Make sure superceeded election ops are ignored
Low: crmd: Remove unneeded indirection
Low: crmd: Improved election logging
Dev: crmd: Reduce election logging now that it seems reliable
High: PE: Bug bnc#490418 - Correctly determine node state for nodes wishing to be terminated
Medium: cib: Overhaul the processing of synchronous replies
Medium: crmd: Re-check the election status after membership events
High: RA: Bug bnc#490641 - Shut down dlm_controld with -TERM instead of -KILL
Medium: Tools: crm_shadow - Support -e, the short form of --create-empty
Medium: Tools: crm_shadow - Return 0/cib_ok when --create-empty succeeds
Low: Move some noisey log messages from info to debug
Low: cts: Only list active primitice resources - otherwise ResourceRecover gets confused
High: crmd: Bug bnc#480977 - Prevent extra, partial, shutdown when a node restarts too quickly
Low: cib: Enhanced logging - include the current cib version number
Dev: crmd: Set an value for terminate/shutdown on startup instead of setting to NULL
High: crmd: Erasing the status section shouldn't be forced to the local node
Dev: cib: Repair operation logging
Low: ais: Fix compile warning
Low: crmd: Log the results of xpath deletions
High: crmd: Bug bnc#480977 - Prevent extra, partial, shutdown when a node restarts too quickly (verified)
Dev: crmd: fix typo
Dev: cib: Provide a default version string when the cib isn't yet populated
Medium: crmd: Limit the scope of resource updates to the status section
Medium: cib: Move more logging to the debug level (was info)
Low: Tools: crm_resource - spelling error in help text
Low: crmd: Log version details only if we're starting the daemon
Medium: PE: Bug bnc#491441 - crm_mon does not display operations returning 'uninstalled' correctly
Medium: crmd: Bug bnc#490426 - Escalated shutdowns stall when there are pending resource operations
Medium: Tools: Reduce pingd logging
Low: cts: Fix the '--stack lha' option
Low: cts: Optionally allow multiple regex's to match a single line in lookforall()
Low: cts: Heartbeat stack - Fix ComponentFail and disable the Reattach test
Hg: Automated merge
Low: Tools: Tweak the operation of crm_node when connecting to a Heartbeat cluster
Low: cts: Fix expected bad news when the ccm exits (heartbeat clusters only)
Medium: PE: Correctly log the actions for resources that are being recovered
Hg: Automated merge
High: Core: Generate the help text directly from a tool options struct
Dev: Core: Send help text to stderr/out depending on exit code
Low: Tools: Convert pingd to the new help-text system
Low: Tools: Some minor help text formatting
Low: Tools: Convert crm_verify to the new help-text system
Low: Tools: Convert crm_shadow and crm_diff to the new help-text system
Low: Tools: Convert ptest and crm_node to the new help system
Low: Tools: Convert crm_resource to the new help system
Low: Tools: Convert cibadmin to the new help system
Low: Tools: Convert crmadmin to the new help system
Low: Core: Allow non-functional elements (ie. just for display with --help) in the options arrary
Low: Tools: Tweak the cibadmin help output
Low: Build: Start generating man pages from help text if help2man is available
Low: Tools: More tweaks to the cibadmin help text
Low: cib: Add common function for get host uname/uuid
Low: Core: Add (and use) library functions for updating node attributes with attrd
Low: Tools - Convert attrd_updater to use the common attrd update routines
Low: Tools: Reindent crm_node
Low: Tools - Convert pingd to use the common attrd update routines
Low: Tools: Standardize on -$ as the short form of --version since -v is often taken
Low: Tools: Standardize on -$ as the short form of --version since -v is often taken
Low: Tools: Standardize on -$ as the short form of --version since -v is often taken
Low: Tools: Convert iso8601 to the new help format
Low: crmd: Convert pingd to use the common attrd update routines
Low: Tools: Convert crm_failcount, crm_master and crm_standby into shell-script wrappers for crm_attribute
Low: ais: Remove unused file
Low: Tools: crm_attribute - Allow a default value to be specified when attributes are not found
Low: cib: Remove dead code
Dev: Tools: Fix the crm_attribute wrappers and update the regression tests
Low: Tools: Massage help text contents and output
Low: Core: Remap common aliases for the nodes and status sections when updating attrd
Low: Tools: crm_node - Allow the node's cluster id to be queried
Medium: Tools: haresource2cib.py - Patch from horms to fix conversion error
High: Tools: crm_resource - Repair the ability to query meta attributes
Low: ais: Implement something for pcmk_exec_dump
Dev: Build: Fix the use of SUPPORT_HEARTBEAT in crm_node
Dev: Build: Fix the use of SUPPORT_HEARTBEAT in crm_node
High: ais: Fix compilation for the latest Corosync API (v1719)
Dev: ais: More macros so that we can compile against both ais versions
Dev: Tools: crm_master - Ensure options are processed
Low: Build: Define our own field names
High: lha: Restore compatability with older versions
Low: Core: Indicate which stack we're connecting to
Low: Build: Modify some of the OBS helpers
Low: cts: Fix creation of stonith resource for the 0.6 syntax
Low: cts: Simplify CIB installation
Low: cts: Remove dead code
Low: cts: CTSproxy is no longer required
High: attrd: Don't perform all updates as complete refreshes
Dev: Core: Aid debugging of compatibility issues
Dev: lha: More changes for compatibility with older versions
Dev: Core: Aid debugging of compatibility issues
Low: Tools: crm_mon - Don't publish options that are not supported by this build
High: PE: Make running a cluster with STONITH enabled but no STONITH resources an error and provide details on resolutions
Low: Core: Add some format templates for help text examples
Medium: Tools: Add examples to help text outputs
High: Core: Handle differences in 0.6 messaging format
Low: Build: Update helpers
Low: cts: Better logging of failed commands
Low: cts: Add a rolling upgrade test
Medium: Tools: Add examples to help text outputs
Medium: Tools: Add examples to help text outputs
Dev: Tools: cibadmin - mark --delete-all as a dangerous option
Low: PE: Indicate processing and config errors to clients
Dev: Core: Option flags can be or'd together, so dont use a switch
Dev: ais: Update for corosync 2162
Dev: ais: isolate corosync specific headers
Hg: Automated merge with stable
Dev: ais: Fix service constructor
Dev: ais: Fix whitetank support
Low: ais: Sanitize the code masking the differences between whitetank and corosync
Low: ais: Simplify the ais_common header
Low: ais: Merge ais_common.h into (the practically empty) ais.h
Dev: Build: Clean up library usage
High: crmd: Correctly calculate how long an FSA action took to avoid spamming the logs with errors
Low: crmd: Don't track how long functions take to execute
Low: PE: Update note-to-self
Medium: Tools: Two new options (--stack,--constraints) to crm_resource for querying how a resource is configured
Low: Core: Use the string INFINITY without a leading '+'
High: PE: Rewrite the notification code to be understanable and extendable
Low: PE: Create a stonith-complete pseudo action that actions can depend on
High: PE: Provide inter-notification ordering guarantees
Low: PE: Remove redundant inter-resource notification consttraints
Low: PE: Have post-notification of stops be an input to the all_stopped pseudo event
Low: PE: Ensure the relative ordering of pre and post notifications
Low: PE: Remove dead code
Low: PE: Remove the notify_keys hack
High: PE: Bug bnc#495687 - Filesystem is not notified of successful STONITH under some conditions
High: Tools: crm_mon - Fix smtp notifications
Dev: PE: Remove dead code
Low: Tools: crm_mon - Fix log message in print_html_status
Low: Tools: ptest - Add the -a to the list of documented options
Medium: PE: Correctly log the occurance of promotion events
Low: cts: Correctly look for fields in RollingUpgradeTest
Low: Extra: showscores - Don't complain when cleaning up
Low: PE: Fix the english in a log meesage
Low: PE: Don't complain about probes that find active resources
Medium: crmd: Record in the CIB, the cluster infrastructure being used
Medium: PE: Build a statically linked version of ptest
Medium: Extra: Refresh showscores script from Dominik
Medium: PE: Implememt node health based on Mark Hamzy's original patch
Medium: Tools: crm_mon - Add option to run custom notification program on resource operations (Patch by Dominik Klein
Low: Tools: crm_mon - Fix the notification logic
Dev: PE: Add a semi-custom node health strategy (same as the others but with custom scores)
Dev: PE: More help text for node health
Low: PE: Comments and spacing
Dev: Core: Fix the value of node_score_infinity
Medium: cib: Don't bother validating updates that only affect the status section
Dev: Build: Get the ptest.static target working
Dev: Build: Get the ptest.static target working
High: cib: Fix huge memory leak affecting heartbeat-based clusters
Dev: Build: Give up on the dream of a static ptest
Low: cts: Simplify the installation of cluster configurations
Low: PE: Remove developmental logging
Medium: cts: Don't combine crm_standby arguments - the wrapper can't process them
Dev: Tools: Add better option parsing to wrapper scripts by using getopt(1)
Dev: Tools: Support the old -U|--uname form for supplying node names
Low: cts: Repair the CIB audit
Low: cts: Remove dead code
Low: cts: Further simplify CIB creation logic
Dev: Tools: Clean up the crm_standby wrapper
Low: Build: Clean up spec file
Low: Build: Clean up spec file
High: crmd: Fix another large memory leak affecting Heartbeat based clusters
Low: cts: Formatting
Low: Tools: crm_diff - Add the -f option to the help text
Medium: cts: Fix the CIBAusdit class
Dev: Tools: Fix the option processing for crm_master and crm_failcount
Low: cts: Add all the resource's tags and be done with it
Dev: cts: Fix the PrimitiveAudit and remove dev logging
Dev: cts: Message improvement
High: crmd: Bug lf#2120 - All transient node attribute updates need to go via attrd
Low: cts: Improve the PrimitiveAudit
Hg: Automated merge
Dev: Core: Correctly initialize the cache in get_uname()
Low: cts: Align the test summary
Low: cts: Tag audit related CTS logs
Low: cts: This particular message isn't triggered anymore
Medium: Core: Include supported stacks in version information
Low: cts: 'calls' is more interesting
Dev: Core: Fix compilation
Dev: Core: Fix get_uname() for Heartbeat-based clusters
Low: Core: Standardize uname lengths
High: Tools: attrd - Prevent race condition resulting in the cluster forgetting node's wish to shut down
Dev: Tools: attrd - Don't print NULL
Dev: Tools: attrd - The docs lie, 0 is a valid source id.
Dev: Tools: More fun with attrd
High: PE: Prevent use-ofNULL when using resource ordering sets
Dev: Tools: attrd - No need to disconnect from the CIB if we never connected
Medium: Tools: Include stack information in crm_mon output
Low: Core: Simplify failure logic
High: Core: Allow xpath queries to match attributes
Medium: Tools: crm_resource - Allow --cleanup to function on complex resources and cluster-wide
High: ais: Bug lf#2126 - Messages replies cannot be routed to transient clients
Low: cts: Include the random seed
Low: Build: Bump version numbers
Low: Build: Use standard name for configure input
Low: Build: Update debian changelog
Low: Build: Fix compile warning
High: attrd: Support the value++ and value+=... syntax required for failcounts
Dev: Tools: attrd - Fix value++ expansion
Hg: Fix tags
High: Tools: attrd - Prevent use-of-NULL
Low: Build: Older versions of automake didnt have abs_builddir
Low: Build: net-snmp-config only allows one option at a time
Low: Build: Compile against cluster-glue
Low: Build: Streamline configure
Dev: Build: Fix compilation
Dev: Core: Fiddle with header files
Dev: Core: Fiddle with header files
Dev: Build: Fix compilation of crm_uuid
Dev: Build: Fix directory installation
Low: Build: Remove unused make targets
Medium: cib: Supply an empty status section for replace operations
Low: Build: Remove duplicate definition of UUID_FILE
Low: Core: Reduce logging noise
Low: Tools: Ensure crm_resource -q prints a trailing newline
Low: ais: Better handling of to_file: and to_syslog: options
Low: extra: Add a dummy parameter to 'Dummy' so we can test reload
High: lrm: Look in the correct location for stonith agents
High: PE: Fix reload for master/slave resources
Dev: PE: Turn down the debug logging
High: Core: Show help text and exit with rc 1 if option processing failed
Low: Tools: ptest no longer supports --xml-pipe
High: Tools: Differentiate between --help and an unknown option
Low: Build: Fix warnings in autogen.sh
High: ais: fix compilation against latest corosync
Low: Build: Add shebang to shell script
High: Build: Fix compilation when snmp and esmtp are not available
High: crmd: Terminate if we are ever evicted from the membership
High: crmd: Unset any existing DC value before querying for a new one
High: ais: Fix cluster connection when using corosync 1.0
High: PE: Bug bnc#515172 - Fix the boolean-op attribute of rules
High: PE: Bug bnc#515172 - Correctly process location constraint rules which contain multiple expressions
Low: PE: Update regression tests for the bnc#515172 fixes
Low: PE: New test for bnc#515172
Low: Build: Use a service define name that doesn't clash with corosync
Low: Build: RPM helpers
Medium: Extra: Add tools, an RA and tests for the System Health feature written by Mark Hamzy
Low: cts: All standard ditributions store the default python in /usr/bin. Use that
Medium: Tools: crm_resource - Prevent use-of-NULL by requiring a resource name for the -A and -a options
Medium: PE: Prevent use-of-NULL in find_first_action()
High: PE: Only prevent migration if the clone dependancy is stopping/starting on the target node
Low: PE: New migration regression test
Low: crmd: Include cib version of the updates which caused a transition abort
High: PE: Bug 2160 - Dont shuffle clones due to colocation
Low: PE: Bug 2160 - New regression test
Low: crmd: Extra line-breaks for viewing clarity
Low: PE: Fix find_first_action() when no node filter is specified
Low: PE: ptest - Suppress deleted ordering constraints from the .dot graph
High: PE: New implementation of the resource migration (not stop/start) logic
Dev: PE: Ensure regression tests pass again
Dev: PE: Update the system health regression tests
Low: Build: Include licensing files
Low: Build: Add AUTHORS file for packaging
Dev: PE: Downgrade developmental logging
Low: Build: Prefer glue_config.h over hb_config.h - allows building against heartbeat and openais
Dev: PE: Remove testing artifacts
Low: Build: Prefer unversioned auto* binaries
Dev: Build: Prefer glibtool over libtool for non-Linux systems
Low: Build: Clean up the system health logic in configure
Dev: Tools: crm - Make sure err_buf is defined
Medium: crmd: Note that dc-deadtime can be used to mask the brokeness of some switches
Dev: crmd: Correctly locate the xml diff
High: Tools: crm_resource - Advertise --move instead of --migrate
Medium: Extra: New node connectivity RA that uses system ping and attrd_updater
Low: RA: Create an appropriate state file depending on globally_unique
Low; Build: Bump the version numbers
Low: Build: Remove dead reference
Low: fencing: Beginnings of a new stonith daemon
Low: fencing: Add the IPC callback glue
Low: fencing: Add a basic client library with support for callbacks and notificiations
Low: fencing: Implement the add/remove/call device API calls
Low: fencing: Add test code for the new fencing API
Low: fencing: Allow the device's port name to be specified when invoking the device
Low: fencing: Implement port discovery and port mapping
Low: fencing: Hook up call_optons and implement the unfence API call
Low: fencing: Hook up command replies, implement unfence
Low: Use synchronous calls and log the result
Low: fencing: Implement a query command
Low: fencing: Hook up notifications
Low: fencing: Provide peer disconnect notification
Low: fencing: Some more defines to ensure consistency
Low: fencing: Create a new message class for stonith-ng
Low: fencing: Implement delegated fencing
Medium: fencing: Allow stonith-ng to be built
Low: fencing: Implement a passive mode for the test client
Low: Fencing: Wrapper for driving heartbeat stonith agents according to the Red Hat API
Low: fencing: Add the ability to query the list of registered devices
Low: fencing: Add an admin CLI for fencing
Low: fencing: Allow query ops with no target - return the complete list of devices
Low: Agents: Add an agent for driving fencing
Low: fencing: Clean up the notification and callback APIs
Low: fencing: Finish cleaning up the notification and callback APIs
High: Replace stonithd with the new fencing subsystem
Medium: fencing: Implement a metadata API call and rework the lrm plugin to talk to the new stonith daemon
Low: fencing: Clean up some error codes
Low: fencing: Remove dead code
Low: fencing: Use the lrm plugin afterall
Low: fencing: Use the lrm plugin afterall
Low: fencing: Implement proper host name matching
Low: fencing: Additional development notes
Low: fencing: Drop the old stonith implementation
Low: fencing: Simplify things by using the old stonithd binary location and library (with a higher version)
Low: fencing: fence_legacy - monitor is better implemented as -S with no port description
Dev: fencing: Fix parse_host_list()
Dev: crmd: fencing api wants the timeout in seconds
Low: crmd: Ensure all callbacks from the fencing API log something
Low: fencing: Catch requests for metadata
Dev: fencing: Send the correct message type
Dev: fencing: The standard indicates the host name should be specified with 'port'
Low: fencing: Include some of the fencing op output
Dev: fencing: Test for exec and operation failure when fencing hosts
Low: fencing: Clean up the fencing API
Low: fencing: fence_legacy - Suppress the parameters passed to stonith
Low: fencing: Add additional logging
Dev: fencing: Only look for F_STONITH_ACTION when its needed
Low: fencing: Supply a sane default for provider
Dev: fencing: Ensure we actually terminate after SIGTERM
Medium: fencing: Do not block while agent calls are in progress
Medium: fencing: Fix incorrect providerd supplied by the shell or anyone else
Low: fencing: Clean up the error and option codes
Low: fencing: More flexibly locate the RA metadata
Low: fencing: Run metadata queries in the background too
Hg: Automated merge
Dev: fencing: Add missing headers
Low: Tools: Add a tool for injecting fake operations into the CIB
Dev: Tools: Refactor crm_inject
Low: crmd: Kill unused parameter
Low: crmd: Refactor lrm updates so they can be used elsewhere
Dev: Tools: Allow crm_inject to perform transitions
Dev: Remove useless test for NULL before free
Medium: Extra: Mark the pingd RA as deprecated
Low: cib: Return success when signing off from a file-based cib
Low: Build: Update version numbers in preparation for the next release
High: ais: Ubuntu needs a leading zero for directory modes
High: crmd: Bug bnc#527530 - Wait for the transition to complete before leaving S_TRANSITION_ENGINE
High: ais: Avoid excessive load by checking for dead children every 1s (instead of 100ms)
Medium: PE: Stop ptest from logging to syslog
Medium: Core: Disable syslog for any process that doesn't want its arguments logged
Medium: PE: Correctly log resources that would like to start but can't
High: PE: Bug lf#2171 - Prevent groups from starting if they depend on a complex resource which can't
Low: PE: Add regression test for lf#2171
High: PE: Bug lf#2170 - stop-all-resources option had no effect
High: PE: Disable resource management if stonith-enabled=true and no stonith resources are defined
Low: PE: Update regression tests for the previous change (disabling of resource management when no stonith resources are defined)
High: Core: Bug lf#2169 - Allow dtd/schema validation to be disabled
Medium: PE: Bug lf#2180 - Include node information for all failed ops
Medium: PE: Bug lf#2178 - Indicate unmanaged clones
Medium: ais: Include version details in plugin name
High: PE: Bug lf#2106 - Not all anonymous clone children are restarted after configuration change
Low: Build: Clean up top-level makefile, add spec file for fedora
Low: PE: New regression test for bug lf#2106
Medium: ais: Attempt to enable core file generation if it was disabled
Medium: crmd: Requery the resource metadata after every start operation
Low: PE: Remove developemental logging from output
Low: crmd: Fix generation of FSA graphs
Low: Tools: attrd - decrease the frequency with which we attempt to connect to the cib
Medium: Tools: crm_shadow - log diffs to stdout instead of stderr
High: ais: Gracefully handle changes to the AIS nodeid
Low: PE: Fix the metadata to advertise the correct type for stonith-timeout
Low: Build: Ignore warnings from libtoolize
High: cib: Repair the ability to connect to the cluster from non-cluster machines
Medium: Tools: crm_attribute - Suppress all logging with --quiet
Medium: Tools: Use -q as the short form for --quiet (for consistency)
Medium: Tools: attrd_updater - Suppress all logging with --quiet
Dev: Build: Remove debugging reminants
Low: crmd: Don't log operation rc if status != LRM_OP_DONE
Dev: Tools: Extra include needed for attrd_updater to use basename()
Low: Build: Force 'make dist' to be consistent with 'make tgz'
Low: Build: Fix 'make dist'
Medium: PE: Provide details for the maintenance-mode option
High: Extras: Add sctp support to the controld RA
Medium: Extra: Include SNMP MIB file for crm_mon (from Michael Schwartzkopff)
High: crmd: Prevent use-after-free with LOG_DEBUG_3
Medium: PE: Bug lf#2189 - Incorrect error message when unpacking simple ordering constraint
High: PE: Don't include master score if it would prevent allocation
High: ais: Bug rh#525589 - Prevent shutdown deadlocks when running on CoroSync
Dev: ais: Log child wait events based on time not iterations
Medium: Build: Re-enable asciidoc documentation
High: PE: Bug bnc#515172 - Provide better defaults for lt(e) and gt(e) comparisions
Low: PE: Fix regression test that did string-based lt and gt comparisions
High: PE: Bug lf#2197 - Allow master instances placemaker to be influenced by colocation constraints
Low: cib: Add missing license information
Dev: PE: First process peers that the master must be colocated with, then worry about peers that are colocated with us
High: PE: Make sure promote/demote pseudo actions are created correctly
Low: PE: Active additional regression test
Medium: PE: Correctly log the state of orphaned clone instances
Low: PE: Suppress debug messages concerning the stonith-complete action being unknown
Low: PE: Clean up the use of next_role
Medium: PE: Only re-instate target-role if it is less than the calculated one
High: PE: Prevent target-role from promoting more than master-max instances
Low: PE: New regression test for master target-role handling
Low: Doc: Add a docbook version of the confguration explained PDF
Low: doc: Clean up conversion, re-add missing text
Low: doc: Sanitize the first 3 chapters
Low: Doc: More formatting
Low: doc: Use regular tables (so they have a title and are listed in the index)
High: cib: Prevent use-after-free for remote plaintext connections
Medium: cts: Prepare for corosync testing
Medium: cib: Remove delay for remote plaintext connections
:High: cib: Correctly clean up when both plaintext and tls remote ports are requested
Low: ais: Use 'round' logging intervals
Dev: cib: Fix compile error
High: ais: Bug lf#2199 - Prevent expected-quorum-votes from being populated with garbage
High: cib: For validation errors, send back the full CIB so the client can display the errors
Low: tools: Use the XML_ATTR_EXPECTED_VOTES macro instead of literal text
Low: cib: Use a more appropriate temp file prefix
High: ais: Prevent deadlock - dont try to release IPC message if the connection failed
Low: doc: Increase TOC level to 2
Low: Build: Finalize the build process for generated html/pdf docs
Low: doc: Minor change to the generated index file
Low: Build; Ensure authors and licence files are packaged
Low: doc: Remove duplicate files
Low: Build: Minor cleanup of the doc makefile
High: crmd: Bug lf#2201 - Guard against possible cause of a segfault
Medium: PE: If no migrate_(from|to) action is defined, look for migrate instead
Dev: crmd: Check for _invalid_ votes, not valid ones
Dev: Build: Fix installation of chunked html docs
Dev: doc: Install ascii source of html docs
Medium: Build: Shell templates arent documentation
Low: Build: Honor docdir from autofoo if set
High: Tools: pingd - Fix a number of critical bugs (patch via Kazunori INOUE)
Dev: crmd: Downgrade to a debug message, seems normal on heartbeat
Dev: Tools: Fix pingd compilation on Darwin
Low: PE: Allign resource summaries nicely
Dev: PE: Only print an extra space when printing to std(err|out)
Low: cts: Convert handling of non-blocking commands to Popen(), ensure stdout is logged somewhere
Low: Build: Update version numbers
Hg: Automated merge
Low: Build: Fix compilation with old autofoo
Low: tools: RHEL-4 doesn't define a number of IPv6 and MLD message types
Low: Build: Move the 'build package' logic to GNUmakefile where it is usable on unconfigured source trees
Hg: Automated merge
Low: Build: Don't check for possible candidates for format attributes (-Wmissing-format-attribute)
Hg: Automated merge
Low: Build: Replace our home-grown --enable-pretty with AM_SILENT_RULES
Medium: Build: Don't automatically add -L/lib, it could cause 64-bit arches to break
Low: ais: Support the old name for PCMK_SERVICE_ID that had to be changed because corosync also used it briefly
Medium: PE: Provide a default action for resource-set ordering
Low: extras: Add the o2cb resource agent
Medium: xml: Add stopped as a supported role for operations
Medium: doc: Use Publican for docbook based documentation
Hg: Automated merge
Medium: cib: Send password prompt to stderr so that stdout can be redirected
Medium: cib: Indicate where the remote connection came from
High: cib: Ensure the loop for login message terminates
High: cib: Remote plaintext - Retry sending parts of the message that didn't fit the first time
Low: doc: Indicate the correct password field for remote cib connections
Low: doc: Set the correct doctype
High: cib: For remote connections, default to CRM_DAEMON_USER since thats the only one that the cib can validate the password for using PAM
Low: cib: Downgrade development logging
Medium: cib: Clean up logic for receiving remote messages
Medium: cts: Fix rsh handling when stdout is not required
Medium: tools: Make crm_mon functional with remote connections
High: cib: Finally fix reliability of receiving large messages over remote plaintext connections
Low: cts: Add a plaintext connection too
Low: Build: Add a mock build target
Low: cts: Minor cleanup
High: cib: Fix remote notifications
Low: tools: Only display the passord prompt if one wasnt supplied
Medium: cib: Create valid notification control messages
Dev: cib: Repair the TLS send/recv logic for remote connections
Dev: cib: Fix minor build issue
Low: cib: Re-calling cib_recv_remote_msg() when there is no TLS message causes a deadlock
Low: tools: Dont waste verticle space in crm_mon output
Medium: xml: Bug bnc#552713 - Treat node unames as text fields not IDs
Medium: PE: Bug lf#2206 - rsc_order constraints always use score at the top level
Medium: xml: Bug lf#2215 - Create an always-true expression for empty rules when upgrading from 0.6
High: PE: Bug lf#2209 - Clone ordering should be able to prevent startup of dependant clones
Low: PE: Bug lf#2209 - Add regression test
High: PE: Bug lf#2216 - Correctly identify the state of anonymous clones when deciding when to probe
High: PE: Bug lf#2225 - Operations that require fencing should wait for 'stonith_complete' not 'all_stopped'.
High: PE: Bug lf#2225 - Prevent clone peers from stopping while another is instance is (potentially) being fenced
Low: crmd: C_TIMER_POPPED is now quite normal thanks to the recheck timer, downgrade log message
Medium: PE: Silently fix requires=fencing for stonith resources so that it can be set in op_defaults
Low: doc: Add vectorized PE graphs (and include their source)
Low: cib: Allow get_channel_token() to return errors
Low: xml: Bug 2230 - Preserve the old stonith-enabled=false symantics from 0.6 during an upgrade
Dev: xml: Allow the auto-id base to be specified, because generate-id() isn't always unique
Medium: doc: Update the docs to reflect the 0.6/1.0 rolling upgrade problem
High: PE: Bug 2213 - Ensure groups process location constraints so that clone-node-max works for cloned groups
Low: PE: Bug lf#2213 - New regression test
Low: Build: Remove generated file
Hg: Automated merge
Hg: Automated merge
Hg: Automated merge
Medium: extra: Add the daemon parameter to the controld metadata
Low: fencing: Use the name 'portmap' for the mapping parameter
Low: Build: Target for mock builds of uncommitted code
Medium: ais: Some clients such as gfs_controld want a cluster name, allow one to be specified in corosync.conf
Low: doc: fix the defaults for pe-*-series-max
High: PE: Correctly unpack ordering constraints for resource sets to avoid graph loops
Low: PE: Regression test to set ordering
Low: extra: Two new system health RAs from Michael Schwartzkopff
Low: PE: Bug lf#2251 - Don't log uninstalled resource agents as errors
Low: cts: Start an instance of the lrmd so that the shell works
Low: cts: Also tell the shell to get out of the way
Low: cts: Make sure stonith-enabled is set to false so that subsequent shell command dont fail
Low: cts: Remive duplcate shutdown pattern
Dev: fencing: Fix the local escalation logic
Dev: crmd: Fix stonith sign-in/out
Hg: Automated merge
Low: cts: Look for stonith-ng patterns
Low: cts: Initiate stops asynchronously, but confirm with the synchronous version before returning
Dev: cts: Repair previous commit
Dev: cts: Use the higher-level StopaCM call
Dev: crmd: Resolve use-of-NULL during fecing shutdown
Low: cts: More flexible Process object
Dev: cts: Revert previous commit, the higher-level call is too smart and filters the stop action for 'stopped' node
High: PE: Bug lf#2153 - non-clones shouldn't restart when clones stop/start on other nodes
High: PE: Bug lf#2153 - Update regression tests
Low: PE: Add missing testcase
Low: Tools: Shell templates for gfs2
Low: PE: Turn of development logging
Medium: doc: Fill in the section on removing a node from an AIS-based cluster
Dev: PE: Revert development logging change
High: PE: Correctly anti-colocate with a group
Hg: Automated merge
Dev: fencing: Apprently 'hostlist' is the standard command name
Medium: Tools: Bug lf#2286 - Allow the shell to accept template parameters on the command line
Dev: cts: Fix copy/paste error
Medium: PE: Prevent non-multistate resources from being promoted through target-role
Medium: PE: Simplify the rsc_order syntax - don't make funky inferences based on score
Medium: PE: Implement serializing order constraints that dont cause restarts or inhibit migration
Medium: PE: Support serialized sets of resources
Low: PE: Minor clean up
Low: PE: Reuse get_flags() for consistency when unpacking ordering constraints
Medium: PE: Allow resource sets to be reused between ordering and colocation constraints
Low: PE: Update regression tests for the prior commit
High: PE: Bug lf#2153 - non-clones shouldn't restart when clones stop/start on other nodes - improved
High: PE: Bug lf#2153 - More improvements, not finished
High: PE: Bug lf#2153 - Finished
Hg: Automated merge
High: PE: Bug lf#2153 -
Low: PE: Minor code cleanup
Medium: PE: Only complain about target-role=master for non m/s resources
High: crmd: Ensure batch-limit is correctly enforced
Low: PE: Don't overwrite the value of batch-limit
Low: cts: Add the 'cmd' to the output of external commands
Dev: pengine: crmd: Ensure help text includes correct binary name
Low: tools: Rename crm_inject as crm_simulate
Medium: PE: Make crm_simulate a full replacement for ptest
High: crmd: Always connect to stonith
Dev: Tools: Cleanup of crm_simulate internals
Hg: Automated merge
Hg: Automated merge
Dev: Tools: crm_simulate - Implement --set-datetime
Low: PE: Use crm_simulate for regression tests and allow it to be run from anywhere
Low: Tools: crm_node - Accept -f as the short form of --force
Low: cts: CM_lha - Tweak some component fail patterns
Medium: Tools: Bug lf#2307 - Provide a way to determin the nodeid of past cluster members
Low: doc: Fix instructions for purging node status from the cib
Low: Build: Update version numbers
High: crmd: Ensure we have the latest status after a transition abort
Low: Build: Another helper target
Low: cts: Correctly process stonith options
Low: fencing: Allow the retrieval of device metadata with the admin tool
Low: Build: Tweak helper targets
Medium: PE: Include node attributes for the node to be fenced
Low: fencing: Add compatibility code to fence_legacy
High: Fencing: Overhaul the fencing daemon
Dev: Core: Filter out meta attributes beginning with a hash symbol
Dev: Fencing: Add missing header
Dev: Fencing: Move proctrack include to the correct location
Dev: Core: Filter attributes containing a colon too
Dev: Fencing: Include missing header
Medium: Fencing: Default to looking for hosts in the output of 'list'
Dev: Fecning: Ensure remote_op_list exists
Low: Build: New approach to shared spec files - write for fedora and account for differences with sed
Low: Build: Fix hg version in changelog
Low: Core: Improve assert diagnostics
High: crmd: Make sure we wait for fencing to complete before continuing
Low: Build: Fix building/installation of documentation in the absense of publican
High: fencing: Fix can_fence_host_with_device() logic and improve hostlist output parsing
High: Fencing: Include F_STONITH_CLIENTID in the remote query so that peers can create the remote_op_list entry
High: Fencing: Correctly parse arg maps and do not return a provider for unknown agents
Medium: Fencing: Reject unknown agents
Medium: cib: Pick an appropriate shadow directory for regular users
Medium: tools: crm_simulate - Fix in-place simulations
High: crmd: Prevent use-of-NULL when non-DCs get stonith callbacks
Low: Tools: crm_simulate - ensure stdout is flushed so that logging sent to stderr comes out in the correct order
Low: crmd: Give stonith-ng some breathing space to report timeouts itself
Low: crmd: Log text version of fencing error codes
Low: crmd: Sign up for fencing notifications - but don't act on them
Low: cts: Improve detection of idle DCs
Medium: Fencing: Fix the ability for fencing notifications to be sent to all active clients on all nodes
Low: Fencing: Universally add 10% padding to callback timeouts to allow the daemon time to reply
Medium: Tools: crm_simulate - Update failcount for injected operations that failed
Low: PE: Add missing regression test
Hg: Automated merge
Low: Build: Install and include regression tests in the -devel subpackage
Low: Build: Fix generation of suse-compatible spec file
Low: cts: Add support for QARSH as an SSH alternative
Low: Tests: Indicate regression test result with return status
Medium: Tools: crm_mon - Display the true fail-count, not the effective value
High: Agents: Prevent shell expansion of '*' when there are files in /var/lib/heartbeat/cores/root (Patch from Sébastien PRUD'HOMME)
Low: doc: Remove references to the old linux-ha.org site
Medium: Agents: ping - Clean up from Maros Timko
Medium: Tools: attrd - Only ignore the update if the attributes value is completely stable (ie. supplied, current, and stored all match)
Low: cts: Add documentation for --clobber-cib | -c option
Low: Tools: CLI regression tests no longer need to be run as root
Hg: Automated merge
Medium: Doc: Update the location for the project rpms
Low: Doc: Add the Cluster Labs brand for publican docs
Low: Agents: ping - Fast-track stop actions - Patch from Maros Timko
Medium: Tools: Bug LF#2312 - crm_mon - Prevent zombie child processes when using custom traps (Patch from Bernd Schubert)
Low: Doc: Tweak stonith example
Low: Tools: Log crm_node invocations
Low: cib: Remove redundant if-clause
Low: Build: Target for making TAGS include directives relative to current directory
Medium: Fencing: fence_legacy - Always parse stdin options
High: Fencing: Bug bnc#577007 - Correctly parse the hostlist output from stonith agents
Hg: Automated merge
Medium: Tools: Bug lf#2302 - Use the same resource printing logic for html and non-html output
Medium: ais: Set permissions to allow 'to_file' logging to function correctly
High: fencing: Account for stonith_get_info() always returning a pointer to the same static buffer
Low: Tools: crm_mon - Fix compilation on Darwin
Low: cts: Also look for stable DCs
Low: PE: Allow tests to exit with a known error code
Low: PE: Improved graphs when probing partially up nodes
High: crmd: Bug bnc#578644 - Improve handling of cancelled operations caused by resource cleanup
Hg: Automated merge
Low: PE: Repair minor merge conflict
High: ais: Bug lf#2340 - Force rogue child processes to terminate after waiting 2.5 minutes
Hg: Automated merge
Low: cts: Fix qarsh compatibility
Low: Build: Remove references to non-existant debian directory
Low: cts: Move CTS modules into python's site-packages directory
Low: build: Formatting in configure.ac
Low: cts: OCFIPraTest is a program not a module
Low: cts: Fix the install location for cts
Low: Build: Include some helpful defines in the common header file
Low: Build: Ensure py_sitedir is available in spec file
Low: Extra: ping(d) - Fix metadata
Hg: Automated merge
Low: Tools: Purge haresources2cib.py
Low: cts: Force shell commands to ignore validation
Low: doc: Build with the default brand unless the clusterlabs one is available
Low: Build: Update the spec file for 1.1
Low: Doc: Fix up-to-date workaround
Hg: Automated merge
Hg: Automated merge
Low: Build: Update versions
Low: cts: Allow some cluster managers to define alternate names to search for when stopping a node - Patch by Angus Salkeld
Dev: ais: Correctly determine when to force rogue children to terminate
High: PE: Suppress duplicate ordering constraints to achieve orders of magnitude speed increases for large clusters
Medium: PE: Some minor speed improvements to a performance bottleneck
Low: Build: Don't link against snmp libraries if we dont support snmp
Low: PE: Notes on future optimization
Low: PE: Minor speedup to heavily used functionality
High: PE: Allow startup probes to be disabled - their calculation is a major bottleneck for very large clusters
Med: cts: Support clusters without centralized logging
Low: Build: Update spec changelog
Hg: Merge with stable-1.0
Hg: More merging with stable-1.0
Low: PE: Fix regression test script
Low: xml: Prepare schemas for 1.1/1.2
Dev: xml: Add missing schema file
Dev: ais: Fix the child shutdown escalation logic
Medium: Core: Give signal handlers higher priority - patch based on Lars Ellenberg's work
Low: cts: Make further use of key_for_node() for corosync testing - Angus Salkeld
Low: RA: controld - Choose better defaults if the resource name contains gfs
Low: cts: Auto-detect the type of logging
Low: doc: Include a link to the LSB spec in the appendix for testing compatibility
High: PE: Correctly implement optional colocation between primitives and clone resources
Medium: Build: Improved check for the correct uint64_t print format
Low: cts: Simplify the remote logger
Hg: Automated merge
Medium: crmd: Bug bnc#578644 - Don't send operation updates for deleted resources
Dev: cts: Fix up log watching
Hg: Automated merge
Hg: Automated merge
Low: cts: Make LogAudit less verbose during normal operation
Medium: crmd: Log stonith notifications more succinctly
High: ais: Bug rhbz#525552 - Move non-threadsafe calls to setenv() to after the fork()
Low: Build: Make doc generation more robust
Low: cts: Minor tweak to LogAudit output
Low: fencing: Further refine stonith notifications
Hg: Automated merge
Low: fencing: Make sure notificaitons always have a sane originator listed
Medium: ais: getpwnam() is also not thread safe, move after the call to fork()
Low: Tools: crm_mon - Remove deprecated option from the html output
High: PE: Bug lf#2358 - Fix master-master anti-colocation
Hg: Automated merge
High: ais: Bug lf#2359 - expected-quorum-votes not correctly updated after membership change
Low: cts: Clean up and simplify testing to make it more consumable by other projects
Low: cts: Less special handling for BSC testing
Dev: cts: Fix path to random choice generator
Hg: Merge with stable-1.0
Low: cts: install the scenarios module
Low: cts: install the scenarios module
Low: cts: Use the ping RA instead of the deprecated pingd one and require the 'connected' attribute to also be defined
Medium: cts: Make sure we are really, _really_ stable before Reattach tests to prevent false negatives
Low: cts: Better offset logging for LogWatcher
High: ais: Bug lf#2359 - Default expected votes to 2 inside Corosync/OpenAIS plugin
Hg: Merge from stable-1.0
Low: cts: Fix busted Makefile merge
Low: Build: Update spec file
Medium: Tools: Transplant an updated hb_report tool from cluster-glue
Low: cts: Prefer local sources if they're available
Low: cts: Some minor speedups
Hg: Automated merge
Low: Build: Simplify the spec file
Low: Fix various spelling issues, as found by lintian - Patch by Frederik Schüler
Low: cts: Some minor speedups
Hg: Merge with stable-1.0
Low: tools: crm_simulate - Return error code when an invalid transition is produced
Low: PE: Update regression tests for the stonith-ng changes
Low: Tools: crm_report - Search for the pattern we need later when calculating node list
High: PE: Repair handling of unordered groups in RHS ordering constraints
Remove hash-bang from script that should not be executed directly
Low: Build: Remove execute permissions from c-files
Medium: PE: Correctly locate test directory when installed
High: ais: Do not count votes from offline nodes and calculate current #votes before sending quorum data
High: ais: Do not count votes from offline nodes and calculate current #votes before sending quorum data
Low: PE: Clean up regression test output
Medium: Build: Do not require supported stacks, the admin should select which stack to use and install it
Low: doc: Clean up creation of generated docs and make the brand configurable
Hg: Automated merge
Hg: Automated merge
Low: doc: Make sure the ascii txt files get pushed to clusterlabs.org too
Low: PE: Bug lf#2364 - Only log when orphans are found to be active
High: PE: Bug lf#2361 - Ensure clones observe mandatory ordering constraints if the LHS is unrunnable
High: PE: Bug lf#2317 - Avoid needless restart of primitive depending on a clone
Low: doc: Document resource reloads
Medium: PE: Bug bnc#586710 - Make sure migration ops use the correct meta options (eg. timeouts)
Low: Build: Update version numbers
Medium: Tools: crm_mon - Add support for older versions of SNMP - Patch derived from the work of sato yuki
Low: doc: Remove the proto-section on constraint lifetimes
Low: RA: ping - Indicate if we couldn't contact attrd
Medium: PE: Deprecate the lifetime tag in constraints
Low: Build: Include man pages back in the main package
Medium: PE: Bug bnc#586710 - Make sure migration ops use the correct meta options (eg. timeouts)
Low: Doc: xslt for producing man page docbook from RA metadata
Low: Doc: Allow man page filename to be specified
Medium: Provide man pages for PE And CRMd options
Medium: Core: Only change to core-dir if the caller was root or the daemon user
Low: Build: Split the CRM_DAEMON_DIR and HB_DAEMON_DIR variables but give them the same value for now
Low: Build: Only build docbook based man pages if xsltproc is present
Medium: RA: Create man pages for Pacemaker agents
Low: Build: Use the configure variable for xsltproc in Makefiles
Low: Build: Use the configure variable for xsltproc in Makefiles
Low: PE: Use the .scores files for finding current regression tests to package
High: crmd: Do not allow the target_rc to be misused by resource agents
Medium: Tools: Bug lf#2330 - Use \r in addition to \n for line endings
Medium: Tools: Bug lf#2330 - Add a blank line after the subject to indicate the beginning of the mail body
Medium: Tools: crm_mon - Use node uname in snmp/smtp/etc events
Medium: Tools: Bug lf#2330 - Move the blank line before the body text instead
Low: cts: Cache the quorum value to avoid redundant lookups
Low: cts: Improve signal handling
Low: cts: Summarize the results so far before tearing down the scenario if an exception is raised
Low: cts: More reliably detect when resource activity is still occuring
Low: RA: ping - Return error if we couldn't update the attribute, accept more values for 'debug'
Medium: cts: Finish after the requested number of iterations were completed
Low: cts: Reduce audit logging
Low: cts: Turn on debugging for the ping RA
Low: cts: Install dummy lsb agent if needed
Low: cts: Replace LSBDummy with a script that is actually LSB compliant
Low: cts: Provide a fallback for HA_VARRUN in case .ocf-directories doesn't exist
Low: Build: Update the changelog
Low: doc: Include analytics tracker in the clusterlabs brand
Low: Build: Pacemaker now supports the older net-snmp versions
Low: PE: Downgrade a warning to 'info'
Low: Build: Support building a specific version
High: crmd: Remove race condition that could lead to multiple instances of a clone being active on a machine
Low: PE: Considate and unify logging of fencing causes
Low: cts: Wait a little longer for fencing logs
Low: Build: Override broken py_sitedir on opensuse11.2
Low: Tools: crm_report - Minor formatting change
Low: cts: Fix spelling error
Medium: Tools: crm_report - Extract log master logs to the correct location
Hg: Merge with stable-1.0
Dev: Build: Fix HB_DAEMON_DIR definition
Low: cts: Add the ability to repeat a series of named tests in a loop
Low: Build: Force a valid value for py_sitedir on openSUSE
Low: Build: Fix dependancies for -devel package on SUSE
Low: cts: Refer to the correct env variable when deciding to re-enable fencing
Low: cts: Remove invalid logfile check and prevent hang when ssh peers die
Low: cts: Clean up scenario selection
Low: cts: Only return applicable audits
Medium: Tools: crm_mon - Display node attributes. Patch by Yuusuke IIDA
Low: Build: Fix py_sitedir on opensuse
Merge with 1.0
High: ais: Look for the correct conf variable for turning on file logging
Low: doc: Publican barfs on long lines
Low: doc: Clean up artifacts in the LSB instructions
Low: doc: Publican barfs on long lines
Low: Build: Build the index after the publican docs
Low: doc: Publican thinks this is a long line when using computeroutput
Low: doc: Publican thinks this is a long line when using computeroutput
High: crmd: Bug lf#2379 - Ensure the cluster terminates when the PE isn't available
Low: Build: Update the changelog for 1.0.8-2
High: crmd: Ensure we dont get stuck in S_PENDING if we loose an election to someone that never talks to us again
Hg: Last merge with 1.0, from now on we cherry pick
Hg: Automated merge
Low: crmd: Fix duplicate definition of O_DC_TIMER_RESTART
Medium: RA: Fix o2cb metadata
Medium: RA: Install man pages with ocf:pacemaker: prefix
High: crmd: Fix memory leaks exposed by valgrind
Low: Tools: Update regression test output for new schema version
Low: Tools: Update regression tests for some minor output changes
Low: Build: Add RA man pages to the spec file
Add rpmlint control file
High: crmd: Do not ignore action timeouts based on FSA state
Low: PE: A default cib
Low: Build: Bug lf#2394 - Only require lm_sensors-devel on arches that support it
Low: doc: Update the default value for globally-unique
Low: PE: Bug bnc#586710 - Make sure migration ops use the correct meta options (update regression tests)
Medium: PE: Bug lf#2391 - Ensure important options (notify, unique, etc) are always exposed during resource operations
Low: PE: Minor loop optimization
Low: PE: Bug lf#2391 - Ensure important options are always exposed during resource operations (update regression tests)
Low: Tools: Bug bnc#591168 - Support short form of crm_simulate --quorum
Low: PE: Correctly detect test cases
Hg: Merge with 1.1
Hg: Merge with 1.1
Low: Shell: Allow the shell to be run from the source tree
Low: Shell: Allow crm binary and tools path to be specified on command line for BuildBot
Low: Shell: Fix regression test option parsing
Low: Shell: Add option for saving user output to a file for buildbot
Low: Build: Add a default target that dose something useful if autofoo hasn't been initialized
Low: PE: Use a custom directory for CIB shadow files
Low: Tools: Use a custom directory for CIB shadow files (also for buildbot)
Low: PE: Remove test artifacts during make clean
Low: Tools: Allow a path prefix to be set so the tests can run with binaries from the build tree
Low: cib: Provide a user-specific shadow directory when possible
Low: Fencing: Delay system initialization until after we handle possible metadata queries
Low: Tools: Add the ability to locate the shadow config file
Low: Build: Remove spurious executable permissions from source files
Low: Build: Update rpmlint filters
Low: Build: Do not specify explicit lib package dependencies (rpmlint)
Low: Build: Resolve more rpmlint errors
Low: Shell: Include important files in regression output for buildbot mode
Low: Shell: Found a better way to get results when called form buildbot
Hg: Merge with 1.1
Low: Build: Add package for generated manpages from xml
Low: Build: Add missing shell install dependancy
Low: Build: Add another runtime prereq for the shell
Low: RA: Drop the default timeouts for the Dummy agent
Low: shell: Make regresion tests less fragile by using ocf:pacemaker:Dummy instead of ocf:heartbeat:Dummy
Low: Core: Only try to change the current directory for cluster users
Low: shell: Use Dummy instead of IPaddr for metadata test
Low: Shell: Make sure we get the metadata from ocf:pacemaker:Dummy
Low: RA: Uodate Dummy metadata
Low: Shell: ocf:pacemaker:Dummy just uses /var/run for state files
Low: PE: Fix verbose mode for regression tests
Low: PE: Remove accidentally checked in files
High: crmd: Bug 2401 - Improved detection of partially active peers
High: PE: Bug lf#2384 - Fix intra-set colocation and ordering
Low: PE: Bug lf#2384 - Fix intra-set colocation and ordering (regression tests)
Low: Tools: Prefer strlen() in crm_mon
Low: PE: Change the regression test output filenames to simplify the Makefile.am install logic
High: PE: Bug lf#2383 - Combine failcounts for all instances of an anonymous clone on a host
Low: PE: Bug lf#2383 - Combine failcounts for all instances of an anonymous clone on a host (regression tests)
Low: Build: Prepare for --with-rsctmpdir from agents/glue
Low: crmd: Bug lf#2387 - Add missing fields to the metadata for use by the shell
Low: crmd: Log message doesn't need to be an error
Low: PE: Fix verbose output handling
Medium: PE: Check for use-of-NULL in dump_node_scores()
High: PE: Rewrite native_merge_weights() to avoid Fix use-after-free
Low: PE: Rewrite native_merge_weights() to avoid Fix use-after-free (regression test)
Low: PE: Fix log message, reduce to debug
Low: PE: Bug bnc#586613 - Don't create scaffolding for ordering sets with a single resources
Medium: PE: Ignore colocation weights from clone instances
Medium: PE: Print allocation scores as +/-INFINITY when appropriate
Low: PE: Print allocation scores as +/-INFINITY when appropriate (regression tests)
Low: doc: Include a reasonable conversion of the Clusters from Scratch document
Low: doc: Update diagrams and generate from svg
Low: doc: Replace openais with corosync where appropriate
Low: doc: Refresh the intro from Pacemaker Explained
Low: PE: Remove development logging
Low: doc: Fix comparison between groups and ordered sets
Low: doc: Alternate table background for alternate rows
Low: doc: Build the brand images too
Low: Doc: Swicth to a blue-based brand consistent with the website
Hg: Merge with 1.1
High: ais: Use the threadsafe version of getpwnam
High: ais: Fix two more uses of getpwnam() in non-thread-safe locations
High: ais: Need to find a better and thread-safe way to set core_uses_pid. Disable for now.
High: PE: Bug lf#2403 - Enforce mandatory promotion (colocation) constraints
Low: PE: Bug lf#2403 - Enforce mandatory promotion (colocation) constraints (regression test)
Medium: PE: Bug lf#2410 - Don't complain about missing agents during probes of a-symetric clusters
Medium: PE: Bug lf#2169 - Allow constraints to apply to specific instances of clones
Medium: Core: Only use the named schema if the validation passes
Medium: Core: Only update configurations conforming to unsupported schemas
Medium: PE: Bug lf#2169 - Allow constraints to apply to specific instances of clones (regression tests)
Medium: PE: Allow ordering constraints to apply to specific instances of clones
Low: PE: Allow ordering constraints to apply to specific instances of clones (regression tests)
Medium: PE: Bug lf#2066 - Clean up failcount after it expires
Low: PE: Bug lf#2066 - Clean up failcount after it expires (update regression tests)
High: Core: Bump the feature set due to the new failcount expiry feature
Low: PE: Update feature set field in regression test outputs
Medium: PE: Add support for failover domains
Medium: PE: Allow node's to have a non-zero default score
High: ais: Ensure the list of active processes sent to clients is always up-to-date
High: ais: Fix previous commit, actually return a result in get_process_list()
Hg: Merge with 1.1
Low: PE: Update a regression test output with the new version number
Low: PE: Reduce logging
High: PE: Don't be so quick to pull the trigger on nodes that are coming up
High: crmd: Use global fencing notifications to prevent secondary fencing operations of the DC
Low: cts: Reduce noise in CTS output
Medium: crmd: Fix call to send_stonith_update()
Medium: cib: Bug lf#2352 - Changes to group order are not detected or broadcast to peers
Low: cib: Remove development logging
Low: doc; Update for F-13 installer
Low: crmd: Improved logging for fencing notifications
High: crmd: Send erase_status_tag() calls to the local CIB when the DC is fenced, since there is no DC to accept them
Medium: ais: Correctly set logfile permissions in all cases
Low: crmd: Provide more logging info for Bug lf#2382
Low: doc: Almost finished the installation section of Clusters_from_Scratch
Medium: ais: Make sure debug messages make it into the logfiles too
Low: crmd: Include operation interval in logged action
Low: cts: Be slightly more verbose during scenario setup/teardown
Low: cts: Refine watch patterns for stonithd test
Medium: crmd: Arrange for only one node to clean up after a fenced DC node
Low: tools: crm_report - Add an additional 5s fuzz when collecting logs
Hg: Automated merge
Hg: Automated merge
High: PE: Bug lf#2412 - Correctly locate clone instances by their prefix
Low: ais: Make a better attempt at determining an accurate born-on value from corosync
Low: ais: Make a better attempt at determining an accurate born-on value from corosync (take 2)
Low: Build: Minor cleanup of temp directory handling
Low: Tools: Handle unset time formats gracefully
Medium: tools: crm_mon - Enable 'connectivity' mode for 'ping' resources too
Low: cts: Add extra ssh options to detect connection failure. Indicate logfile truncation
Low: RA: Improved logging from ping for those that want it
Low: cts: A slightly better pattetn to look for in ComponentFail
Low: cts: More noise reduction for LogAudit
Low: tools: Tweak crm_report help text
Low: Build: Clean up release Makefile helpers
Low: Build: Bump release version
High: crmd: Bug lf#2414 - Prevent use-after-free of the PE connection after it dies
Low: crmd: Remove dead code
Low: cts: Correct minor spelling mistake
High: crmd: Bug lf#2414 - Prevent use-after-free of the stonith-ng connection
Low: ais: Bug lf#2414 - Indicate the valgrind binary being used
Low: Build: Fix the date format for the changelog
Low: cts: Make sure the valgrind suppression file exists and isup-to-date
Low: cts: Update the suppression file for corosync
Low: cts: Do not attempt to shoot components being profiled by valgrind
High: Core: Bug lf#2414 - Prevent use-after-free reported by valgrind when doing xpath based deletions
Low: cts: Add stonith-ng to the list of components to trace with valgrind
Low: cts: Some valgrind leaks to ignore
Low: tools: Clean up all memory to make spotting PE leaks easier
Medium: cib: Correctly free the cib contents at signoff when in file-based mode
Low: PE: Fix valgrind handling in regression tests
High: PE: Fix memory leaks reported by valgrind
High: Core: Fix memory leak in replace_xml_child() reported by valgrind
Low: Build: Tweak changelog generator
Low: Build: Update changelog
Low: PE: Remove developemnt logging
Low: Build: Finalize the 1.1.2 release
Low: Resolve coverity dead-code defects
Low: Resolve coverity has-no-effect defects
Medium: Resolve coverity NULL_RETURNS defects
Medium: tools: Resolve coverity UNINIT defects
High: ais: Resolve coverity CONSTANT_EXPRESSION_RESULT defects
Low: RA: ping - Factor in the number of hosts in the timeout calculation
Low: doc: Finish conversion to blue theme
Low: doc: Fix imagedata element and set correct pacemaker version in Book_Info
Low: doc: Add version history and author info for Clusters_from_Scratch
Low: doc: Start building Clusters_from_Scratch
Low: stonith: Fix coverity CHECKED_RETURN defect
High: ais: Revert part of an accidental commit
medium: Fix coverity NEGATIVE_RETURNS defects
Medium: ais: Fix coverity FORWARD_NULL defects
Medium: PE: Fix coverity FORWARD_NULL defects
Medium: Tools: Fix coverity OVERRUN_DYNAMIC defects
Medium: Core: Fix coverity OVERRUN_DYNAMIC defects
Medium: Fencing: Fix coverity REVERSE_INULL defects
Medium: Core: Fix coverity SIZECHECK defect
Medium: Fix coverity UNUSED_VALUE defects
High: Core: Bug lf#2401 - Backed out changeset 6e6980376f01
Hg: Merge with bugfix branch
Low: Build: Bump spec version
Hg: Automated merge
Hg: Automated merge
Medium: ais: Backed out changeset 2e096a41a5f9
Hg: Branch merge
Low: doc: Update and format the DRBD section of clusters from scratch
Low: doc: Update links to Fedora installation docs in cluster from scratch
Low: doc: Re-indent the active/active chapter and move the lead paragraph to the right place
Low: doc: Minor tweaks
Low: doc: DRBD was upstreamed in 2.6.33
Low: doc: Merge the ClusterFS and Active/Active chapters. Highlight user commands.
High: ais: Bug lf2401 - Improved processing when the peer crmd processes join/leave
Low: Tools: Add valgrind support to regression tests
Low: Tools: Fix memory leaks in CLI tools so that we can spot leaks in the libraries we're using
Low: cib: Fix memory leaks in file-based connections
High: cib: Also free query result for xpath operations that return more than one hit
Medium: cib: xpath - Allow all hits to be deleted, allow the no_children option to return multiple hits
Low: PE: Simplify set_working_set_defaults() and only destroy domains in cleanup_calculations() if its non-NULL
Low: PE: Arrange for memory leaks to cause regression tests to fail
Low: Tools: Resolve memory leak and avoid stderr
Low: Tools: crm_resource - Report crmd connection errors
Low: Shell: Update expected outputs
Low: Tools: rpmlint doesn't like empty files
Low: Build: Package the PE suppressions file for valgrind
Low: Build: Package the PE suppressions file for valgrind in the right location
Low: PE: Remove unneeded valgrind suppressions
Low: Doc: lf#2424 - Ensure man pages created from automake scripts contain the correct binary name
High: Tools: Drop the 'pingd' daemon and resource agent in favor of ocf:pacemaker:ping
Low: Tools: Add man pages for crm_standby crm_failcount and crm_master
Low: Tools: Add man pages for cibpipe
Low: Fencing: Provide man pages for installed binaries
Low: shell: Add support for --version so that help2man can generate a manpage
Low: Build: Just use the changeset ID for BUILD_VERSION
Medium: Simplify --version output and add BUILD_VERSION and CRM_FEATURES to a separate command
Medium: Fencing: Initial implementation of manual fencing ACK - might even work
High: ais: Drop support for whitetank in the 1.1 release series
Low: RA: Restore the pingd metadata to keep the shell happy
Low: RA: Ensure timeout is set even when no hosts are defined (ie. meta_data ops)
High: Resolve coverity RESOURCE_LEAK defects
Low: Build: Debug man page generation in chroot
Low: tools: Fix crm_report --version so that its usable with help2man
Low: Build: Make sure required binaries are in the path
Low: Build: Fix PATH update
High: Resolve coverity RESOURCE_LEAK defects
High: tools: crm_simulate - Resolve coverity USE_AFTER_FREE defect
High: PE: Fix potential use-after-free defect from coverity
Low: tools: crm_report - Preprocess --help and --version to avoid needless initialization
Low: Tools: Allow --version option to be used on legacy scripts from the source directory
High: Tools: Fix use-after-free defect from coverity
Low: Build: Purge old and unmaintained man pages
Low: Tools: crm_mon - Fix compilation on OSX
Medium: PE: Ensure deallocated resources are stopped
High: PE: Bug lf#2422 - Ordering dependencies on partially active groups not observed properly
Medium: PE: If there are no compatible peers when interleaving clones, ensure the instance is stopped
Low: PE: Bug lf#2422 - Ordering dependencies on partially active groups not observed properly (regression test)
High: PE: Make the current data set a global variable so it doesn't need to be passed around everywhere
High: PE: Fix colocation for interleaved clones
High: PE: Fix colocation with partially active groups
Hg: Merge with devel
Low: doc: Highlight user commands in the installation section of C-from-S
Low: cts: Always clean out valgrind logs
High: PE: Fix previous merge
High: cib: Fix the application of unversioned diffs
High: cib: Remove old developmental error logging
High: PE: Prevent endless loop when looking for operation definitions in the configuration
Medium: PE: Don't overwrite existing meta attributes (like timeout) for notify operations
High: PE: Bug lf#2424 - Use notify oepration definition if it exists in the configuration
Medium: PE: Bug lf#2426 - stop-all-resources shouldn't apply to stonith resources
Low: Fencing: Bug lf#2431 - Reduce stonith-ng logging
Low: Shell: Don't exit with an error for the --version option
Low: Shell: Provide access to cibadmin and friends so we can build the crm.8 manpage
Low: Build: Finally get manpages built for wrapper scripts
Low: shell: Debugging man page generation
Low: Shell: Differentiate between all three calls to usage()
High: Tools: Fix recently introduced use-of-NULL
Low: shell: Still debugging man page generation
Low: Shell: Ensure we always use the shell modules from the source tree when building manpages
Low: Shell: Remove development logging
Low: Build: We need all the python bits now that we run the shell as part of the build
Low: Build: Purge old and unmaintained man pages
High: PE: Bug lf#2433 - No services should be stopped until probes finish
Low: crmd: Reduce logging when a resource to be deleted isn't found
Low: PE: Pre-expand probe ordering constraints for complex resources
Low: Build: OSX needs libgen.h for basename
Low: Tools: Allow ptest and crm_simulate to work with configs that lack a status section
Low: Core: Reduce logging for failed xpath searches
Low: PE: Fix up regression tests
Low: PE: Remove redundant if-clause
Medium: Build: Do not enable the -ansi compiler option by default, prevents use of strtoll()
Medium: PE: Bug lf#2435 - Support colocation sets with negative scores
Low: PE: Bug lf#2435 - Support colocation sets with negative scores (regression tests)
Medium: ais: Allow Pacemaker to consume quorum and membership from cman
Medium: PE: Fix memory leak reported by valgrind in unpack_order_set()
Low: cman: Incorporate feedback from chrissie
High: PE: Fix use-after-free in order_actions() reported by valgrind
High: PE: Bug lf#1959 - Fail unmanaged resources should not prevent other services from shutting down
Low: RA: controld - Start the regular RHCS daemons if we're using membership from cman
Medium: cman: Tell fenced about completed stonith events
High: crmd: Bug lf#2439 - Handle asynchronous notification of resource deletion events
High: crmd: Bug lf#2439 - cancel_op() can also return HA_RSCBUSY
Low: tools: crm_node - Do not loop when connecting to corosync
Medium: Build: Fix the installation directory for the Shell's python modules
Low: Build: Use pyexecdir for cts too
Low: cts: Make use of the ccm_ignore patterns
Low: crmd: Pass in the correct result when processing synchronous resource deletion operations
High: ais: Fix unterminated recursion in init_ais_connection()
Medium: tools: Make sure crm_resource waits for the completion of crmd operations it sent
High: crmd: Fix handling of async resource deletion results
Low: tools: crm_resource - Skip past the welcome message
Low: tools: crm_resource - wait for results of --fail operations too
Medium: stonith: Support the poweron/poweroff actions
Hg: Merge with 1.1
Medium: ais: Add the ability to use quorum directly from corosync
Low: ais: Strip out the remaining support for openais
High: ais: Never allow quorum plugins to affect connection to the pacemaker plugin
High: cman: Grab the initial membership when we connect
High: cman: Keep the list of peer processes up-to-date
Low: cman: Debugging
Medium: ais: Use the original service address unless building for cman
High: cman: Correctly process membership/quorum changes from the pcmk plugin. Allow other message types through untouched
Medium: cman: Switch to CPG for peer process messaging
Medium: cman: Autodetect when cman membership should be used
High: cman: Populate sender details for cpg messages
Medium: ais: Fix autodetection of cman-based quorum
High: cman: Filter directed messages not intended for us
Medium: Core: Switch to coredir when not running from a regular login shell
High: mcp: New master control process for (re)spawning pacemaker daemons
Low: ais: Move assertion to a more relevant location
High: Introduce two new cluster types 'cman' and 'corosync', replaces 'quorum_provider' concept
Low: Build: Hook up building of the new MCP
Low: Build: cman based clusters don't use a plugin at all, so no need to change the service id
High: cman: Make sure we can compile without cman present
High: mcp: Make sure we can compile the mcp without cman present
High: mcp: Make sure we can compile the mcp without cman present
High: Support starting plugin-based Pacemaker clusters with the MCP as well
Low: Build: add cman support as a non-default build option
Low: Build: Easily build package variants
Low: cts: Clean up after a leaking corosync
High: mcp: Rename the mcp binary to pacemakerd and create a 'pacemaker' init script
Low: Build: Allow building against CMAN if present
High: mcp: Fix macro expansion in init script
High: crmd: Only need to request quorum update for plugin based clusters
Medium: Build: Include calls to chkconfig in the post and preun scripts for rpm
Low: cts: Fix call to scenario.TearDown()
High: crmd: Allow non-DC nodes to clear failcounts too
High: crmd: Allow non-DC nodes to clear failcounts too
High: PE: Correctly handle pseudo action creation
High: stonith: Map poweron/poweroff back to on/off expected by the stonith tool from cluster-glue
High: PE: Correctly handle pseudo action creation
High: crmd: Disable age as a criteria for cman based clusters, its not reliable enough
High: crmd: Base DC election on process relative uptime
High: mcp: Resend our process list after every CPG change
Low: crmd: Extra debugging
High: mcp: Read config early so we can re-initialize logging asap if daemonizing
High: Core: Correctly unpack HA_Messages containing multiple entries with the same name
Low: Remove debug messages now that proc updates are correctly parsed
High: tools: crm_report - corosync.conf wont necessarily contain the text 'pacemaker' anymore
Low: tools: crm_report - Some extra debug information
Low: tools: crm_report - Grab everything up to EOF for tests that haven't finished yet
Low: Build: Correctly expand package features in shell scripts
Low: Build: Allow easy building of the working directory
High: crmd: Fix assertion failure when performing async resource failures
Hg: Automated merge
High: ais: Sign everyone up for peer process updates, not just the crmd
Low: cts: Add two new stacks - flatiron with mcp, and cman
High: ais: Get an initial dump of the node membership when connecting to quorum-based clusters
High: ais: Move the code for finding uid before the fork so that the child does no logging
Low: crmd: Remove unused function prototype
Low: mcp: Modify the logging of exiting client daemons
High: mcp: Handle CS_ERR_TRY_AGAIN when connecting to libcfg
High: crmd: Factor the nanosecond component of process uptime in elections
High: cman: Update the ringid for cman based clusters
Low: crmd: Additional logging for fenced notifications
Low: crmd: Improved debugging for an error case
High: cib: Restructure the 'valid peer' check for deciding which instructions to ignore
High: crmd: Make sure the membership cache is accurate after a sucessful fencing operation
High: cman: Make sure our common hooks are called after a cman membership update
High: crmd: Offer crm-level membership once the peer starts the crmd process
High: crmd: All nodes should see status updates, not just he DC
High: crmd: Include the action for crm graph operations
Hg: Automated merge
High: ais: Handle CS_ERR_TRY_AGAIN in more cases
High: tools: crm_node - Repair the ability to connect to openais based clusters
Low: tools: crm_resource - Minor output change
High: PE: Correctly detect when there is a real failcount that expired and needs to be cleared
High: PE: Correctly detect when there is a real failcount that expired and needs to be cleared
High: Core: crm_count_member() should only track nodes that have the full stack up
Low: cts: Explicitly specify a stack type for crm_node commands
Low: crmd: Only log the failcount op if something is wrong
High: ais: Don't supply a process list in mcp-mode
Low: PE: Update regression tests for the failcount changes
Hg: Automated merge
High: crmd: Make sure we always poke the FSA after a transition to clear any TE_HALT actions
Low: crmd: Receiving I_STOP in S_INTEGRATION is actually pretty normal if the DC is going down at the same time as some of its peers
Low: Core: Display 'heartbeat' instead of NULL when HA_cluster_type is unset
Low: doc: Account for recent Fedora packaging changes
Low: doc: Adjust example cluster.conf to reflect best practices
Low: doc: Minor clarifications and reformatting
Low: tools: Generally make crm_report more robust
Low: PE: Don't log resource states unless debug is on
Low: tools: Minor cleanup to crm_report output
Low: crmd: More informative error message when fenced cannot be contacted
Low: cts: Ensure fenced is started/stopped for cman based testing
Low: cts: Merge the --stonith and --fencing options
Low: tools: Correctly handle --crm-log in crm_report
Low: doc: Simplify the sample cman configuration in Clusters from Scratch
Low: cts: Shortcuts for the two most common fencing types
Medium: tools: Backport some minor enhancements from hb_report (up to cs: 596a5c4c0df4)
Low: cts: Use the cman init script instead of starting fencing manually
Medium: stonith: Default 'pcmk_host_check' to static-list if a lost list is provided
Medium: ais: Clean up initialization and allow us to manually specify the type of cluster to connect to
Medium: Core: Restore API compatability with 1.0 for crm_log_init()
Low: mcp: flags argument to cfg_shutdown_callback() is an enum not a bitfield
High: mcp: Add missing headers when built without heartbeat support
High: ais: The cluster type needs to be set before initializing classic openais connections
Medium: shell: Use crm_attribute directly instead of the deprecated crm_standby and crm_failcount wrappers
High: stonith: Correctly parse pcmk_host_list parameters that appear on a single line
High: PE: Avoid creating invalid ordering constraints for probes that aren't needed
High: stonith: Advertise stonith-ng options in the metadata
High: Shell: Complete the transition to using crm_attribute instead of crm_failcount and crm_standby
Low: stonith: Tweak the metadata output
Medium: Shell: Update the expected regression test outputs
Medium: Shell: Update the expected regression test outputs (again)
Low: ra: Stateful - No need for the timeouts to be so high
Low: cts: We don't care about heartbeat v1 agents anymore
Low: mcp: Add variables for controlling valgrind to the init script
Low: Shell: Another regression test update
Low: cts: Use crm_attribute directly rather than the crm_standby wrapper
Low: cts: Use --openais instead of -A for crm_node as older versions didnt support -A
Low: cts: Increase the timeout for the Stateful resource, crm_master might take a while
Low: mcp: Add (disabled) MALLOC variables for debugging memory issues
High: cib: Attempt to resolve memory corruption when forking a child to write the cib to disk
Low: Build: Explictly require resource-agents so that we can build man pages for the RAs when heartbeat support isnt enabled
High: cib: Correctly free memory when writing out the cib to disk
Medium: stonith: Allow all host-specific parameters to be turned off for devices that control a single machine
Medium: tools: Impliment cman support in crm_node
Low: cman: Avoid needless processing in callbacks for cman based clusters
High: tools: crm_node - Use the correct short option for --cman
Low: tools: crm_report - Add the ability to specify the cts control file
High: ais: Correct the logic for conecting to plugin based clusters
Low: Tools: crm_report - Correct the --features output
Low: mcp: Resolve incoherent-subsys warning from rpmlint
Low: mcp: Add support for --features to the pacemaker daemon
Low: cts: Create fencing resources first so that the shell doesn't complain
Medium: mcp: Use SIGSEGV instead of SIGKILL for daemons that wont shut down so a core is created and we can see what it was up to
Low: ais: Log when CPG or other corosync operations return CS_ERR_TRY_AGAIN
Low: mcp: Minor tweak to the features output
Low: Core: For full compatability, crm_log_init needs to be a real function
Medium: mcp: 5 retries is plenty for sending CPG messages
High: ais: Guard against saturated cpg connections
Medium: PE: Include the prefix for clone children when printing the clone
Low: PE: Enforce clone ordering in the absense of colocation constraints (test case)
High: PE: Bug lf#2453 - Enforce clone ordering in the absense of colocation constraints
Medium: PE: Turn down development logging
High: PE: Correctly order clone startup after group/clone start
Low: PE: Correctly order clone startup after group/clone start (additional test case)
Low: PE: Turn logging back down (again)
High: Core: New developmental logging system inspired by the kernel and a PoC from Lars Ellenberg
Medium: Core: Add tracability to other pacemaker libs. The __start___verbose symbol is not available for a library unless explicitly used.
Low: Tools: crm_report - Improve event patterns
High: crmd: Prevent everyone from loosing DC elections by correctly initializing all relevant variables
Medium: Core: Add tracability to other pacemaker libs. The __start___verbose symbol is not available for a library unless explicitly used.
Low: Tools: crm_simulate, ptest - Add the ability to easily see which tracepoints where updated
Low: Build: Correct the linking order so that libpe_status is linked after libpengine.
Low: Build: Explicitly add CURSESLIBS dependency to libpe_status.
High: crmd: Ensure we activate the DC timer if we detect an alternate DC
Low: crmd: Additional logging
High: PE: Wait until stonith is available, don't fall back to shutdown for nodes requesting termination
Low: Core: Fix logging compiliation for older versions of gcc
Low: PE: Remove redundant if-clause
High: mcp: Correctly initialize the string containing the list of active daemons
Medium: PE: Fix memory leak when unpacking recurring operations
Low: Core: Fix memory leak when activating trace logging
Medium: PE: Fix potential use-of-NULL when calculating the set of required actions
High: mcp: Fix the expansion of the pid file in the init script
Low: cts: Allow alternatives to syslog-ng to be used easily
Low: cts: Correctly log which nodes LogAudit is failing on
High: mcp: Tell chkconfig we need to shut down early on
Medium: Tools: crm_mon - Only display attributes of online nodes (and sort them)
Low: RA: ping - Indicate that host_list is a required parameter in the metadata
Low: Doc: Only generate txt and html versions of Pacemaker documents by default
High: stonith: Bug lf#2461 - Prevent segfault by not looking up operations if the hashtable hasn't been initialized yet
Medium: Tools: Bug lf#2477 - Have crm_resource --locate indicate which instances are promoted
High: PE: Do not demote resources because something that requires it can't run
Low: Build: Disable the -Wcast-qual compiler flag, it borks on glib headers
Low: PE: Bug lf#2479 - Display managed/unmanaged state for resources when grouped by node in crm_mon
Low: PE: Update regression tests
Low: doc: Fix command for creating drbd resource
High: crmd: Bug lf#2458 - Ensure stop actions always have the relevant resource attributes
Low: crmd: Bug lf#2458 - Ensure stop actions always have the relevant resource attributes - update regression tests
Low: doc: Fix the spec file for the ClusterLabs publican brand
Low: Build: Simplify the installation of our publican brand
Low: PE: Indicate failed resources when the crm_mon output is grouped by node
Low: PE: Improve the accuracy of summary data in regression test output
High: Tools: Bug lf#2456 - Fix assertion failure in crm_resource
Low: Core: Add line/file information to trace messages
Medium: cib: Bug lf#2469 - Clean up disconnected remote connections
High: PE: Rewrite the ordering constraint logic to be simplicity, clarity and maintainability
Low: PE: Allow an existing status calculation to be reused for allocation
Medium: PE: Resolve memory leak in new ordering code reported by valgrind
High: PE: Correct use-after-free introduced in the prior patch
Medium: Tools: crm_simulate - Resolve memory leak reported by valgrind
Medium: PE: Ignore failed actions on nodes that have been shutdown - Patch from Hideo Yamauchi
Low: PE: Bump the SO number due to the ordering rewrite
Low: Doc: Remove generated index.html, its no longer useful/relevant
Low: doc: Bug lf#2488 - Include details on enabling the server-status URL for Apache
Low: crmd: Bug lf#2473 - Better logging of failed stonith operations
High: stonith: Bug lf#2473 - Gracefully handle remote operations that arrive late (after we've done notifications)
Medium: Tools: crm_simulate - Resolve memory leak reported by valgrind (really this time)
High: PE: Bug lf#2476 - Repair on-fail=block for groups and primitive resources
High: Stonith: Bug lf#2473 - Ensure stonith operations complete within the timeout and are terminated if they run too long
High: crmd: Prevent assertion failure for stop actions resulting from cs: 3c0bc17c6daf
High: PE: Prevent segfault by ensuring the arguments to do_calculations() are initialized
Low: doc: Fix compilation on F-14, seems make changed its wildcard matching rules
High: Stonith: Use the timeout specified by the user
High: stonith: Bug lf#2473 - Add the timeout at the top level where the daemon is looking for it
High: stonith: Bug lf#2473 - Ensure timeouts are included for fencing operations
Low: Build: Include example sysconfig file
Low: PE: Bump some logging to NOTICE
Medium: PE: Set the default log level to notice (was info)
Medium: Core: Make trace-logging optional, it does increase the binary size somewhat
Low: cts: Drop the message about restarting logs to debug, we'll produce an error if we cant
Medium: crmd: Use an operation's interval when an invalid timeout was specified
Low: cts: Use the master's IP where possible to avoid name lookup failures
High: crmd: Use the correct define/size for lrm resource IDs
Low: Tools: Ensure the line we're grabbling the time from contains a time
Low: cts: Log CTS output to syslog only if an output file wasn't specified
Low: cts: Always supply a timeout for lrmadmin operations
Low: crmd: Reduce a harmless error to a warning
Low: crmd: Provide more information about active processes in log_member_uname()
High: Tools: crm_simulate - Fix use-after-free in when terminating
Low: cts: Update the connectivity data a little more often and warn when its zero
Low: Build: Update version number to 1.1.3
Low: Build: Update changelog
Low: Build: Update changelog
High: mcp: Use 99/01 for the start/stop sequence to avoid problems with services (such as libvirtd) started by init - Patch from Vladislav Bogdanov
High: PE: Fix typo leading to use-of-NULL in the new ordering code
Low: doc: Import an Italian translation of Clusters_from_Scratch by rasca at miamammausalinux.org
Low: doc: Add the site index file for documentation
Low: doc: Give credit to Raoul Scarazzini for the Italian translation
Low: doc: Build the desktop html version by default
High: crmd: Don't hide stop events that time out - allowing faster recovery in the presence of overloaded hosts
Low: Build: EPEL5 only has the old form of the initddir macro
Low: cts: Ensure the Reattach test always renables resource management
Low: cts: Pause if test teardown fails
Low: Core: Hook up the xml logging functions to the new trace code
High: PE: Use hashtables instead of linked lists for performance sensitive datastructures
Low: PE: Remove unnecessary call back into create_actions() for complex resource types
Medium: PE: Skip probe_resources() completely if probing is disabled
High: PE: Free memory in recently introduced helper function
High: PE: Obtain massive speedup by prepending to the list of ordering constraints (which can grow quite large)
Medium: PE: Detect start/stop actions with intervals
High: PE: Fix memory leak introduced as part of the conversion to GHashTables
High: PE: Optimize the logic for deciding which non-grouped anonymous clone instances to probe for
Medium: Tools: crm_node - Fix the return code for operations on corosync clusters
High: Stonith: Fix two memory leaks
Low: PE: Update comment
Low: PE: Fix comment
High: PE: lf#2478 - Implement improved handling and recovery of atomic resource migrations
Medium: PE: Use g_list_prepend() in favor of g_list_append() for performance reasons where possible
Medium: Core: Fix compilation when both heartbeat and corosync are supported
Medium: Core: Fix compilation when both heartbeat and corosync are supported
Low: doc: Fix compilation of Clusters_from_Scratch
Low: PE: Sort the regression test scores
High: PE: Fix memory leak when processing completed migration actions
Medium: crmd: Reduce code duplication by reusing create_operation_update() when updating the CIB with pending operations
Low: Core: Remove duplicate header import
Low: mcp: Add a cast to keep OSX happy
Low: Build: Add a bunch of requires to the devel subpackage
Low: PE: Drop a debug message to trace
Low: Core: Add support for gprof
Low: Core: Fix the previous patch adding gprof support
Low: Build: Disable various compiler optimizations when profiling
Low: Core: Minor optimization for crm_element_name
Medium: cib: Speed up an often used lookup function by using a hashtable
Medium: cib: Optimize cib_notify_client() for speed, profiling indicated it is a hotspot
Low: Core: Make trace logging reachable with -VVVVV
Medium: cib: Optimize a number of hotspots indicated by profiling
Medium: cib: Another minor optimization
Low: Core: Hook asserts up to the trace framework
High: Core: Fix the CRM_CHECK macro
Low: Build: Default trace data to off
High: Core: Fix the CRM_CHECK macro
Low: Build: Try valgrind for profiling instead
High: Core: Cache Relax-NG schemas - profiling indicates many cycles are wasted needlessly re-parsing them
High: mcp: Ensure valgrind is able to log its output somewhere
Low: Core: Log whenever the RelaxNG schema is re-parsed
Low: mcp: Allow callgrind to be enabled dynamically
Low: Build: Rename --with-trace-data to --with-tracedata so the bcond rpm macros function as expected
Low: mcp: Fix the valgrind logic
Low: mcp: kcachegrind wants input files to have a callgrind.out prefix
High: Core: Correctly hook up the RNG schema cache
High: Core: Don't compress large payload elements unless we're approaching message limits
High: cib: Optimize the way we detect group/set ordering changes - 15% speedup
Low: Core: There shouldn't be any diff markers in the CIB left to filter
Medium: ais: Support the use of callgrind and use the same PCMK_(val|call)grind_enabled variables as the mcp
High: Core: Prevent use-of-NULL when cleaning up RelaxNG data structures
Low: Build: Don't warn about GNU-isms in make
High: cib: Repair the processing of updates sent from peer nodes
High: Core: Actually removing diff markers is necessary
High: Core: Provide significant performance improvements by implementing versioned diffs and digests
Low: PE: Bump the feature set in the regression test outputs
High: cib: Include version information on our synthesised diffs when nothing changed
Low: Core: Fix the format of xml trace output
Low: Core: Add some trace logging to help diagnose a problem when applying diffs
Medium: cib: Reject updates with a feature set greater than our own
High: Core: Modify the v2 digest algorithm so that some fields are sorted
Medium: Core: Some performance enhancements when comparing xml objects
Low: Core: Preparation for the new diff format
High: Core: Correctly use lazy_xml_sort() for v2 digests
Low: mcp: Improve init script message
High: crmd: Use the supplied timeout for stop actions
Low: cib: Avoid slightly more sorting for v2 digests
High: Core: Enable the improve v2 digests
High: cib: Ensure non-change diffs can preserve the ordering on the other side
High: Core: Ensure ordering is preserved when applying diffs
High: Core: Ensure ordering is preserved when applying diffs
High: crmd: There's no need to check the cib version from the 'added' part of diff updates
Low: cib: Update a comment
High: cib: Prevent false detection of config updates with the new diff format
Low: crmd: Additional logging for the cause of new transitions
High: cib: Drop invalid/failed connections from the client hashtable
High: Core: Don't insert empty ID tags when applying diffs
High: PE: Use the original digest algorithm for parameter lists
Low: mcp: Minor tweak to our init script output
Low: Core: Proposed (disabled) enhancement to the v2 digest algorithm
High: cib: Ensure all diffs sent to peers have sufficient ordering information
Low: crmd: Consolidate 'did a config change just happen' logic
Low: cts: Watch for digests that don't apply cleanly
High: cib: Reduce unnecessary copying when comparing xml objects
Low: Core: Drop experimental digest code
Low: Core: Purge diff markers before calculating and comparing digests
Low: Core: Minor speedup when comparing xml objects
Low: Core: Development note
High: cib: The feature set version check is only valid if the current value is non-NULL
Medium: Tools: crm_resource - Ensure xml-related memory is cleaned up at exit
Medium: Tools: crm_simulate - Ensure xml-related memory is cleaned up at exit
High: cib: Fix the feature set check
Medium: cib: Do not calculate changes for status updates that wont be broadcast
Medium: cib: Speed up the establishment of remote connections
Medium: cib: Speed up the establishment of remote connections
High: crmd: All pending operations should be recorded, even recurring ones with high start delays
Low: Core: Update the compiler hint for the do_crm_log_xml macro - logging xml is actually 'unlikely'
Low: Tools: crm_resource - Improved output when listing constraints relevant to a given resource
Low: Tools: crm_resource - Fix -x, the short form of --xml-file
Low: Tools: crm_resource - Include location constraints in -A|-a output
High: Tools: crm_shadow - Avoid replacing the entire configuration (including status)
Low: PE: Fix the dump_node_scores() macro
High: PE: Prevent clones from being stopped because resources colocated with them cannot be active
Low: PE: Prevent clones from being stopped because resources colocated with them cannot be active (regression test)
Low: Build: Useless commit to poke buildbot
Low: Tools: Reuse a common function for freeing lrm operations
High: crmd: Include the correct transition details for timed out operations
High: Core: Bug lf#2506 - Drop the compression limit because Heartbeat's IPC code sucks
High: cib: Bug lf#2506 - Don't remove client's when notifications fail, they might just be too big
High: Core: Prevent use-after-free when creating a CIB update for a timed out action
High: Core: Correctly compare against crm_log_level in the logging macros
High: cib: Revert part of a recent commit that purged still valid connections
Medium: Core: Preserve ordering when removing subtrees with a diff
Low: Tools: Calculate the digest after applying a diff
High: Core: Correctly extract the version details from a diff
Low: cib: Drop the severity of a log message
High: crmd: Don't abort transitions when probes are completed on a node
High: PE: Try to ensure atomic migration ops occur within a single transition
Low: cts: Keep an eye out for TRACE logging too
Low: crmd: Determin the target rc for an action once
High: crmd: Prevent use of NULL by making copies of the operation's hash table
Low: Core: Use the function name for the 'format' trace field
Low: cib: Add trace logging
Low: cts: Ignore the trace logging setup logs
Low: Core: More verbose logging for trace initialization
Low: Core: Turn off unneeded checks when applying a diff
Medium: crmd: Allow operations in the PE state
High: crmd: Ensure the CIB is always writable on the DC by removing a timing hole
Low: cib: Deprecate the set_slave_all() API call
High: PE: Ensure fencing of the DC preceeds the STONITH_DONE operation
Medium: PE: Complete all fencing before attempting atomic migrations
Low: PE: Complete all fencing before attempting atomic migrations (regression test)
Low: Core: Tweak log message
Hg: Merge with 1.1 now that everything appears to be back to normal after the performance work
Low: PE: Avoid needless string compares
Low: cib: Fix a log message
Medium: Core: Provide arguments to do_crm_log_alias() in the correct order when logging xml
Low: Core: Include a missing space in the do_crm_log_alias() output
Low: PE: Remove dead code
Low: Build: Add separate changelog file
Low: Build: New 1.1 release
Low: Remove usage of the slist_iter macro to obtain better information from valgrind/callgrind/gcov
Low: Remove usage of the xml_child_iter macro to obtain better information from valgrind/callgrind/gcov
Low: Remove dead code
High: Repair previous commit removing usage of the xml_child_iter macro
High: Build: Fix handling of conditional features in spec file
Low: Build: Script for determining test coverage
Low: Build: Fix a typo in the coverage script
Low: Build: Fix another typo in the coverage script
High: Core: Prevent recently introduced use-after-free in replace_xml_child()
High: Core: Fix previous commit to replace_xml_child()
High: crmd: Update failcount for failed promote and demote operations
Medium: Core: Update the default validation type to pacemaker-1.0
Low: cts: Correctly track failed SpecialTest1 tests
Low: Build: Keep the Hg version int he spec file up-to-date
High: Core: Reinstate the logic that skips past non-XML_ELEMENT_NODE children
High: Core: Remove extra calls to xmlCleanupParser resulting in use-after-free
High: PE: Prevent use-of-NULL when there is status for an undefined node
Low: Build: Explicitly check for a supported version of glib
Low: crmd: Fix error message regarding feature sets
High: PE: Only update the node's promotion score if the resource is active there
High: PE: Only use the promotion score from the current clone instance
High: PE: Prevet use-after-free resulting from unintended recursion when chosing a node to promote master/slave resources
Low: Build: Fix the use of bcond variables in configure
Low: ACL: Move is_privileged() to a more appropriate location and switch to an inline function
Low: cib: Rework the determine_request_user() function for additional clarity
High: Core: Fix typo in previous commit
High: Core: Move the ACL specific header components into the main ifdef to prevent multiple definitions
Low: RA: Use the correct migration variables in the Dummy agent - Patch from Holger Teutsch
Medium: crmd: Bug lf#2536 - Purge status for resourcs that no longer live in the LRMd
High: Stonith: Teach stonith_admin to automagically obtain the current node attributes for the target from the CIB
Low: RA: o2cb - Fix the return code from kill_daemon()
High: tools: Bug lf#2527 - Prevent use-of-NULL in crm_simulate
Low: Tools: Make unmanaged resources stand out in crm_simulate graphs
High: PE: Bug lf#2508 - Correctly reconstruct the status of anonymous cloned groups
Low: PE: Bug lf#2508 - Correctly reconstruct the status of anonymous cloned groups (regression test)
Low: PE: Minor formatting change to LogActions for consistency
Medium: Core: Bug lf#2511 - Correctly parse the resource name from notificiation action IDs
Low: PE: Don't always regenerate the expected results for bug-lf-2508
Low: crmd: Indicate the interval for the timer that just expired
High: crmd: Reschedule the PE_START action if its not already running when we try to use it
Medium: RA: Overhaul of the HealthSMART RA by Matthew Richardson
Medium: RA: Allow a list of filesystems/devices to be queried for free space - by Matthew Richardson
High: Tools: Prevent crm_resource commands from being lost due to the use of cib_scope_local
High: PE: Prevent use-of-NULL resulting from variable shadowing spotted by Coverity
Low: Update the per-file license text to reflect the project's intentions for libraries
Low: There is no such thing as GPLv2.1+, use GPLv2+
Doc: Use the new logo for generated documentation
Low: Doc: Use the correct name for the logo svg
Low: Doc: Update translation base files for Clusters from Scratch
Low: Doc: Add simplified chinese translation of Clusters from Scratch by Charlie Chen
Low: doc: Update brand installation
Low: doc: Move the new logo to the correct location
Low: doc: Use the raw mercurial hash for identifying document versions on clusterlabs.org
Low: doc: Correct the previous commit for extracting the doc version for clusterlabs.org
Low: Core: Support matching multiple filenames in PCMK_trace_files
Low: doc: Correct typo in crm_mon help
High: PE: Bug lf#2495 - Prevent segfault by validating the contents of ordering sets
Medium: PE: Additional sanity checking of constraint set members
Low: RA: Bug lf#2512 - htmlfile is not mandatory
High: PE: Bug lf#2518 - Avoid spamming the logs with errors for orphan resources
Low: crmd: Some extra logging for shutdown-escalation
High: crmd: Bug lf#2545 - Ensure notify variables are accurate for stop operations
Medium: crmd: Bug lf#2533 - Implement reprobe natively in the crmd to avoid transition loops
Low: doc: Create a man page with common options for stonith resources
Medium: crmd: Bug lf#2309 - Cancel recurring monitor ops at shutdown
High: PE: Bug lf#2493 - Don't imply colocation requirements when applying ordering constraints with clones
Low: PE: Bug lf#2493 - Don't imply colocation requirements when applying ordering constraints with clones (regression test)
Medium: PE: Bug lf#2529 - Protect against constraints that require stonith devices to stop before 'normal' ones
Low: PE: Minor trace logging improvemnt
High: PE: Bug lf#2544 - Prevent unstable clone placement by factoring in the current node's score before all others
Low: PE: Bug lf#2544 - Prevent unstable clone placement by factoring in the current node's score before all others (regression test)
Low: Tools: Suport cibadmin --empty for creating a minimal CIB
Low: Tools: Tweak development output
Low: doc: Make sure people know there is two parts to configuring ocf:pacemaker:ping
Low: PE: Remove duplicate test
High: PE: Bug lf#2445 - Do not override configured clone stickiness values
High: PE: Bug lf#2445 - Avoid relying on stickness for stable clone placement
Low: PE: Bug lf#2445 - Avoid relying on stickness for stable clone placement (regression test)
High: crmd: Cancel recurring operations while we're still connected to the lrmd
Low: crmd: Suppress an ERROR by teaching fsa_action2string() about complex actions
Low: PE: Consolidate logging for resources active on multiple nodes
Low: cts: Make the destination used by ocf:pacemaker:ping configurable
Low: doc: Minor logging tweak
Low: cts: Clarify why we're looking for some patterns
High: ais: Bug lf#2550 - Report failed processes immediately
Medium: Core: Do not store last-run for recurring operations, since it is misleading
Medium: PE: Re-organise the logic for determining what considitons an action requires in order to be performed
Low: Tools: crm_simulate - Remove redundant whitespace and update the regression tests
Medium: PE: More accurate logging of restarts and recovery for master/slave resources
Low: PE: Minor improvements to trace logging
High: PE: Bug lf#2554 - target-role alone is not sufficient to promote resources
Low: doc: Fix auto-rebuild of outdated documentation
Low: doc: Update the CfS with details on using the Master Control Process
Low: Build: More tweaks to the docbook related Makefile targets
Low: doc: Include an Appendix on configuring CMAN
Low: doc: Indicate that disabling stonith is incredibly bad for productions systems and update the software version numbers
Low: doc: Describe what to do if the Pacemaker version of dlm_controld is not found and update the software version numbers
Low: doc: Update the software version numbers
Low: doc: Add user-friendly id for the corosync section
Low: doc: No need to specify votes for CMAN clusters
Low: Build: Use versioned tarballs and force the package to report the spec's version
High: PE: Bug lf#2558 - Suppress unnecssary assertion failure
High: Build: Fix srpm generation for EPEL and SuSE based distros
Low: Build: Update changelogs
Hg: Close dangling referrence to the old stable-0.6 branch
Hg: Close dangling referrence to the old profiling branch
Low: Build: Cleanups and make it easier to mock build packages for multiple distros
High: Build: Remove stray newline
Low: doc: Fix the quorum equation
Low: doc: Fix minor typo
Low: Build: Use the older _initrddir macro to satisfy EPEL-5
Low: doc: Minor corrections to Clusters from Scratch from Michael Prokop
Low: Build: Rearrange the spec file customization logic
Hg: Merge with 1.1
Medium: tools: crm_report - Add cman support
Medium: tools: crm_report - Do not overwrite existing outputs
Low: tools: crm_report - Indicate the correct option for specifying custom log files
Low: tools: crm_report - Do not overwrite existing outputs
Low: tools: crm_report - Only log the CTS control file if it was calculated
Low: tools: crm_report - Extract corosync blackbox data
Low: tools: crm_report - Add cman shutdown log pattern
Low: attrd: Reduce the default amount of logging
High: Build: We need asciidoc in order to build rpms
Medium: tools: crm_report - Save a copy of any core files for later use
Hg: Merge with 1.1
Low: tools: crm_report - Indicate the stack type when requesting the current partition
Medium: Core: Improved determination of the correct stack for cli tools to connect to
Low: Build: Minor cleanup
Low: Core: Reduce code duplication
Medium: Tools: attrd - Do not initiate updates for deleted attributes
High: Tools: Bug lf#2528 - Make progress when attrd_updater is called repeatedly within the dampen interval but with the same value
Low: Tools: attrd - Minor logging changes now that LOG_NOTICE is the default threshold for attrd
Low: Tools: crm_report - Slightly improved help handling
Low: Core: Ensure we log cluster connections at a sufficiently high priority
Low: tools: Bug lf#2556 - Do not pass bash specific switches to other SHELLs
High: PE: Bug lf#2575 - Use uname for migration variables, id is a UUID on heartbeat
Low: doc: Fix typo
Low: doc: Update copyright
High: crmd: Bug lf#2509 - Watch for config option changes from the CIB even if we're not the DC
High: PE: Bug lf#2574 - Prevent shuffling by choosing the correct clone instance to stop
Low: PE: Bug lf#2574 - Prevent shuffling by choosing the correct clone instance to stop (regression test)
Low: tools: cibadmin - Support --blank as an alternative to --empty
Low: Tools: crm_simulate - No need to display Transition Summary in verbose mode
High: PE: Only allocate master/slave resources once
Low: doc: Build PNG images from their SVG sources
Low: doc: Remove generated images from the repository
Low: doc: Purge unused makefile variables
High: crmd: Bug lf#2559 - Fail actions that were scheduled for a failed/fenced node
Low: PE: Time to purge the 0.6 regression tests
Low: doc: Correct misplaced closing bracket
Low: doc: Complete partial sentance
Low: doc: Restore reference to IPMI as a common example of an unreliable fencing device
Low: cts: Drop support for 0.6 based configs
Low: attrd: Indicate when attrd clients are updating fields
Low: PE: Slightly improved debug logging
Low: Build: Avoid unnecessary rebuilds of docbook based documentation
Low: PE: Avoid data structure naming conflict with the transition library
Low: PE: Slightly improved debug logging
High: PE: Bug lf#2581 - Avoid group restart when clone (re)starts on an unrelated node
Low: PE: Bug lf#2581 - Avoid group restart when clone (re)starts on an unrelated node (regression test)
High: cts: Fix dangling elif preventing python compilation
Low: tools: Allow regression tests to run from the source tree
Low: doc: Fix typo
Medium: cib: Downgrade development logging from ERROR
Medium: PE: Set reasonable limits for the number of input files saved
Low: doc: Fix spelling error
Medium: agents: ping - Support timeouts with units
Low: agents: ping - Support 'msec' as well as 'ms' for timeouts
Low: tools: Bug lf#2563 - Do not recursively look for resource meta attributes
Low: doc: Update stale linux-ha URL
Low: crmd: Note to self
Low: PE: Bug lf#2582 - Do not write out duplicated input files
Medium: mcp: Ensure HA_logfacility is set for children to inherit
Medium: cib: Bug lf#2229 - Use SIGPIPE for re-enabling disk writes after an error
High: Build: Correctly generate some special tarballs
Low: Fencing: Clean up the metadata creation
Medium: Stonith: Bug lf#2557 - Allow the fencing agent operations to be overridden on a per-device basis
Medium: Stonith: Bug lf#2557 - Remove redundant logic now that fencing agent operation remapping is handled in a central location
Low: Shell: Update the regression tests for the new stonithd metadata
Medium: Tools: Bug lf#2477 - crm_resource - Indicate master/slave state when querying the parent resource
Medium: Tools: Bug lf#2549 - Display the true number of configured resources in crm_mon
Low: Stonith: Remove unused #defines
Low: doc: Fix publishing of generated docs to the website
Low: Build: Extra items to remove on make clean
Low: doc: Include the block that should be used instead of slist_iter()
Medium: RA: Fix handling of 'devices' in the HealthSMART agent (Patch from Keisuke MORI)
Low: doc: No longer say that Pacemaker doesn't use the OCF reload action (Patch from Holger Teutsch)
High: ais: Followup to previous patch - Ensure we drain the corosync queue of messages when Glib tells us there is input
High: crmd: Catch fence operations that claim to succeed but did not really
High: ais: Correctly disconnect from Corosync and Cman based clusters
High: cman: Correctly override existing fenced operations
High: cman: We need to run the crmd as root for CMAN so that we can ACK fencing operations
Low: crmd: Drop the default deadtime to 20s
Medium: fencing: Fix builds when MANPAGE_XSLT is not found
Medium: ais: Bug lf#2587 - Prevent memory leak in error path
High: crmd: Bug lf#2528 - Introduce a slight delay when creating a transition to allow attrd time to perform its updates
Low: cts: Avoid using timers to detect when the cluster reacts to standby settings
Low: Core: Fix output format of logged xml
Low: Stonith: Improve logging usefulness
Re-map unknown errors to st_err_generic
High: mcp: Enable stack detection logic instead of forcing 'corosync'
High: PE: Restore the ability to save inputs to disk
High: Stonith: Ensure device parameters are passed to the daemon during registration
Medium: Stonith: Correctly disable port list queries upon the first failure
High: Stonith: Correctly finish sending agent options if the initial write is interrupted
Medium: Stonith: Remove node attributes from the server side since they will always be unset
Low: Build: Enable support for trace logging by default
Low: Build: Clean up the way package versions are calculated
Medium: cib: Make a minimal effort to sanitize generated IDs for node attributes
Low: Build: Allow compilation when /tmp is mounted with noexec
Low: cts: Remove a lot of outdated information from the CTS README
High: crmd: Store only the last and last failed operation in the CIB
Medium: Include: Remove arch specific values from public header to prevent multi-arch issues
Low: Core: Add some basic pkg-config control files
Low: Build: Target for coverity scans
Low: Build: Fix up make clean
Low: tools: Clean up some issues reported by coverity
High: stonith: Coverity - Remove badly constructed memory allocation from the query API call
Medium: tools: Coverity - Correctly hook up the crm_shadow --all option
Low: PE: Coverity - Prevent possible use-of-NULL
Low: Core: Coverity - Fix various issues reported by coverity
High: stonith: Coverity - Correctly construct result list for the query API call
Medium: PE: Coverity - fix potential use-of-NULL
Low: mcp: Coverity - Fix memory leak
Low: tools: Coverity - Fix memory leak
Low: Build: Allow partial builds when doing coverity runs
Low: tools: Coverity - Fix memory leaks
Low: doc: Correctly document the default for migration-threshold
Low: crmd: Display the name and vlaue of the transient attribute(s) that caused a new transition
Low: Build: Explicitly require libtool-ltdl-devel in the spec file
Low: doc: Undo more recent 'corrections'
High: stonith: Correctly handle synchronous calls
Medium: tools: Bug lf#2614 - crm_report - Fix suppression of passwords
Low: Build: Keep the compiler from complaining about unused variables
Low: Build: Resolve configure and compilation warnings
High: Core: Use dlopen to load heartbeat libraries on-demand
Medium: mcp: Bug rhbz#708722 - Check permissions on the logfile's parent directory
Low: Build: Allow debugging of failed mock builds
Low: Build: LDFLAGS such as --as-needed require strict and ordered library listings
Low: Build: Fix manpage building when libperl.so is in a non-standard location
Low: Build: Partial revert of previous patch. MIB install directory was correct
Low: Build: Allow compilation without a cluster stack
Low: Core: Support compilation against newer versions of GNUTLS
High: Core: Partial revert of accidental commit in previous patch
Revert controversial commit pending proper review
Revert controversial commit pending proper review
Low: Core: Horrible function for dropping the const flag in glib iterators when we know its safe
Low: Build: Use the detected location of help2man
High: PE: Bug lf#2613, lf#2619 - Group migration after failures and non-default utilization policies
Medium: mcp: Add systemd-style init script
Low: Build: Automatically re-run configure if we run autogen.sh and config.log exists
High: Fencing: Improved pcmk_host_map parsing
High: Build: Correctly locate the pacemaker systemd inti script
High: Fencing: Consolidate pcmk_host_map into run_stonith_agent so that it is applied consistently
High: Fencing: Deprecate pcmk_arg_map for the saner pcmk_host_argument
Low: Build: Create a cli subpackage
Low: Build: Default systemd support to off
High: Tools: crm_report - Correctly extract data from the local node
Low: Tools: crm_report - Suppress error message if the stonith binary is not available
Medium: Tools: crm_report - strip leading directory paths from tarball contents
High: Tools: crm_report - Repair the ability to extract data from the local machine
Low: shell: Update regression tests for new stonith metadata
High: mcp: dirname() modifies the input path - pass in a copy of the logfile path
High: Tools: crm_report - Remove newlines when detecting the node list
High: pengine: Correctly determine the state of multi-state resources with a partial operation history
Low: pengine: Correctly determine the state of multi-state resources with a partial operation history (regression test)
Low: cts: Watch for unexpected resource recovery
Low: Build: Prevent GREP_OPTIONS from interefering when rerunning configure
Low: PE: Make custom_action_order() slightly smarter
High: Core: Cancelled and pending operations do not count as failed
High: PE: Ensure restarts due to definition changes cause the start action to be re-issued not probes
Low: PE: Ensure restarts due to definition changes cause the start action to be re-issued not probes (regression test)
High: Fencing: Prevent use-of-NULL by g_hash_table_lookup
Low: Build: libtool-ltdl-devel doesn't exist on SUSE
High: crmd: Prevent use-of-NULL by g_hash_table_iter_next()
Low: Build: More reliable package building when systemd is available
High: Tools: crm_report - Report on all detected backtraces
Low: cts: Suppress recovery messages in the ResourceRecover test
High: mcp: Shut down if corosync becomes unavailable
Low: Build: Fix library package name on suse
Low: Build: Fix library package name on suse - pt2
Low: Tools: crm_report - Minor language improvment
Low: Build: Fix automated building in some timezones
High: mcp: systemd control file is now functional
Low: cts: Use StopTest in PartialStart for improve debugging
Low: cts: Wait for all nodes to be up in before testing for cluster stability
Low: cts: Tweak some startup logic
Low: cts: Detect values for at-boot and syslogd automatically
High: mcp: Fix spelling mistake in systemd service script that prevents shutdown
Low: cts: Attempt to determine a good starting point for test IPaddr2 resources
Medium: fencing: Add support for --reboot to stonith_admin
High: cts: Teach cts how to cope with nodes that are fenced due to quorum changes resulting from startup-fencing=True
Medium: Fencing: Downgrade an error message for a valid scenario
Medium: tools: crm_report - provide feedback for activities that may take a long time
High: crmd: Ensure we do not attempt to perform action on failed nodes
Low: Fencing: admin - Disable stderr logging by default
Medium: Fencing: Update the pcmk_host_map example to reflect reality
High: Fencing: Fix use-of-NULL by g_hash_table_lookup
Medium: Tools: Bug lf#2621 - Correctly log corosync node IDs
Medium: extra: Incorporate fping functionality from Dan Urist
High: crmd: Recurring actions shouldn't cause the last non-recurring action to be forgotten
High: crmd: Do not wait for actions that were pending on dead nodes
High: crmd: Cancel timers for actions that were pending on dead nodes
Medium: Extras: ping - use a fixed backoff of 1.0 to avoid pushing complex timeout calculations to the user
High: Fencing: Prevent use-of-NULL when no Linux-HA stonith agents are present
High: PE: Correctly recognise which recurring operations are currently active
Low: cts: Downgrade a warning that occurs naturally when startup-fencing=True
High: PE: Ensure role is preserved for unmanaged resources
Low: PE: Ensure role is preserved for unmanaged resources (regression test)
High: PE: Ensure unmanaged resources have the correct role set so the correct monitor operation is chosen
Low: PE: Ensure unmanaged resources have the correct role set so the correct monitor operation is chosen (regression test)
Low: shell: Update regression tests for new stonith metadata
Low: crmd: Dev note to my future self
Medium: doc: Refresh the stonith section of Pacemaker Explained
Low: tools: crm_simulate - Send logging to stdout when additional detail is requested so that grep works as intended
High: PE: Bug suse#707150 - Prevent services being active if dependancies on clones are not satisfied
Low: PE: Bug suse#707150 - regression test
Medium: PE: Prevent resource shuffling when the 'utilization' placement strategy is used
Low: PE: Prevent resource shuffling when the 'utilization' placement strategy is used (regression test)
Low: PE: Minor code cleanup
Hg: Merge with upstream
Low: Build: Get publican to stop checking our SRPMs
High: PE: Partial revert of 'Minor code cleanup CS: cf6bca32376c On: 2011-08-15'
Low: PE: Move get_action_flags() to a more appropriate location given its lack of general usefulness
Medium: stonith: Clean up the stonith_key_value_t API elements
Medium: stonith: Add an API call for listing installed agents
Low: doc: Indicate the correct command for listing installed stonith agents
Low: stonith: Correctly indicate the number of installed devices
High: stonith: Fix use-of-NULL in stonith_api_device_list()
Medium: stonith: Allow the fencing history to be queried
Low: stonith: Remove test code
High: stonith: Ensure completed operations are recorded as such in the history
Low: stonith: admin - Print how many matching devices were found after the contents of the list for --list*
Medium: stonith: Support --quiet to display just the seconds since epoch at which a node was last shot
Low: Core: Generate the short-form option string automatically to ensure it is in sync with the long-form
High: stonith: stonith_admin - Prevent use of uninitialized pointer by --history command
High: Core: Ensure there is sufficient space for EOS when building short-form option strings
Hg: Merge devel into 1.1
Low: cts: Prevent use of None in SimulStartLite
Low: Build: Bump the version number
Low: Build: Repair the --with gcov specific blocks
Medium: stonith: Serialize actions for a given device
Low: stonith: Fix stand-alone mode
Hg: Merge with ygao
Low: cts: Detect fencing for 2-node clusters
Low: cts: Cloning fencing agents is no longer best-practice
High: stonith: Add missing entries to stonith_error2string()
High: Core: Fix variable expansion in pkg-config files
Medium: Build: Fix the spec file for opensuse
Low: doc: Only package en-US documentation by default
Low: extra: Helper that works around wildcard expansion issues in buildbot
Low: extra: Add an extra case to the buildbot helper
Low: Build: Resolve and/or ignore a number of rpmlint issues
Low: Build: Spec file tweaks to make it valid for opensuse
Low: PE: Suppress an invalid error caused by a known bug in valgrind
Low: tools: Suppress an invalid error caused by a known bug in valgrind
High: PE: Resolve memory leak reported by valgrind
Hg: Automated merge
High: PE: Fix memory leak for re-allocated resources reported by valgrind
Low: Build: Ensure the SRPM gets installed for gcov in buildbot
Low: Build: Keep trying to get gcov tests working
Low: Build: Keep trying to get gcov tests working
Low: If/when we run indent, these directives will preserve the existing formatting of some code blocks
Low: More prep-work for consistent code formatting
Low: cib: Remove the remaining uses of the xml_child_iter() macro
Low: Tweak some indent options
Low: PE: Remove dead code
High: crmd: Bug lf#2584 - Allow nodes to fence themselves if they're the last one standing
Medium: Build: Force a versioned dependancy on the libs and cli subpackages
Low: cts: Ignore some additional BadNews items
Low: doc: Allow publishing of multiple sets of annotated source code
Medium: Resolve a number of coverity use-of-NULL defects
Medium: Resolve a number of coverity leak defects
Low: More coverity appeasement
High: PE: Demote from Master does not clear previous errors
Low: PE: Demote from Master does not clear previous errors (regression test)
Medium: cib: Attempt a graceful sign-off for remote TLS connections
High: cib: Remove disconnected remote connections from mainloop
High: cib: Call gnutls_bye() and shutdown() when disconnecting from remote TLS connections
Medium: Core: Treat GNUTLS_E_UNEXPECTED_PACKET_LENGTH as normal termination of a TLS session
Low: cib: Development note
High: crmd: Bug lf#2632 - Correctly handle nodes that return faster than stonith
Low: cts: Prevent qarsh from hanging permanently
Low: Build: Update helper targets
Medium: crmd: Log duplicate DC detection as a WARNING not ERROR
Medium: crmd: Prevent secondary DC fencing resulting from CIB updates that are lost due to elections
Low: cts: Suppress BadNews for renamed clones in ResourceRecover
Low: doc: crmd, pengine and stonith option manpages should be section 7
Low: tools: crm_report - Include a description for the help text
Low: doc: Push manpages to the website
Low: PE: Regression test filenames dont need to be too detailed
Low: doc: Update changelog and spec file for 1.1.6
Low: mcp: Re-format with indent
High: Build: Fix srpm generation
Low: Build: Fix scratch builds
Medium: Build: Use git for the build hash
Medium: Build: Fix processing of git archive strings
Low: cib: Reformat with indent for consistency
Low: crmd: Reformat with indent for consistency
Low: indent labels at column 2
Low: tools: Reformat with indent for consistency
Low: PE: Reformat with indent for consistency
Low: Reformat with indent for consistency
Low: Build: Update indent settings
Low: Teach git which files to ignore
Low: PE: Remove unused artifacts from the regression tests
High: cib: Fix compilation of ACL code (via ygao)
Low: Tools: Tweak man pages
Low: Build: Filter some unhelpful automatic configure arguments
Low: Build: Tweak the build order for the www helper
Low: Core: Remove use of the xml_prop_iter macros - they confuse parsers, analyisers and indent tools
Low: Reformat with indent for consistency
Low: Build: Simplify the indent target by revert the 1 file that indent still mangles
Low: doc: Fix uploading of generated docs
Low: Git: Ignore build results
Low: cts: Remove remote logging setup from cluster_test as it is no longer needed
Low: cts: always use the short node name when performing log audits
Low: cts: Add more stonith presets
Low: cts: Prevent use of undefined variable in ResourceRecover
Low: extra: Three helper utilities for setting up clusters
High: extra: cluster-helper - Prevent funky interaction with calls to read
Low: extra: cluster-init - add option to accept defaults
Medium: tools: crm_report - Only perform host detection on nodes we're extracting data from
Medium: mcp: Additional error logging if we can't connect to corosync
High: tools: Avoid problems caused by differences between node name and uname -n
Medium: tools: crm_report - Give syslog time to log our discovery message when detecting log files
Medium: tools: crm_report - Capture rpm --verify error text
Medium: Extra: cluster-helper - Remove stray character in option processing
Low: cts: Lines containing only whitespace are not valid nodes
Low: Build: Split out pacemaker-cluster-libs rpm package from pacemaker-libs
Low: Build: Improve the description text for subpackages
Low: Build: Move cluster-aware tools to the main package to avoid -cli depending on -cluster-libs
Low: Build: Rename the 'ais' bcond variable to 'corosync'
Low: Build: Only force compilation with -O0 for profiling builds
Low: Build: Drop unhelpful warning
Low: Fencing: Stop linking against unused libraries
Low: Build: Greatly simplify the bootstrap process
High: controld: Convert crm_terminate_member...() to use stonith-ng directly instead of attrd->PE->crmd
High: Build: Move libcrmcluster to its own directory to avoid build loops
Medium: Extra: Import the ccs2cib conversion tool for those coming from rgmanager
Low: extra: Move the rgmanager conversion tool to its new home after import
Low: Extra: Re-indent rgmanager conversion tool to conform to project standards
Low: Build: ccs2cib is a script, saying otherwise confuses 'make tags'
Medium: fencing: Helpers for initiating fencing that avoid the need for install-time dependancies
High: fencing: Implement the server-side of the st_opt_cs_nodeid flag
Low: extra: Move conversion tool readme to proper location
Medium: Fencing: Simply return 0 (aka. never) if stonith_api_time_cs_helper() fails
High: fencing: Support checking in-progress operations for all nodes
High: Fencing: Support admin configurable timeouts for API operations
High: New fencing agent for cman based clusters
Low: extra: Teach cluster-init how to configure cman to use Pacemaker fencing
High: crmd: Do not block when trying to notify fenced of a successful fencing operation
Low: Build: Update the spec file
Low: extra: Support specifying the set of nodes to operate on as an environment variable in cluster helpers
Medium: doc: Update the GFS2 section to use CMAN
Medium: doc: Refresh the stonith section of CfS
Medium: doc: Remove non-unicode quotes from CfS
Low: Build: Remove execute permissions from header file
Low: extra: Fix FSF address in rgmanager conversion tool
Low: Build: Tabs->spaces in spec file
Low: Build: Use a tarball name/prefix consistent with what github produces
Low: extra: Fix FSF address in rgmanager conversion tool
Low: doc: Tidy up some cman related text
Low: doc: Add images lost in the conversion to git
Low: Build: Remove execute permissions from headers
Low: Build: Move man pages to the correct subpackage
Low: PE: Move master based on failure of colocated group (regression test)
Low: PE: Move master based on failure of colocated group
Low: doc: A README for github and casual observers
High: mcp: Set the HA_quorum_type env variable consistently with our corosync plugin
High: Stonith: Implement a manual override for when nodes are known to be safely off
Medium: PE: Resolve memory leak when retrieving failure counts
High: Stonith: Ensure all nodes recieve and deliver notifications of the manual override
High: stonith: Fix compile error
Medium: Build: Older versions of autofoo don't like the use of PKG_ as a variable prefix
Low: Build: Skip schema validation when building man pages with a2x for RHEL5
Low: Build: Remove stale autofoo file left over from the rgmanager conversion tool import
High: stonith: Bug rhbz#745526 - Ensure stonith_admin actually gets called by fence_pcmk
Merge pull request #9 from dfcluster/dfcluster
Merge pull request #10 from gao-yan/stonith-metadata
Merge pull request #8 from gao-yan/stonith-timeout
Merge pull request #7 from gao-yan/ra-fixing
Merge pull request #5 from gao-yan/meta-env
Low: Tools: crm_report - highlight the need for quotes when specifying date and time
Low: Build: The GNUtls library package was renamed in opensuse-11.0
Low: Build: Allow just the core pieces to be installed
Low: Extra: cluster-helper - specify a username when using pdsh
Low: Extra: cluster-init - Fix many issues preventing setup of a sane cluster
Medium: PE: cl#5002 - increment_clone() does not terminate on invalid input
Medium: Build: Install pacemaker daemons to their own directory
High: Fencing: fence_pcmk - don't block if nothing is specified on stdin
Low: cts: Allow CTS to use a nodes listed in DSH group
Low: cts: Clean up the help text
Low: cts: Default to putting logs in ~/ for non-root users
Low: Tools: Remove strange quote characters from help text
Low: Build: Tool for checking the ABI compatability of two releases
Low: Build: Support ABI checks
Low: PE: Restore log message to its correct severity
Low: Build: Leave a note how to make other build commands run silently
High: PE: Correct return from error path
Medium: Build: Fix srpm generation when no build counter is present
Merge pull request #11 from yuusuke/iidayuus
Low: Build: Update paths in spec file
Low: Build: Fix generation of packages on openSUSE
Medium: PE: Indicate the difference between resources that are restarting and reloading
High: crmd: Leave it up to the PE to decide which ops can/cannot be reload
High: PE: Detect reloads of previously migrated resources
High: Tools: Bug cl#5003 - Prevent use-after-free in crm_simlate
Low: cib: Bug cl#5018 - Check the return status of scanf to satisfy -Werror=unused-result
Medium: Tools: Bug cf#5003 - Specifying an XML output file for crm_simulate shouldn't imply a PE simulation
Medium: Tools: Bug cl#5005 - Prevent glib error that resulted in duplicate crm_simulate output when redirceted to a file
Low: Build: Remove buildroot after ABI dump completes
Low: crmd: Remove unnecessary use of the CRM_CHECK macro
Low: Core: New helper for extracting integers from const xmlNode pointers
Low: PE: Re-arrange sort_op_by_callid() for clarity
High: Core: Bug cl#5016 - Prevent failures in recurring ops from being lost
High: crmd: Bug cl#5015 - Remove the failed operation as well as the resulting fail-count and last-failure attributes
High: Fencing: Supress errors when listing all registered devices
Low: Fencing: Earlier initialization of the device list
Merge pull request #12 from gao-yan/abi-compliance
Merge pull request #13 from gao-yan/stonith-metadata
Low: ais: Glib aborts compilation unless you include /only/ the main header file
Low: PE: Update the regression tests for the bump of CRM_FEATURE_SET
High: PE: Reload of a resource no longer causes a restart of dependant resources
Low: Build: Don't install scratch binary
Low: Build: Enable silent rules by default
Low: PE: Provide an easy way to get overall profile data for the PE
Low: Build: Don't complain when building on OSX
Low: Build: Clean up .dSYM files when extracting header defines
Medium: Make better use of libqb logging
Low: Build: List libqb as a feature
High: Core: Correctly initialize the log level when using libqb
Git: Merge in libqb logging support
Merge branch 'master' of github.com:ClusterLabs/pacemaker
Medium: Core: Provide header file for basename(3)
High: Core: Fix compilation when libqb is not available
Low: Build: Enable libqb support on Fedora
High: Core: Remove duplicate definitions of CRM_LOG_ASSERT() and CRM_CHECK()
High: Core: Fix the last remaining issues when building without libqb
Merge pull request #14 from gao-yan/ptest-xml-text
Merge pull request #15 from gao-yan/cib-new-segfault
Medium: Tools: crm_uuid needs core libraries too
Medium: Core: Fix compilation on distros with -Werror=unused-result
Low: PE: Correctly indicate when resources are being reloaded
High: mcp: Remove unused entry in output format string
High: Core: Rempove stray character from qb_ipc_response_header definition
Medium: Fencing: New API functions for adding and removing topology entries
Low: Fencing: Optimize the order in which we compare API calls
High: Fencing: Implement support for advanced fencing topologies: eg. kdump || (network && disk) || power
Low: Build: Fix the NEXT_VERSION macro
Low: Build: Some versions of mock crash with -q
Low: extra: Remove trace logging from cluster-init
Low: Build: Limit NEXT_RELEASE to 3 digits
Low: extra: cluster-init - Prevent the use of cached repo files when installing from a new yum repo
High: Core: Libqb crashes if 'file' is unset
Medium: Fencing: Add support for fencing levels to the admin tool
Low: Build: Don't build with heartbeat by default, add cman on fedora
High: Fencing: Fix use-of-NULL when listing installed devices
High: Fencing: Prevent duplicate free when listing devices
Medium: Core: Support multiple files separated by commas in PCMK_trace_files
High: Fencing: Prevent uninitialized pointers being passed to free
Medium: mcp: Using Type=forking for the systemd file appears to be broken
Medium: Core: Ensure we set useful libqb format strings
High: Core: Consistently apply trace and other logging directives
Low: Fencing: Two dummy agents for testing
High: Fencing: Correctly remove the fencing levels when requested
High: Fencing: Fix use-of-NULL in process_remote_stonith_query()
Medium: fencing: Append list items instead of prepending them
High: Fencing: Prevent use-after-free, we may need the query result for subsequent operations
High: Fencing: Not all devices might be running locally when creating a fencing level
Low: Fencing: Additional trace logging
Low: Fencing: Don't try to parse empty strings in parse_host_line()
Medium: Core: strchrnul() does not exist on Darwin
High: Core: Fix logging of xml with libqb
Medium: Core: Finer grained enablement of XML logging
High: cib: Don't halt disk writes if the previous digest is missing
Medium: Cluster: Prepare for corosync 2.0
High: Cluster: Hook up the callbacks to the corosync quorum notifications
Medium: Core: Record the stack which led to an assert if tracing was enabled for the caller
Medium: mcp: Implement pacemakerd --shutdown and prevent duplicate daemons
High: Corosync: Correctly extract the node id from membership udpates
Low: Corosync: Debug patch
Low: Core: Tweak our log format when using libqb
Low: crmd: Drop some logging to debug
Low: Replace crm_debug_* macros with crm_trace
Low: PE: De-emphasis logs for resources that are unchanged
High: Fencing: Fix coverity FORWARD_NULL error
Medium: Fencing: Fix coverity MISSING_BREAK defect
High: mcp: Fix coverity FORWARD_NULL defect
High: PE: Fix coverity FORWARD_NULL defect
High: tools: crm_resource - Fix coverity FORWARD_NULL defect
Low: mcp: Optimize the default sysconfig file for systemd based systems
High: cib: Fix coverity RESOURCE_LEAK defect
High: Fencing: Fix coverity RESOURCE_LEAK defect
High: PE: Fix coverity REVERSE_INULL defects
Low: PE: Fix coverity UNUSED_VALUE defect
High: Core: Fix compilation when libqb is not available
High: Build: Fix breakages in configure when corosync is disabled
High: Core: More compile fixes for when libqb is not available
High: Core: More compile fixes for when libqb is not available
High: Fencing: Add missing format parameter
Medium: Fencing: Make the compiler happy by casting pointers to char*
Medium: Fencing: Make the compiler happy by casting pointers to char*
Merge pull request #18 from gao-yan/qb-log
Medium: Core: Simplify the wrapping logic for writing out sequences
High: Core: Ensure signals are handled eventually in the absense of timer sources or IPC messages
High: Corosync: Default to using the nodeid as the node's uuid (instead of uname)
High: corosync: Consistently set the correct uuid with get_node_uuid()
High: Fencing: Standardize on /one/ method for clients to request members be fenced
High: Fencing: Fix segfault caused by passing garbage to dlsym()
Medium: Fencing: Remove our logging from an API call designed to be used externally
Medium: Fencing: crm_system_name likely wont be set by callers of stonith_api_kick()
High: Core: correctly inherit the debug option when using libqb
Low: Convert use of do_crm_log_unlikely to crm_trace
Low: tools: crm_diff - use the relevant version when calculating the xml digest
Low: Core: Better tracing of what xml was used to create a given digest
Low: Extra: Support configuration of clusters using corosync's native quorum
Medium: Core: Don't do any work constructing arguments if the log will be discarded
Medium: Tools: Add the ability to generate one of the on-the-wire md5 digests
Medium: cib: Share the diff massaging code between diff_cib_object() and cib_perform_op()
Medium: Fencing: Only initiate logging for the lrm plugin once
High: Core: Strip text nodes from on disk xml files
High: Tools: Implement crm_node functionality for native corosync
High: corosync: Hook up CPG membership
High: mcp: Do not specify a WorkingDirectory in the systemd unit file
High: Core: Prevent use-of-NULL in crm_update_peer()
Low: PE: Don't log the scheduling of recurring actions so verbosely
Low: crmd: Logging tweaks
Low: extra: Trick systemd into realizing the cluster processes are dead and should be started next time
Low: mcp: Less verbose logging
Low: corosync: Drop developmental logging from the initial libqb port
Low: Fencing: Less verbose logging
Low: More logging macro consolidation
Low: Fencing: Logging tweaks
Low: crmd: Logging tweaks
Low: fencing: Drop start-delay from the agent metadata
Low: cts: 'log' is not specific enough when looking for syslog packages
Low: cts: Reliably parse the highest IP in our network segment
Low: cts: Detect if the cluster uses systemd
Low: cts: Sane names for the corosync variants
Low: cts: Also support the standard syslog LOG_PID format
Low: cts: Clean up after corosync if it's using libqb
Low: cts: libqb uses lowercase tags for all log levels
High: fencing: Fix log format error
High: cman: Allow cpg connections to be established
High: cman: Fix compilation when cman is supported
High: tools: Fix compilation against corosync 1.x
Medium: fencing: Indicate in what way execution of an agent failed
Low: corosync: Logging tweaks
Low: cts: Avoid referencing uppercase log level tags in patterns
Low: fencing: Transparently handle the interruption of waitpid()
High: fencing: More gracefully handle failed 'list' operations for devices that only support a single connection
Low: Build: Turn off cman support now that its gone from fedora
High: corosync: Unset the process list when nodes dissappear on us
Low: crmd: Logging tweaks
Low: cts: Look for the new pattern that indicates stonith reconnect
Low: crmd: Logging tweak
Merge remote-tracking branch 'ygao/set-ref-template'
Medium: tools: crm_report - avoid assuming that all logs in a file share the same format
Medium: Tools: crm_report - Correctly locate configured logfiles
Medium: tools: crm_report - Obtain syslog and configured logfiles where available
Low: logging tweaks
High: corosync: Correctly infer lost members from the quorum API
High: crmd: Election - perform the age comparison only once
High: crmd: Remove rusage differences of less than a second from election calculations
Low: doc: Bug cl#5033 - Fix link to SLES documentation
High: PE: Ensure post-migration stop actions occur before node shutdown
Medium: Build: Don't fail when overwriting daemin compatability symlinks for heartbeat
Low: Logging tweaks
Low: tools: Advertise the cibadmin --empty command
Low: Build: Ignore pengine test files
Medium: crmd: getrusage().tv_usec is not as unstable as first thought, re-instate as an election input
High: fencing: Correctly remove a single fencing level for a target
High: crmd: Look for changes to the stonith topology and push them into stonith-ng
High: fencing: Add the fencing topology section to the 1.1 configuration schema
High: Core: basename() may modify its input, do not pass in a constant
Low: Core: Ensure trace data is sent to stderr if/when it gets enabled
High: corosync: Ensure we catch nodes that leave the membership, even if the ringid doesn't change
High: corosync: Treat CS_ERR_QUEUE_FULL the same as CS_ERR_TRY_AGAIN
Medium: Tools: corosync-objctl was moved to corosync-cmapctl and changed argument names. Support both in crm_report
Medium: mcp: Wait a little while for corosync to finish coming up
Low: mcp: Logging tweaks
High: mcp: Shut down if one of our child processes can/should not be respawned
Low: doc: Generate horrid but useful docbook from asciidoc
Low: doc: Correctly create a non-visible index term
Low: doc: Generate more docbook sections from asciidoc
Low: doc: Munge asciidoc output into book chapters
Low: doc: Generate more docbook sections from asciidoc
Low: doc: Remove stale file
Low: doc: Generate more docbook sections from asciidoc
Low: doc: Center align images
Low: Doc: Finish off the conversion of Clusters from Scratch to asciidoc
Low: doc: asciidoc is now a pre-req for publican docs
Low: Build: Massage the default localstatedir and sysconfdir values
Low: Build: Add a clang validator target
Low: doc: Update revision history
Low: doc: Fix the Logo size to avoid scrollbars
Low: doc: Update the copyright year
Low: doc: Auto-populate the --features and --help text in Ch-Tools.txt
Low: doc: Convert elements of Pacemaker Explained to asciidoc
Merge pull request #20 from gao-yan/template-validation
Low: Tools: crm_report - Record a dump of corosync config and runtime variables
Low: extra: cluster-init - Install the abrt package for tracking and interrogating core dumps
Low: crmd: Avoid the string 'error:' in log formats
High: crmd: Fast-track shutdown if we couldn't request it via attrd
Low: doc: Add P-E credits
Low: doc: Import chapter on utilization and placement
Low: extra: Fix cman initialization
Medium: mcp: Start stonith after the cib now that it connects for fencing topology
Low: Git: Ignore built docbook files
Merge branch 'master' into public
Low: doc: Updates from ygao to the Utilization chapter
Low: doc: Restore necessary asciidoc section title
Low: corosync: Fix some compile warnings in the pacemaker/corosync plugin
Medium: corosync: Bug rhbz#729035 - Build pacemaker/corosync plugin with the correct options
Low: Build: Always build packages verbosely
High: crmd: Supply format arguments in the correct order
High: mcp: Fix compilation against corosync 1.4.x
Low: corosync: Use the relevant enum instead of hard-coding 0 for cluster messages
High: corosync: Use the correct structure when libqb is present but corosync isn't built to use it
Low: corosync: Hide the IPC message header variants behind a typedef
High: corosync: Relax a development assert and gracefully handle the error condition
High: corosync: Follow-up to previous patch, prevent use-of-NULL
High: Corosync: Remove a testing assert
Low: tools: Fix logfile linking even though the old form should have worked too
High: Core: Correctly determine when an XML file should be decompressed
High: corosync: Remove deprecated member of the CFG API
High: fencing: Provide enough data to construct an entry in the node's fencing history
Medium: mcp: Provide additional information for resolving logfile permission issues
High: mcp: corosync_cfg_state_track was removed from the corosync API, luckily we didnt use it for anything
Low: tools: Avoid linking to files that don't exist
Low: Build: Build with cman support by default on fedora < 17
Medium: Tools: Have crm_report autodetect the cluster nodes more often
Merge pull request #24 from davidvossel/stonith_standalone_cfg_2
Medium: fencing: Make the stand-alone config file optional and compile when enabled
High: fencing: Move the CFLAGS modification to where it will work
Merge pull request #26 from davidvossel/build_failure
Low: Build: Fix opensuse packages
Merge pull request #27 from yuusuke/iidayuus
Medium: crmd: Badly configured resources can't be reliably stopped, treat them as not running for logging purposes
Low: doc: Indicate the correct node address
Low: Core: Correctly determine if we should compile xslt related functionality
Low: PE: Indicate the human readable version of agent error codes
Merge pull request #29 from davidvossel/xml_parse_segfault
Merge branch 'master' of github.com:beekhof/pacemaker
Low: Build: Find the corosync_common library if necessary
Low: Build: Make it easier to test compilation on other fedora versions
Low: doc: Refresh translation templates
Low: doc: Fix the macros for refreshing the tranlation templates
Merge pull request #31 from gao-yan/rsc-template
Medium: Doc: Avoid tag duplication by specifying one manually for common titles
Low: fencing: Remove unneeded header
Medium: Doc: Correctly avoid tag duplication by specifying one manually for common titles
High: PE: Bug cl#5038 - Prevent restart of anonymous clones when clone-max decreases
Low: PE: Bug cl#5038 - Prevent restart of anonymous clones when clone-max decreases (regression test)
Merge pull request #34 from davidvossel/xml_mon_stdout
Medium: fencing: Bug rhbz#720218 - Advertise support for start and stop actions to make crm shell happy
Medium: fencing: Bug rhbz#720214 - Advertise that 'port' is not required, for most setups stonith-ng populates it automatically
Low: PE: Implement LOAD_STOPPED filtering in a manner more consistent with the overall design
Low: PE: Make order_actions() and get_pseudo_op() available during unpacking
Medium: tools: Include pseudo and cluster actions in the crm_simulate transition summary
Medium: PE: Bug rhbz#789397 - Failcount and related info should be reset or removed when the resource is deleted
Low: PE: Bug rhbz#789397 - Failcount and related info should be reset or removed when the resource is deleted (regression test)
Merge pull request #35 from nozawat/master
High: fencing: fence_legacy - Drop spurilous host argument from status query
Low: PE: Bug cl#5000 - Additional regression tests for A after (B || C)
High: PE: Bug cl#5000 - Ensure ordering is preserved when depending on partial sets
Low: PE: Indicate when we'd like to start a resource but an ordering constraint prevents it
Low: Core: Allow alternate locations for schema files to be specified for testing
Merge pull request #38 from gao-yan/documentation
Merge pull request #40 from davidvossel/optional-env-file
High: Fencing: fence-legacy - Ensure port is available as an environment variable when calling monitor
High: XML: Revert accidental schema change from b6f8a73 - Low: Core: Allow alternate locations for schema files to be specified for testing
Low: PE: Log the location of schema files
Low: Core: Support tracing on systems that don't have the __verbose section
Low: Core: Ensure the trace logging format stays applied
High: Core: Bug rhbz#800054 - Correctly retrieve heartbeat uuids
Low: Core: Fix memory leak in trace logging
Low: Core: Fix potential use of NULL in trace logging
Low: PE: Improved trace logging
Low: PE: Ensure we test against the current action flags
High: PE: Bug cl#5028 - Unmanaged services should block shutdown unless in maintainence mode
Low: PE: Bug cl#5028 - Unmanaged services should block shutdown unless in maintainence mode (regression tests)
High: PE: Log as loudly as possible when we cannot shut down a cluster node
Low: Build: Avoid expression errors when not building on fedora
Low: fencing: Always include the project header
Merge with upstream
Merge pull request #44 from davidvossel/bug-5032
Merge pull request #41 from gao-yan/simulate-tickets
Merge pull request #43 from davidvossel/bug-5015
Low: Tools: Bug rhbz#801351 - Fix crm_report help text
Low: PE: Fix dumping of node scores to trace logging
Medium: PE: Bug rhbz#799070 - Report resources as active in crm_mon if they are located on an unclean node
Low: doc: Newer versions of asciidoc abort on level1 headings in apendicies
Low: cts: Make migrator slightly sticky (since we have no other location constraints) to avoid relocation during Reattach
Low: cts: Check if the dlm is blocking shutdown
Low: crmd: Fix log message
Low: PE: Reduce the severity of a log message
Low: PE: Fix a trace log
Low: crmd: Additional logging to help track down a bug
Merge pull request #45 from gao-yan/crm_ticket-force
Low: Build: Support cman on rhel based systems
Medium: Core: Ensure non-libqb trace logging comes out at debug and only when appropriate
Low: Tools: crm_simulate - Improve action logging
High: crmd: Also purge fencing results when we enter S_NOT_DC
Medium: PE: Only complain about mis-matched values of clone_node_max is interleaving was requested
Merge branch 'master' of github.com:ClusterLabs/pacemaker
Low: cib: If we received the message, the peer is active enough for us to process it
Low: Build: Fix changelog generation
Low: Build: Fix the ABI comparision helper
Low: Fencing: Disable the config file until we decide if we want to support it
Low: Build: Tweak the indent helper
Low: Run indent on the C source and headers
Low: ais: Fix coverity NO_EFFECT error
Low: Core: Suppress coverity CONSTANT_EXPRESSION_RESULT warning
Low: Core: Suppress coverity FORWARD_NULL warning
Low: mcp: Resolve coverity CHECKED_RETURN error
Low: Core: Suppress coverity FORWARD_NULL false positive
Low: Fencing: Suppress coverity RESOURCE_LEAK false positive
Low: Core: Fix two coverity RESOURCE_LEAK errors in XML and XSLT handling
Low: PE: Suppress coverity NEGATIVE_RETURNS false positive
Low: PE: Resolve coverity FORWARD_NULL error
Low: PE: Keep coverity happy that FORWARD_NULL is impossible here
Low: cib: Resolve coverity RESOURCE_LEAK in error path
Low: tools: crm_mon - resolve coverity UNINIT error
Merge pull request #50 from davidvossel/cts-rate-limiting-bad
Merge pull request #52 from gao-yan/ticket_state
High: Build: Fix instalation by removing duplicate reference to crm_ticket.8
Merge remote-tracking branch 'public/master'
Low: doc: Fix 'Corsync' typo in clusters from scratch - thanks to dann.frazier at canonical.com
Low: extra: Allow manpages to be built for RAs when ocf-shellfuncs is not present
Low: Build: Fix compilation on rawhide and simplify doc logic
Low: tools: crm_simulate - Reliably send the transition summary to the terminal
High: PE: Correctly resolve a coverity FORWARD_NULL error, the previous attempt 'b6b9e1830c5570702563f5e636f262acc5dc3466' broke regression tests
High: PE: Correctly migrate a resource that has just migrated
Low: PE: Correctly migrate a resource that has just migrated (regression test)
Merge pull request #53 from gao-yan/coverity
Low: Build: Opensuse doesnt have publican
Low: Build: Ensure an editor is available for automated shell regression tests
Medium: Core: Use the default schema location when PCMK_schema_directory is defined but empty
Medium: PE: Correctly locate crm_simulate for automated regression tests
Low: Tools: Attempt to clean up memory in crm_simulate so that valgrind output is useful
Medium: Tools: Fix memory leak in crm_simulate (valgrind)
High: PE: Fix memory leak in order_rsc_sets (valgrind)
Low: PE: Suppress two valgrind errors from libqb that are now fixed upstream
Low: PE: Suppress error relating to the non-recurring creation of crm_system_name (valgrind)
High: Core: Correctly track the length of a string without reading from uninitialzied memory (valgrind)
High: Tools: Fix compilation by using the LIBQB_LOGGING define correctly
Low: Build: Fix the automated coverage tests
Low: Build: Drop unused config option
Low: Build: Man pages don't depend on the 'doc' specfile conditional anymore
Medium: Corosync: Reset connection variables after disconnect
High: cib: Ensure no additional messages are processed after we begin terminating
High: Corosync: Correctly disconnect from corosync variants
High: cib: Correctly shutdown in the presence of peers without relying on timers
High: cib: Determine when there are no peers to respond to our shutdown request and exit
Medium: corosync: Fix complation warning when corosync doesn't use libqb
Low: cib: Log connected clients at shutdown
Medium: cib: Even with no peers, we still need to wait for connected clients to disconnect
Low: Build: Update version and changelog
Angus Salkeld (6):
configure: ac_check_funcs requires ac_check_lib
Medium: mcp: change the link order
Medium: tools: Prefer built binaries when invoking help2man
High: ais: Prepare for corosync versions using IPC from libqb
High: Core: Support libqb for logging
Low: ais: Change some log_printf to qb_log_from_external_source
Barney Desmond (1):
High: crmd: Prevent use-after-free when calling delete_resource due to CRM_OP_REPROBE
Bernd Schubert (1):
Fix xml parser bug.
Dan Frîncu (42):
Low: doc: Plural form of arguments required
Low: doc: Consequence requires an extra e at the end
Low: doc: Extra _the_ removed
Low: doc: Re-worded the phrase
Low: doc: Reference could be of singular or plural, changed it to singular
Low: doc: Fixed typo
Low: doc: Fixed typo
Low: doc: Fixed typo
Low: doc: Fixed typo
Low: doc: Fixed typo
Low: doc: Added missing past tense
Low: doc: Fixed typo
Low: doc: Added missing subject
Low: doc: Added first Romanian translation draft for Pacemaker Explained
Low: doc: Proofreading Ap-Changes complete
Low: doc: Fixed typo
Low: doc: Proofreading Ap-Debug complete
Low: doc: Filled in header information for translation files
Low: doc: Proofreading Ap-FAQ complete
Low: doc: Backported fixed typos done on en-US to the initial EN part of ro-RO translation for consistency
Merge branch 'master' of git://github.com/ClusterLabs/pacemaker into dan
Low: doc: Proofreading of Ap-Install complete
Low: doc: Proofreading of Ap-LSB and Ap-OCF complete
Low: doc: Fixed typo
Low: doc: Proofreading of Ap-Samples and Ap-Upgrade-Config complete
Low: doc: Proofreading of Ap-Upgrade, Author_Group and Book_Info complete
Low: doc: Proofreading of Ch-Advanced-Options complete
Low: doc: Proofreading of Ch-Advanced-Resources complete
Low: doc: Proofreading of Ch-Basics complete
Low: doc: Proofreading of Ch-Constraints complete
Low: doc: Proofreading of Ch-Intro complete
Low: doc: Proofreading of Ch-Nodes complete
Low: doc: Proofreading of Ch-Options complete
Low: doc: Proofreading of Ch-Resources complete
Low: doc: Proofreading of Ch-Rules complete
Low: doc: Proofreading of Ch-Status complete
Low: doc: Proofreading of Ch-Stonith complete
Low: doc: Proofreading of Pacemaker_Explained complete
Low: doc: Proofreading of Preface and Revision_History complete
Low: doc: Fixed typos and spelling mistakes
Low: doc: Updates to the Romainian Pacemaker Explained translation
Low: doc: Initial Romainian translation of Clusters from Scratch
Dave Blaschke (11):
Bug 555 - Fix CTS bandwidth test when port not specified
Bug 665 - Updated to reflect latest 2.0 design, provide additional info
Bug 846 - Allow making rpm without lynx.
Fixed syslong-ng misspellings, added syslog-ng options to exerciser
Bug 892 - Enable CTS Split_brain tests on legacy cluster
Bug 892 part deux - Getting Split_brain to actually work required adding a
Bug 926 - TypeError in re.search() call.
fix one missed typo
OSDL bug 1553: suicide plugin lacking ST_DEVICENAME
OSDL bug 1553: suicide plugin lacking ST_DEVICENAME
OSDL bug 1563: ssh STONITH plugin cleanup
David Lee (46):
to handle a variety of uuid implementations, including our own for last-resort use
Let uuid_parse.c build when builddir and srcdir different
insert missing '#include <portability.h>' into 'lib/crm/common/*.c'
#include <portability.h> was absent
Somes OSes don't have getopt_long()
'#include <portability.h>' was absent
'#include <portability.h>' was absent
Somes OSes don't have getopt_long()
'#include <portability.h>' was absent
Somes OSes don't have getopt_long()
Somes OSes don't have getopt_long()
common code for syslog facility name/value conversion
migrate m4 inclusion of 'config/snmp_subagent.m4' from 'configure.in' to new 'acinclude.m4'
Allow for @HTML2TXT@ being null
More tolerant of non-GNU 'make' (still fine with GNU/make)
improve portability of [n]curses aspects
Compile-time portability fixes
Compile-time portability fixes
lib/crm/common/iso8601.c portability (with configure.in)
Remove the "libltdl.tar" aspects of the July 10th
Fix compiler portability problem: separate declarations from non-trivial initialisers
Fix a pathname portability glitch
strcpy() gave compiler warning/error. So '#include <string.h>'.
Fix segmentation fault on OSes whose 'struct dirent' has a minimal-length, placeholding 'char d_name[1];'
#include <portability.h>
For 'ctype' macros/functions, some platforms warn if argument is not of type (int)
replace/{setenv.c,unsetenv.c}: rationalise declarations and convey return data
AM_CHECK_PYTHON_HEADERS
Fix a compiler warning on some platforms
fix compilation warning on some platforms about undeclared 'strcmp()'
Fix a compilation warning when 'replace/daemon.c' is required
The recent change to 'include/stonith/stonith.h' (its rev. 1.17) requires extra '-I...' here for some environments.
Tidy a couple of config-time directory variables
Declarations should precede code
Update (1.196) yesterday exposed some declarations in the middle of code, causing compile-time errors.
On uninstall should ideally remove python's derived *.pyc and *.pyo files.
cts/Makefile.am rev 1.13 was a fudge. Should use automake's proper python support
rev 1.14 was good in principle. But Andrew noticed that it lost depedndency tracking. Fix.
Determine GNUTLS libs centrally in 'configure.in' rather than in individual Makefiles
Bourne shell 'test' does not support '-e' flag. Using '-r' instead.
Improve portability across versions of 'gcc'.
Fix a Linux compiler warn/error
Remove an unnecessary 'include' that broke building on some OSes. It continues to build OK on various OSes.
uname() return code: add (heartbeat/heartbeat.c) and improve error message (crm/admin/crm_attribute.c)
CTS: (minor): autoconfiscate a pathname
CTS: Improve a pathname construction.
David Vossel (16):
Low: tools: Bug rhbz#754216 - Add XML as an output option to crm_mon
Low: fencing: Add standalone configuration file for stonith_ng
Low: lib: Fixes segfault in xml.c when xmlDocGetRootElement returns NULL
High: pengine: Bug #5014, Prevent asymmetrical order constraints from causing resource stops
Low: pengine: Adds regression tests for bug #5014.
High: pengine: Bug #5007, Fixes use of colocation constraints with multi-state resources
Low: pengine: Adds regression tests for bug #5007
Low: tools: crm_mon's as-xml option by default prints to stdout.
Low: tools: Update crm_mon's help documentation concerning --as-xml option
High: pengine: Fixes regression in default use of symmetrical=true for ordered constraint sets caused by revision 03f9d869bd
Low: pengine: Adds additional regression tests for bug #5014 specific to the use of sets
High: pengine: Bug cl#5000 - Implements ability to create rsc_order constraint sets such that A can start after B or C has started.
Low: pengine: Bug cl#5000 - Implements ability to create rsc_order constraint sets such that A can start after B or C has started (regression tests)
Low: mcp: Pacemaker service file should ensure no error occurs when pacemaker environment file does not exist.
Low: crmd: Bug cl#5015 - Remove failure operation from lrm resource cache when failure-timeout occurs.
Low: tools: Bug cl#5032 - Fixes issue in iso8601 calculations.
Dejan Muhamedagic (809):
High: stonithd: ais support
High: build: enable building and packaging of stonithd
High: crmd: include stonithd subsystem
High: TE: enable stonith for openais
High: build: replace -lstonithd with a ref .la in the build tree
LRM: convert the LSB meta-data to XML.
LRM: fix my own fix, it broke lrmd.
CTS: a new script to collect pe input files for one CTS test.
Bug 1613 - stonith: convert long/short descriptions to xml.
Bug 1613 - stonith: convert long/short descriptions to xml.
LRM plugins: skip dot-files.
tools: ocf-tester: add capability to test with lrmadmin/lrmd
LRM (LF 1684): a meta-data operation fix introducing short sleep on zero reads
lrmd: performance 2: remove unnecessary memstat
tools: haresources2cib.py: update parameter lists and configuration defaults
raexecocf: a fix for bug introduced in 2a787bbd73f6; longer sleep in get_resource_meta
raexecocf: include time.h (thanks to Keisuke MORI)
tools: a reporting utility
tools hb_report: lookup logd.cf
lrmd (LF 1715): increase the max dispatch time for lrmd
hb_report: include net interface stats
hb_report: check if strings(1) is available
hb_report: rpm/install
hb_report: support both logging formats
hb_report: suppress some warnings and stderr output
hb_report: update the hb_report location
lrm stonith plugin: fix exit code handling
lrm stonith plugin: fix exit code handling
hb_report: a tiny doc update
hb_report: an update
merge with dev
cts (ComponentFail): error messages to be ignored: move the one belonging to the stonith library to common and add one for crmd
hb_report: fix collecting pengine graphs when not logging through syslog
lrmd (LF 1729): revise return codes
lrmd (LF 1729): revise return codes
hb_report: add logd.cf to the list of files to be analysed
hb_report: remove cib.xml from the list of files to be analysed
hb_report: support cts (use -f cts:nn)
hb_report: run crmadmin in background (it sometimes won't quit)
hb_report: analyze cib.xml, but don't consolidate it if crm_diff says that they are equal
hb_report: update documentation on extracting CTS tests
hb_report (LF 1763): multiple fixes
hb_report: include package verification
hb_report: fix the previous fix (sorry)
hb_report: use getopts instead of getopt(1) for the sake of portability
hb_report: add more packagers support (thanks to Sebastian Reitenbach)
hb_report: fix handling of debug option
hb_report: reduce the number of ssh invocations by 1
haresources2cib.py: update ra parameters lists
haresources2cib.py: set default-action-timeout to the default (20s)
High: build: include stonith LRM plugin
High: build: add Makefile.am for the stonith plugin
Low: build: add Makefile.am to lib/plugins
Low: packaging: drop spurious directory
merge with upstream
merge with upstream
Low: build: include libstonithd.la for crmd
Low: build: improve the dist target
Low: build: replace ref to heartbeat.spec
Low: CTS: set/define node_list (otherwise CTS with heartbeat fails)
Low: haresources2cib: replace _ with - in stonith_enabled
High: add the crm cli program
Low: packaging: add the crm program
Low: cli: forgot the license
Low: cli: don't ask questions when not interactive
Low: cli: shuffle file based code
Medium: cli: update cib sanitizing process
Low: cli: drop debug code
Medium: cli: sort children in containers
High: cli: cib not initialized before erase
High: cli: improve (python) instance management
Low: cli: print usage for commands on syntax error
Low: cli: update param usage
Low: cli: fix user preferences (a bit)
High: cli: keyword completion (static)
Low: cli: update usage
Medium: cli: update the configure delete command to remove more than one object
High: cli: when removing resources, remove the related constraints too
Low: cli: rename some variables
High: cli: fix removing related constraints
High: cli: completion (dynamic)
Low: cli: allow comments
High: cli: fix a typo bug (whole missing line)
Medium: cli: improve error reporting on refused update
High: cli: del and not .remove for dictionary entries
High: cli: fix check for objects to be updated (and not created)
High: cli: shlex allows newlines to creep in
High: cli: verify (new configure command)
High: cli: add cib attributes to the cib
High: cli: add interval=0 for ops if user didn't supply that
merge with upstream
High: new hb2openais.sh conversion script
High: hb2openais.sh: improve pingd handling; several bugs fixed
High: cli: new features (rc file,show options)
High: cli: support for cluster properties
High: build: rename hb2openais.sh to .in and replace paths with vars
Low: documentation: update the CLI document
Medium: cli: add completion for cluster properties (from pengine metadata)
High: cli: properties--fix copy-paste typo
High: cli: wrong lookup on xml edit
High: cli: configure refresh (new command)
Low: cli: commands started with '_' are only for debugging
Low: cli: update help
High: cli: change syntax (add ':' to score)
High: cli: fix (some) problems found by pylint
Low: cli: drop some unused code
Low: cli: move some code around
Low: cli: allow processing of nvpairs within the attributes tag
Low: cli: various code and performance improvements
merge with andrew
Low: cli: improve completion cache
High: cli: configure ptest (new command)
merge with andrew
High: cli: configure ptest (doc update)
Low: hb2openais: change provider to pacemaker
High: hb2openais: fix clone creation; replace EVMS strings
Low: Tools: cli: don't supply provider for stonith resources
Low: Tools: cli: don't show second prompt in batch mode
Low: Tools: cli: create ids the same like crm_resource
High: stonithd: fix handling of timeouts
Low: Tools: cli: convert '_' to '-' in attributes
Low: Tools: cli: add quotes to generated attributes
Medium: Tools: cli: refuse to configure constraints which reference non-existing resources
Low: Tools: cli: more consistent error reporting
High: Tools: cli: fix editing of container resources
Low: Tools: cli: ask user on EOF if they want to commit changes
Low: Tools: cli: save changes when editing from the command line
High: Tools: cli: fix the edit operation (on syntax errors)
Low: Tools: cli: replace tabs with spaces (zero code change)
Low: Tools: cli: improve style and maintainability (1)
Low: Tools: cli: improve style and maintainability (2)
merge with andrew
Low: Tools: cli: convert '_' to '-' in score_attribute
Low: Tools: cli: CibObject.propagate_updated (improve interface)
Medium: Tools: cli: fix id store updates on failed edit
Medium: Tools: cli: support for multiple attribute sets
High: Tools: cli: append crm properties to the objects list
Low: tools: hb2openais: move python code to a separate file
High: tools: hb2openais: add testing option
High: tools: hb2openais: fix nodes removal
Low: tools: hb2openais: improve logging
High: tools: hb2openais: fix processing of multiple ucast directives
Low: tools: hb2openais: check if the heartbeat's running
Medium: tools: hb2openais: HA_VARRUNDIR is HA_VARRUN
Low: tools: hb2openais: remove cib.xml.last too
High: tools: hb2openais: add the lha_crm service stanza to openais.conf
High: Tools: hb2openais: backup improved and multiple fixes
High: Tools: hb2openais: add option to upgrade the CIB to v3.0
Low: Tools: hb2openais: update user messages
High: Tools: hb2openais: add U option to getopts and update usage
High: Tools: hb2openais: fix class/provider reversal
Low: Tools: hb2openais: update user messages on CIB upgrade
High: Tools: hb2openais: move the CIB update to the end
High: Tools: hb2openais: fix testing
Low: doc: hb2openais: conversion document
Low: Tools: hb2openais: fix grep of ps output for crmd
Low: Tools: crm: fix usage message for shadow new
Medium: Tools: crm (bnc#441028): check for key error in attributes management
Low (bnc#438278): Tools: cli: replace popen2 with the subprocess module
Low (bnc#443792): Tools: cli: don't delete shadow cibs which are currently in use
Low: ais: add use_logd option
High: Tools: hb2openais: update logging and set logfile appropriately
Low: Tools: hb2openais: update openais.conf configuration to match what andrew set in the openais package
merge with andrew
Low: Tools: hb2openais: no encryption, no generate keys
High: Tools: crm cli: parser: return False on syntax error and None for comments
High: Tools: crm cli: unify template and edit commands
Low: Tools: crm cli: fix common_err call
High: Tools: hb2openais: start mgmtd if appropriate
Low: Tools: crm cli: up the number of Vs for ptest
Low: Tools: hb2openais-helper: include license
Low: packaging/build: make hb2openais-helper.py executable
Low: Tools: crm cli: define ptest options once only
Low (bnc#443339): Tools: crm cli: add completion for help commands
High: Tools: crm cli: sometimes, on errors, level would change on single shot use
Low: Tools: crm cli: don't print error if cib new fails
Low: Tools: crm cli: cib new implies cib use
Low: Tools: crm cli: use validate-with attribute to check if the CIB is supported
Low: Tools: crm cli: implement force for cib new
Low: Tools: crm cli: remove the symmetrical option from colocations
Low: Tools: crm cli: recognize INFINITY as special score
Medium: Tools: crm cli: replace inf with INFINITY
Medium: Tools: crm cli: supply provider only for ocf class primitives
High: Tools: crm cli: exit with an appropriate exit code
High: Tools: crm cli: prevent key errors on missing cib attributes
High: Tools: crm cli: catch system errors on file operations
High: Tools: crm cli: new command: configure upgrade
Low: Tools: crm cli: python doesn't like two lines without backslash
Low: Tools: crm cli: cib upgrade referenced an undefined variable
Low (bnc#447216): Tools: crm cli: test if the tty is terminal
merge with Andrew
High: Tools: crm cli: fox wrong exit code
High: Tools: crm cli: improve handling of cib attributes
Medium: Tools: crm cli: keep order of help topics
Low: Tools: crm cli: unify help topics order
Low: doc: crm cli: replace exit with quit
Low: Tools: crm cli: fix lines ending with backslash in help strings
High: Tools: crm cli: short node preference location constraint
High: Tools: crm cli: return on syntax error when parsing score
Low: Tools: crm cli: translate INFINITY score to inf on conversion to the cli format
High: Tools: crm cli: rsc_location can be without nvpairs
High: Tools: crm cli: drop '-rsc' from attributes for order constraint
merge with andrew
Low: Tools: crm cli: use the right INFINITY thing for the mandatory keyword
High: Tools: crm cli: fix wrong order of action and resource in order constraint
Medium: Tools: crm cli: add back symmetrical for order constraints
Low: Tools: crm cli: infrastructure for different cli output
Low: Tools: crm cli: tell user that we're starting dotty to show the graph
Low: Tools: crm cli: prefix info messages with INFO:
Low: Tools: crm cli: allow '-' as stdin/stdout for load/save/template
Low: Tools: crm cli: report syntax error properly when there's no rule keyword in location
Low: Tools: crm cli: improve syntax error reporting
Medium: Tools: crm cli: generate role in location when converting from xml
High: Tools: crm cli: completion for ids in configure
Medium: Tools: crm cli: refine completion for ids in configure
High (bnc#450815): Tools: crm cli: don't generate id for the operations tag
Low: Tools: crm cli: further refine completion for ids in configure
High: Tools: crm cli: new command: configure rename
High: Tools: crm cli: allow rename only if the resource is stopped
High: Tools: crm cli: syntax: drop a bunch of commas (remains of help tables conversion)
Low: Tools: crm cli: replace _ with - in many examples/help texts
Low: Tools: crm cli: don't ask for confirmation if configure is run from the top level
Low: Tools: crm cli: update usage text
High: Tools: crm cli: print long help for help topics
High: Tools: crm cli: new command: node delete
Medium: Tools: crm cli: streamline cib load and parsing
Low: Tools: crm cli: allow cib use live to switch to the cluster cib
Low: Tools: crm cli: drop shadow. from shadow name in list
Medium (bnc#445622): Tools: crm cli: improve the node show command and drop node status
High: build: create and install doc files
High: Tools: crm cli: verify user input for sanity
Low: Tools: crm cli: improve output of node show
Low: Tools: crm cli: protect is_id_valid from bad input
Low: Tools: crm cli: properties usually do not need ids
Low: Tools: crm cli: check and store properties' ids
Low: cts: os.exit is sys.exit
Low: cts: improve syslog facility processing
Low: Tools: hb2openais: inform users that the logd service should be running
Low: cts: update ignore message when aisexec fails
Low: cts: update ignore message for ComponentFail/LinuxHAv2
Low: cts: more update ignore message for ComponentFail/LinuxHAv2
Low: Tools: hb2openais: generate encryption keys
High: Tools: hb2openais: set no-quorum-policy to ignore for 2-node clusters
High: Tools: hb2openais: adjust path to the supporting script and set CONF variable
Medium: Tools: hb2openais: reuse code
Medium: Tools: crm cli: handle shlex exceptions
Low: Tools: crm_uuid: crm_uuid is obsolete on openais
Medium: Tools: hb2openais: replace crmadmin with crm_mon
Low: Tools: crm_mon: exit with proper code on one_shot mode
Low (bnc#468262): Tools: crm: support for --help
Low: cts: SplitBrain: two more messages to ignore
Low: Tools: ocf-tester: add -v to usage
Low: Tools: pingd: reduce severity for the update sent message
Low: cts: Reattach: make sure that the resource management is reenabled
Low: cts: standby2: make sure that a node is set back to active (standby off)
Medium: Tools: hb2openais: replace the lsb class with ocf for o2cb
Medium: Tools: hb2openais: do not remove Evmsd resources, but replace them with clvmd
High: Tools: hb2openais: openais supports up to two rings
Low: Tools: hb2openais: enable encryption (secauth) by default
High: Tools: hb2openais: improve media processing and set rrp_mode for multiple rings configuration
High: Tools: hb2openais: add constraints for clvmd to controld
Low: doc: crm_fencing.txt: new document about fencing
Low: doc: crm_fencing.txt: add descriptions for special devices
High: Tools: crm: find expressions within rules (don't always skip xml nodes due to used id)
Low (LF 2079): Tools: crm: fix usage for node attribute and set-attribute
Low (LF 2078): Tools: crm: fix usage for resource cleanup
Low: Tools: hb2openais: set rrp_mode to active for multiple media clusters
Low: Tools: hb2openais: parametrize the supported number of interfaces
Low: Tools: hb2openais: make CIB upgrade to v1.0 default
Low: Tools: hb2openais: set threads to the number of cpus if authentication is enabled
Low: Tools: hb2openais: instead of zapping the nodes section, replace node ids with unames
Low: Tools: hb2openais: set expected-nodes to the number of nodes found in heartbeat
Low: Tools: hb2openais: streamline setting cluster properties
High: Tools: hb2openais: create clvmd clone for respawn evmsd in ha.cf
High: Tools: hb2openais: fix a serious recursion bug in xml node processing
High: Tools: hb2openais: fix ocfs2 processing
Low: Tools: cts: rearrange order so that list-tests works without connecting to nodes
Low: Tools: cts: it is experimental (not exprei...)
Low: Tools: cts: ha_logd is not anymore controlled by the cluster init script
Low: Tools: crm: refine CibFactory/CibObject
High: Tools: crm: rsc/op_defaults
Low: Tools: cli: replace repeating code with a function
High: Tools: cli: fix and improve delete command
Low (bnc#478062): Tools: crm: on delete of composite objects, leave their children around
Low (bnc#479248): Tools: crm: clones are started/stopped by manipulating clone-max
Low: Tools: crm: node standby/online node defaults to this node
Low (bnc#478649): Tools: crm: resource cleanup expand clones/ms
Medium: Tools: crm: ignore comments in the CIB
High: Tools: crm: support for no value attribute in nvpairs
High: Tools: crm: fix parsing attributes for rules (broken by the previous changeset)
Low: Tools: crm: let people know that graphviz can show transition graphs
Low: Tools: crm: add line numbers to messages
High: Tools: crm: create top configuration nodes if they are missing
Low: Tools: crm: don't create configuration nodes unless necessary
High: Tools: crm: resource agents information management
merge with andrew
Low: Tools: crm: make lrmadmin happy
Medium (bnc#479049): Tools: crm: add validation of resource type for the configure primitive command
Dev: Tools: crm: fix confusion with lists/strings in cs a23bbc615e46
Dev: Tools: crm: pass user string to ra_type_validate
Low: Tools: crm: remove the enter command from the documentation
High: Tools: crm: new ra set of commands
High: Tools: crm: add support for command aliases and some common commands (i.e. cd,exit)
Low: Tools: crm: improve output of ra types
Low: Tools: crm: add documentation for the ra command
Dev: Tools: crm: catch exceptions on parsing the RA meta-data
Low: Tools: cts: add more patterns to ignore for the Stonithd test
Low: Tools: crm_diff: don't dump core if xml parse failed
Low (bnc#484298): Tools: crm: unbreak ptest
High: Tools: crm: the new configure node command
Dev: Tools: crm: include nodes when generating xml from an object list
Low: doc: crm_fencing: review
Low: Tools: crm: replace hb_libdir with libdir
Low: Tools: cts: drop colon from one statement
Dev: Tools: crm: rename is more complex
Low: Tools: crm: no need to copy nodes; we have our own now
Low: Tools: crm: add several command aliases
Low: Tools: crm: optimize overview help (join commands and their aliases)
Low: Tools: crm: help code cleanup
Dev: Tools: crm: preserve order of help topics (use a list not a dictionary)
Low: Tools: crm: print short help if there is no long help; replace "desc" with "help"
Low: Tools: crm: issue warnings on all unrecognized elements of the cib configuration
High: Tools: crm: the new configure monitor command
Low: Tools: crm: return exit code from configure commands
Low: Tools: crm: check shadow name for sanity in the cib use command
Low: Tools: crm: more shadow name sanity checks
Low (LF 2092): Tools: crm: check operation name
Low (LF 2092): Tools: crm: monitor requires interval
Medium (bnc#479050): Tools: crm: add tab completion for primitive params/meta/op
Medium (bnc#479050): Tools: crm: add help for RA parameters in tab completion
Medium (bnc#479050): Tools: crm: reimplement cluster properties completion
Low: Tools: crm: update line number while parsing edit/load/template files
Low: Tools: crm: lrmadmin moved to @sbindir@
Low: Tools: crm: lrmadmin moved to @sbindir@ (part 2)
Low: Tools: remove hb_report and ocf-tester
Medium (bnc#486968): Tools: crm: listnodes function requires no parameters (don't mix completion with other stuff)
Low: Tools: crm: add shadow management to the configure level (improve usage)
Low: Tools: crm: streamline subprocess usage (code cleanup)
Low: Tools: crm: add tab completion for shadows
merge with andrew
merge with andrew
Low: Tools: crm: consolidate tab completion help and don't print it too often
Medium: Tools: crm: refuse to delete running resources
Low: doc: crm_fencing: minor changes
Low: Tools: crm: reset cache when we know it's outdated (and some name changes)
Low: Tools: crm: allow crm conf[igure] to enter a configure session
Medium: Tools: crm: fix id-refs and allow reference to top objects (i.e. primitive)
Low: Tools: crm: make sure that master works as alias to ms
Low: Tools: crm: allow only y and n for answer
Low: Tools: crm: replace obsolete to be os.popen by subprocess
Low: Tools: crm: lrmadmin may need sudo
Low: Tools: crm: don't commit if crm_verify reports ERRORs
Low: Tools: crm: don't invoke pager unless really necessary (i.e. big output)
Low: Tools: crm: don't remove parents twice in delete
Dev: Tools: crm: process delete in correct order and on edit pass a list of objects to be deleted
Low: Doc: crm_cli: documentation update
Dev: Tools: crm: enable default for node standby/online
Medium: Tools: crm: add more user input checks
Low: Tools: crm: replace obsolete to be os.system by subprocess
Low: Tools: crm: fix error messages buffering
Low: Tools: crm: show childless attribute list nodes if they have an id
Dev: Tools: crm: fix handling of default or missing attributes
Low: Tools: crm: looks like boolean-op defaults to and
Low: Tools: crm: check if stdin is a tty early
Low: Tools: crm: show offending xml if it can't be represented in cli
Dev: Tools: crm: cib use is more complex when invoked from configure
Low: Tools: crm: don't confuse people if cib use fails
Dev: Tools: crm: combine multiple creates (-C) into a single update (-U) cibadmin invokation
Dev: Tools: crm: try invoking a method joe
Medium: Tools: crm: don't check resource status of we're working with a shadow
Medium: Tools: crm: multiple column output wouldn't work with small lists
Low: Tools: crm: set min column length to 8
Low: Tools: crm: make info an alias for meta (ra)
Low: Tools: crm_regression: move crm_regression files to tools/shell/regression
Low: Tools: crm: add references to collocation
High: Tools: crm: add and implement templates
Low: Tools: crm: replace otherwise unused prefixes in templates
Low: Tools: crm templates: the ip address name is not that important
Dev: Tools: crm: wrong template id set
Dev: Tools: crm: replace prefixes more thoroughly
Medium: Tools: crm: rudimentary if-else for templates
Low: Tools: crm templates: add filesystem
Low: Tools: crm: catch exceptions in case of os errors (templates)
Low: packaging: install the filesystem template
Low: Tools: crm: define lineno early
Low: Tools: crm: output error messages to stderr
Low: Tools: cts: tune messages for ComponentFail stonithd and aisexec
Dev: Tools: crm: don't remove nodes on erase
Medium (LF 2107): Tools: crm: revisit exit codes in configure
Low: Tools: crm: more support for regression testing/add proper options parsing
Low: Tools: crm: drop symlink silliness
Medium: Tools: crm: work with parents only when managing xml objects
Dev: Tools: crm: fix cli multiple object edit
Dev: Tools: crm: fix verify exit code
Dev: Tools: crm: don't assign to tuples (in collocation)
Dev: Tools: crm: drop 'and' as default boolean-op
Medium: Tools: crm: implement erasing nodes in configure erase
Low: Tools: crm: improve check for short node preference
Low: Tools: crm: resolve id-refs for rules
Low: Tools: crm: don't add id for a rule if it contains an id-ref
Low: Tools: crm: sort objects for cli output (more predictable)
Low: Tools: crm: sort ra classes/providers
Low: Tools: crm: don't print ra types/providers in multiple columns when regression testing
High: Tools: crm: regression tests
Low: Tools: crm: replace sed script (regression)
Low: Tools: crm: change source for the lsb ra (regression)
Low: Tools: crm: add quotes around the sed expression
Low: Tools: crm: use -N instead of -U for nodes
Low: Tools: crm: update regression expect files to reflect the latest changes in wording
High (bnc#488291): ais: don't rely on byte endianness on ptr cast
Low: Build: -Wno-pointer-sign is available as of gcc 4
Dev: Build: automake can't digest ifeq (unbreak build)
Low: Tools: hb2openais: fix a possible bad reference
High (LF 2114): Tools: crm: add support for operation instance attributes
Low: Tools: crm regtest: improve empty.xml
Low: Tools: crm regtest: add operations instance attributes test
Medium (bnc#507255): Tools: crm: handle empty rsc/op_defaults properly
Medium: Tools: crm: catch syntax errors for configure load
Medium (bnc#507255): Tools: crm: use the right obj_type when creating objects from xml nodes
High (bnc#507255): Tools: crm: import properly rsc/op_defaults
High (bnc#507255): Tools: crm: delete rsc/op_defaults (these meta_attributes are killing me)
Medium: Tools: crm: fix the verify exit code properly
Dev: Tools: crm: recreate constraints on grouping resources
Low: ais: set HA_LOGFACILITY too
High: Tools: crm: support for color output
Dev: Core: getopt and friends return int (fixes 41e9a756cf15)
Dev: ais: ais_ipc_handle is for corosync
Dev: core: comment out debug printf
Low: doc: mention that python-curses is needed for color output
Low: Tools: remove spurios newline output to stderr
Dev: Tools: crm: do not allow empty attribute names
Low: Tools: crm: make check for operations id attribute stricter
Dev: Tools: crm: implement binary op types properly (thanks to Alain St-Denis)
Dev: Tools: crm: allow id-refs to stand by themselves (update for fix 37077ed56af5)
High (bnc#507255): Tools: crm: implement date expressions
Low: Tools: crm regtest: include date expressions
Low: Tools: crm: let the user know that curses is needed for colors
High: Tools: crm: new display type (uppercase keywords)
Low: ais: spelling
Low: Tools: crm: make pengine metadata globally accessible
Low: Tools: crm: fix one wrong function reference
High: Tools: crm: improve the edit command
Dev: Tools: crm: fix edit change from the 5e8f9f4cfad7 changeset
High: Tools: crm: add semantic checks depending on the meta-data from resource agents
Low: Tools: crm regtest: update tests to pass the semantic checks
merge with upstream
Low: Tools: crm regtest: match the latest output from crm_resource
Dev: Tools: crm: fix wrong logic in is_check_always()
Low: doc: update crm_cli date/release
Dev: Tools: crm: replace lookup_cli() with lookup() in the xml processing part
Dev: Tools: crm: change interface of create_from_cli (and fix create in edit)
Dev: Tools: crm: change interface of create_from_node (and fix create in xml edit)
Dev: Tools: crm: make get_stdin show stderr output
Low: Tools: crm: update the all_committed flag everywhere
Low: Tools: crm: print message on constraint changes while deleting resources
Low: Tools: crm: variable name change
Low: Tools: crm: comment out some debugging code
Low: Tools: crm: add which(1) to the list of prerequisites
Low: Tools: crm: check prerequisite programs at the beginning
Medium: Tools: crm: expand template documentation
Medium (LF 2164): Tools: hb_report: expand the crm status command
Medium (bnc#520707): Tools: crm: new templates ocfs2 and clvm
Low: Tools: crm: update resource regression expect file
Medium (LF 2185): Tools: crm: add resource promote/demote commands
Low (LF 2175): Tools: crm: add help for node status
Low (LF 2177): Tools: crm: add multiple instance attributes
Dev: Tools: crm: fix initialization of a dict in node show
Low (LF 2179): Tools: crm: show that the node is offline if it's offline in node show
Low (LF 2181): Tools: crm: page output of cib diff
Medium: Tools: crm: add extra flag to CibObject for invalid objects
Dev: Tools: crm: add default for offline in node show
Medium: Tools: crm: first child of a removed parent inherits constraints
Low: RA: Stateful: add roles for monitor in the metadata
Low: Tools: crm: show queues command (debugging)
Low: Tools: crm: show cib objects command (debugging)
Low: Tools: crm: fix a wrong variable name
High: Tools: crm: improve processing of group edit and constraints
merge with upstream
Low: Build: add crm_fencing and html files to specfiles
Low: cts: replace heartbeat class IPaddr with ocf
Low: cts: fix bad time specification (P10S?)
Low: cts: replace heartbeat class properly
Low: cts: undo d5083751e9ac (P10S valid)
Low: cts: add back the heartbeat class resource
Low: Tools: crm: don't verify time values for interval and start-delay (no iso8601 support yet)
Low: Tools: crm: fix wrong logic in is_live_cib
Medium: Tools: crm: do return cached resources dom node
Medium (LF 2184): Tools: crm: extend ptest command
Medium (LF 2198): Tools: crm: add node fence command
Low (LF 2181): Tools: crm: crm_shadow(8) now prints to stdout
Low: Tools: crm: add standard syntax for ra meta
Low: cts: fix pattern for resource start
Medium: cts: add --benchmark for scalability tests
Low: cts: log test start/stop times and test runtime
Low: cts: set_starttime and the Running test message exchange places
High (bnc#547579,547582): Tools: crm: status section editing support
Dev: Tools: crm: fix syntax error
Dev: Tools: crm: update shadow regression test case
Low (LF 2183): Tools: crm: resource restart command
Dev: Tools: crm: refine/update status editing
merge with upstream
Low: Tools: crm: add help at the top level for ra
Low: Tools: crm: rudimentary debug support
Low: Tools: crm: fix getopt for debug
Low: Tools: crm: don't print xml always
High: Tools: crm: resource sets (bnc#550923)
Dev: Tools: crm: fix action/role in resource sets
Dev: Tools: crm: fix parentheses parsing in resource sets
High: Tools: crm: support for description attribute in resources/operations (bnc#548690)
High: Tools: hb2openais: add EVMS2 CSM processing (and other changes) (bnc#548093)
Medium: Tools: crm: don't add score which doesn't exist
Dev: Tools: crm: copy score from/to resource_set
Dev: Tools: crm: update tab completion for constraints (support resource sets)
Dev: Tools: crm: handle resource_set resource references in delete and rename
Medium: Tools: crm: drop empty attributes elements
High: Tools: hb2openais: don't allow empty rules, clones, or groups (LF 2215)
High: Tools: hb2openais: refuse to convert pure EVMS volumes
Low: Tools: hb2openais: add inf score to order constraints
Medium: Tools: crm: verify changes on exit from the configure level
Medium: fencing: stonithd: add metadata for stonithd instance attributes (and support in the shell)
Low: Tools: crm: drop short description from the metadata if it equals long description
Low: Tools: crm: update resource regression expect file (match e94a683c48af)
Low: Tools: crm: update ra regression expect file (match 2447829cb31f)
Low: Tools: crm: update file regression test (match f49a0cab20aa)
Low: Tools: crm: relax checks for advertised monitor operations (LF 2222)
Low: Tools: crm: replace some re.match with re.search
High: Tools: crm: support for comments (LF 2221)
Low: Tools: crm: test operation name against a list of possible ops
Dev: Tools: crm: issue only warning on unrecognized operations
Dev: Tools: crm: revert cs 1687e90aa5c9 (support for comments goes to dev)
Low: Tools: crm: return on exception in xmlparse
Low: Tools: crm: return on exception in xmlparse (this time really)
Low: Tools: crm: relax check for RA actions
Dev: Tools: crm: convert tupple to list before parsing (LF 2278)
Dev: Tools: crm: fix indentation (LF 2278)
Medium: Tools: crm: revisit configure commit
Medium: Tools: crm: don't consider warnings as errors (LF 2274)
Medium: Tools: crm: direct RA interface for ocf class resource agents (LF 2270)
Low: Tools: crm: define missing index in validate_template (LF 2287)
Medium: Tools: crm: stay in crm if user specified level only (LF 2286)
Medium: Tools: crm: fix exit code on single shot commands
Low: Tools: crm: allow entering level and input file (-f)
Low: Tools: crm: relax check for RA existence (LF 2270)
Low: Tools: crm: fix OCF_ROOT processing
Low: Tools: crm: catch EOFError exception in raw_input (LF 2289)
Medium: Tools: crm: implement -F (--force) option
Low: Tools: crm: rename some misnamed functions
Medium: Tools: crm: direct RA interface for stonith class resource agents (LF 2270)
Dev: Tools: crm: fix and uniq and sort the RA types list
Low: Tools: crm: direct RA interface for lsb class resource agents (LF 2270)
Low: cts: add log mark function
Low: cts: fix one log mark reference
Low: cts: standby: add log marks and drop sleeps
Medium: Tools: crm: exclude locations when testing for pathological constraints (LF 2300)
High: Tools: crm: load help from crm_cli.txt
Dev: Tools: crm: drop a debug message
Low: pengine/crmd: move crm_log_init after version and metadata calls (LF 2272)
Dev: Tools: crm: move one reference of CibShadow down
Dev: Tools: crm: don't fall apart if there's no help text file
Low: Tools: crm: cache help tables (LF 2303)
Low: Tools: crm: more optimizations (LF 2303)
Dev: Tools: crm: build completion tables happens before interactive is set
Low: Tools: crm: create a parent class for the UI classes
Low: Tools: crm: simplify template for ocfs2
Medium: Tools: crm: don't remove sets which contain id-ref attribute (LF 2304)
Low: Tools: crm: handle configure cib use non-interactively too
Dev: Tools: crm: create and use index for help file (perf)
Dev: Tools: crm: fix command aliases processing
Medium: Tools: crm: add update method to template apply (LF 2289)
Low: Tools: crm: make template new more robust, add documentation
Low: Tools: crm: page help text
Dev: Tools: crm: fix check for invalid constraints
Medium: Tools: crm: rename status to cibstatus (LF 2236)
Medium: Tools: crm: fix node delete (LF 2305)
Low: Tools: crm: save options if run from the command line (LF 2310)
Low: Tools: crm: restore setting interactive flag with command line use
Low: Tools: crm: add notify to the list of valid operations
Low: cts: ignore lrmd notify_client error for Reattach
Low: cts: add sleep to the Standby test
Medium: Tools: crm: don't verify empty configurations (lf#2316)
Low: Tools: crm: fix exit codes for show/edit empty configuration
High: Tools: crm: recovery from bad/outdated help index file
Medium: Tools: crm: check for duplicate children when creating groups (lf#2326)
Low: doc: update links in crm_fencing.txt
High: Tools: crm: move scores from resource sets to the constraint element (lf#2331)
Medium: Tools: crm: don't allow score-attribute in orders
Medium: Tools: crm: don't produce empty resource sets when adding roles/actions
Low: Tools: crm: validate roles/actions in resource sets
High: Tools: crm: support for the utilization element
Medium: Tools: crm: install crm_cli.txt also in the datadir
Medium: Tools: crm: use the CRM_HELP_FILE variable if set
Medium: Tools: crm: fix reference replacement in resource sets
Medium: Tools: crm: catch IOErrors when opening files
Medium: Tools: crm: don't fiddle with cib when there's no cib (bnc#575701)
Medium: Tools: crm: fix exit code for template apply
Low: Tools: move tools/shell to the top level directory
Low: cts: refine timers interface (benchmark)
Low: build: fix autoconf machinery for the new shell directory location
Medium: Tools: crm: fix CIB upgrade command (bnc#578637)
High: Shell: split shell into modules
Low: Shell: remove old crm.in
Low: Shell: cache nodes as node_list (a typo)
Low: Shell: set global RA variables on demand
Low: Shell: node delete cannot be tested in regression tests
Low: Shell: fix function reference in debug
Low: Shell: help: don't get into the next level if it starts with the same string (cib/cibstatus)
Medium: Shell: help: unsort help items
Low: Shell: catch startup errors
Dev: Shell: add comm_node initialization
Dev: Shell: disable comments (till they're ready)
build: use autoconf instead of setup.py for shell
High: Shell: improve configure commit (lf#2336)
Low: doc: describe resource sets use in crm_cli
Low: Shell: add tab completion for op/rsc_defaults (nbc#552541)
Medium: Shell: add option to control sorting of cib elements (lf#2290)
Dev: Shell: one backslash missing in the previous cs f6011edefcfd
Dev: Shell: move one utility to userprefs
build: use autoconf instead of setup.py for shell
Medium: Shell: parse lists not tupples
High: Shell: support for the utilization element (old patch for the new structure)
Low: Shell: add tab completion for nodes (nbc#577003)
Low: Shell: add tab completion for nodes (nbc#577003)
Dev: Shell: fix wrong var reference
Dev: Shell: cleanups (cliformat,idmgmt)
Dev: Shell: restore some utilization bits
Low: doc: fix rule in one location example
Medium: Shell: update supported CIB RE to include pacemaker 1.1 and 1.2
High: Shell: update previous node lookup procedure to include the id where necessary
Dev: Shell: more cli format cleanups
merge
Dev: Shell: update the utilization parts
Dev: Shell: reorganize element create and update
Dev: Shell: restore element testing
High: Shell: restore error reporting in options
Low: Shell: don't print prompt in the batch mode
Low: Shell: don't say goodbye either in the batch mode
High: Shell: add support for xml in cli
Medium: Shell: repare clone/ms cleanup (nbc#583288)
Dev: Shell: fix id_store/element updates interaction
Low: Shell: add two convenience error functions
Medium: Shell: load update should update existing elements
Dev: Shell: more cleanups in cib updates
Low: doc: describe xml in cli
Low: Shell: more error checking when converting to the configuration language
Dev: Shell: import shlex in ui.py
Medium: Shell: implement lifetime for rsc migrate and node standby (lf#2353)
Low: Shell: don't ask questions without terminal
Medium: Shell: node attributes update in configure (bnc#582767)
Dev: Shell: correct previous node updates changeset
Low: Shell: update regression tests
Low: Shell: no need to start stonithd
High: Shell: check timeouts also against the default-action-timeout property
Low: Shell: catch KeyboardInterrupt and show appropriate message (thanks to T. Schraitle) (bnc#584638)
Low: Shell: tone down messages on elements which cannot be represented in the configuration language (bnc#584289)
Dev: Shell: repair loading status from shadow files
High: Shell: new cibstatus import command (bnc#585471)
Medium: Shell: don't cache node and resource ids (lf#2368)
Low: Shell: no need to clear cache on commit (lf#2368)
High: Shell: new configure filter command
High: Shell: edit multiple meta_attributes sets in resource management (lf#2315)
High: Shell: edit multiple meta_attributes sets in resource management (lf#2315)
Medium: Shell: fix commit for new clones of new groups (bnc#585471)
Medium: Shell: fix commit for new clones of new groups (bnc#585471)
Low: Shell: update the resource test case
Dev: Shell: don't touch resources within a group when managing groups/clones
Dev: Shell: don't touch resources within a group when managing groups/clones
Medium: Shell: repair node show (thanks to T. Schraitle) (bnc#587883)
Medium: Shell: repair node show (thanks to T. Schraitle) (bnc#587883)
Low: Shell: check python version at start (lf#2369)
Medium: Tools: hb2openais: add support for corosync (and more)
Low: build: fix typo in BUILD_OPENIPMI_SERICELOG
Medium: Tools: hb2openais: add support for corosync (and more)
Low: build: fix typo in BUILD_OPENIPMI_SERICELOG
Low: Shell: templates: fix ocfs2
Low: Shell: templates: fix ocfs2
Low: Shell: check python version at start (lf#2369)
Medium: Shell: support for resource/node utilization
Dev: Shell: switch to single cibadmin for commit
High: Shell: enable comments (lf#2221)
Low: Shell: print warning with sys if no curses support (lf#2372)
Medium: Shell: verify only changed elements on commit (bnc#590033)
Medium: Shell: observe op_defaults when verifying primitives (bnc#590033)
Low: Shell: update meta attributes list
Medium: Shell: verify meta attributes and properties (bnc#589867)
Low: Shell: print warning with sys if no curses support (lf#2372)
Medium: Shell: verify only changed elements on commit (bnc#590033)
Medium: Shell: observe op_defaults when verifying primitives (bnc#590033)
Low: Shell: update meta attributes list
Medium: Shell: verify meta attributes and properties (bnc#589867)
Low: Shell: fix wrong exit code setting logic in resource cleanup
Low: Shell: fix wrong exit code setting logic in resource cleanup
Dev: Shell: fix possible KeyError when building help index
Dev: Shell: fix possible KeyError when building help index
merge with 1.1
Medium: RA: SystemHealth: exit properly when the required software is not installed (bnc#587940)
Low: CTS: update pattern for Stonithd test
High: Shell: implement new cibstatus interface and commands (bnc#580492)
Medium: Shell: add support for utilization in ptest/simulate (bnc#589873)
Dev: Shell: fix regression introduced in c7bb6cb8da24 (bad cib passed to crm_verify)
merge with 1.1
Low: Shell: add two attributes not present in the crmd metadata
Low: Shell: add two attributes not present in the crmd metadata
Low: Shell: templates: update apache documentation
Low: Shell: templates: new sbd template
Low: Shell: add another attribute not present in the crmd metadata
Low: Shell: don't allow empty attributes for properties/*_defaults (lf#2395)
Low: Shell: add another attribute not present in the crmd metadata
High: Shell: always reload status if working with the cluster (bnc#590035)
High: Shell: always reload status if working with the cluster (bnc#590035)
Low: Shell: add new regression tests
Low: Shell: templates: add dlm to the clvm template (bnc#595135)
Low: CTS: reduce dc-deadtime to 5s (lf#2381)
Low: Shell: update and simplify regression tests
Dev: build: remove references to lrmregtest-lsb
Low: Shell: fix validate-with in the source xml
Low: Shell: update the delete test case
Medium: Shell: prefer PATH to CRM_DAEMON_DIR (for crmd,pengine,stonithd)
Low: Shell: don't print message about building help index when regression testing
Low: crm_simulate: add colon after option q (thanks to Lars Ellenberg)
Low: Shell: print message about building help index only on debug
Low: doc: described how to specify the role in operations
Medium: Tools: benchmark test suite
Medium: Shell: new command default-timeouts to set default operation timeouts (bnc#479052)
Medium: Shell: don't error on missing resource agent with asymmetrical clusters (lf#2410)
Medium: Shell: don't verify empty configurations (bnc#602711)
Low: cts: Reattach: sleep before re-enabling management to allow connected attribute update (bnc#598807)
Medium: Tools: crm_mon: refresh screen on terminal resize (bnc#589811)
Low: Build: get HA_RSCTMPDIR from agent_config.h (lf#2378)
Low: Build: agent_config.h is optional
Medium: ais: create the final directory too for resource agents (bnc#603190)
Low: cts: default to corosync for the cluster stack in cluster_test (bnc#602873)
Medium: Shell: support for clone instances in constraints (lf#2169)
Low: Shell: templates: set order to mandatory between the base services and ocfs2 (bnc#605580)
Low: cts: add and process the loop-minutes option
Shell: templates: clvm,ocfs2: add the fs/vg resource to the base group, add timeouts (bnc#605580)
Medium: Shell: skip resource checks for property-like elements (lf#2420)
Medium: Shell: on no id match the first of property-like elements (lf#2420)
Low: Shell: set proper ids for rsc/op_defaults (lf#2420)
Low: Shell: regression: filter out providers other than heartbeat and pacemaker for the ra testcase (lf#2418)
Low: Shell: implement ra info for programs (e.g. pengine)
Low: Shell: hide some attributes of the crmd metadata
Medium: Shell: add the actions option to configure ptest (lf#2360)
Medium: Shell: add exit codes for non-interactive use
Low: Extra: ping,pingd: print correct default attribute name in the meta-data
Dev: Shell: fix changeset 3aa4c83ec3e5
Dev: build: fix reference to lrm_op_t in configure
Medium: Shell: add support for classes to the providers command
High: crmd: several fixes for async resource delete (thanks to beekhof)
Low: Tools: hb2openais: require /bin/bash to run the script
Medium: Shell: test early for duplicate elements in edit (bnc#620460)
High: stonith: pass the configuration to the stonith program via environment variables (bnc#620781)
Low: Shell: print error if a resource is not found in stop/start/manage (bnc#627757)
Medium: Shell: set shadow directory properly (bnc#626638)
Low: Shell: doc: add the cib help item at the configure level
Medium: Shell: refine the node delete command
High: stonith: cleanup children on timeout in fence_legacy
Low: Shell: use uname instead of id in nodes when building other ids
Low: crmd: on version print everything to stdout
Medium: Shell: update shadow file directory processing
Low: PE: reduce logging severity for single rsc sets
Medium: Shell: warn on non-existing nodes in location constraints (lf#2486)
Medium: Shell: repair ability to display multiple property sets
Low: Shell: fix one wrong variable reference
Medium: Shell: repair readline history file save
Low: Shell: revert f36fde0938f7 (repair tab completion)
Dev: Shell: put completion bits into a separate module
Dev: Shell: add the new module to the Makefile
Medium: Shell: merge properties and rsc/op_defaults on load update (lf#2454)
Medium: stonith: use ST_DEVICEID for short description
Medium: Shell: combine crmd and PE meta-data (lf#2373,2419)
Medium: Shell: ask for confirmation for node fence and clearstate (bnc#642268)
Hg: Merge with 1.1
Low: Shell: insert sleep between stop and start in resource restart
High: Shell: acl support
Low: stonith: remove unnecessary logging
Medium: Shell: fix exception when HOME is not set
Medium: Shell: fix rsc/op_defaults edit
Low: Shell: update regression tests expect files
Medium: Shell: repair management of cloned groups
Medium: Shell: wait for the stop transition to finish before invoking start in the resource restart command (lf#2516)
merge with 1.1
High: Shell: support for ACL xpath shortcuts
Medium: Shell: allow adding node attributes in configure when the attributes set doesn't already exist
High: Shell: ACL: enable multiple role references
Dev: Shell: ACL: update location xpath and allow multiple id references
Dev: Shell: ACL: add cib shortcut and explain why it's necessary
High: Shell: don't create empty optional sections (bnc#665131)
Dev: Shell: work around lrmd/lrmadmin blocking when not run as root or hacluster
Dev: Shell: work around lrmd/lrmadmin blocking when not run as root or hacluster
Medium: Shell: more lazy help system initialization
Low: Shell: fix logic for not printing the default monitor depth
Low: Shell: fix ra regression test to match output with missing default monitor
Low: Shell: add acl regression test
Medium: doc: man page for crm
Low: Shell: print error instead of warning and exit with 1 if a resource cannot be deleted (bnc#680401)
Dev: build: creating crm.8 depends on asciidoc
Low: Shell: don't bail out on bad rsc type specification
High: Shell: repair template list command
Dev: Shell: really fix template list
High: Shell: implement -w,--wait option to wait for the transition to finish
Dev: Shell: remove empty lines to match the existing style
Low: Shell: don't throw exception if the RA meta-data does not contain unique attribute for all parameters
Low: Shell: fix output of ra regression test (stonith:external/ssh)
Medium: Shell: invoke cibadmin with --force when requested by user
Low: doc: exec-time is in milliseconds
Low: Shell: improve logic for when to use the pager
Low: Shell: don't print error on Broken pipe
High: Shell: set of commands to examine logs, reports, etc
Dev: Shell: spawn transition command from peinputs
Dominik Klein (4):
Low: Tools: crm_mon now display the correct node status in html mode
RA: pingd: Allow native setting of interval, ping_attempts and ping_timeout
High: RA: pingd: Set default ping interval to 1 instead of 0 seconds
Low: extra/showscores.sh Correctly display stickiness if unset
Federica Teodori (1):
Medium: Shell: print warning if start or stop interval is not zero
Florian Haas (14):
Medium: build: service_crm.so is Corosync only, don't try to remove it if it wasn't built
Medium: build: enable --without heartbeat and --without ais as advertised
Low: build: introduce "--without esmtp" rpmbuild option
Medium: build: add compatibility wrappers for conditional builds in legacy RPM versions
Dev: build: tools/crm.in is gone, remove it from autoconf
Low: build: crm shell depends on python >= 2.4
Low: build: add --without snmp flag to RPM spec
Low: build: fix Requires for libesmtp
Low: build: add conditional build flags to %configure in RPM spec
Medium: shell: find hb_delnode in correct directory
xml: allow upstart primitives in resources.rng
Low: Tools: fix typos in crm_resource --help
Medium: Clusters From Scratch: fix device param in Filesystem resource
Medium: mcp: Bug cf#5024 - set environment variables expected by ocf-shellfuncs library
Frank DiMeo (1):
fix refresh time for html interface of crm_mon
Gao,Yan (71):
High: PE: Before migrating an utilization-using resource to a node, take off the load which will no longer run there (lf#2599, bnc#695440)
Low: PE: Update regression tests (lf#2599, bnc#695440)
High: PE: Before migrating an utilization-using resource to a node, take off the load which will no longer run there (regression tests) (lf#2599, bnc#695440)
Medium: PE: Support of monitor op with role="Stopped"
Low: PE: Support of monitor op with role="Stopped" (regression tests)
Low: PE: Update regression tests
Low: cib: Record and display the originator details of CIB changes for schema "pacemaker-1.2"
Low: Build: Install pkg-config files
Medium: PE: Bug lf#2606 - A demote operation is implied by the stonith operation
Low: PE: Bug lf#2606 - A demote operation is implied by the stonith operation (regression test)
Low: PE: Update regression test (lf#2606)
High: PE: Implement cluster ticket and deadman
Low: PE: Implement cluster ticket and deadman (regression tests)
Low: PE: Implement cluster ticket and deadman (regression tests for rsc_ticket with resource_set)
Low: Build: Add crm_ticket into spec file
Low: Tools: crm_attribute - No need to determine the node for tickets management
Low: Tools: CLI regression tests for crm_ticket
Medium: Tools: crm_mon crm_resource - Display the "operation_key" instead of the operation "id" if available (bnc#707201)
High: PE: Implement resource template
Low: PE: Implement resource template (regression tests)
High: PE: Improve the merging with template's operations
Low: PE: Improve the merging with template's operations (regression test)
Low: PE: Set the name of a xml node in the better way - Fix memory leak
Medium: Tools: crm_attribute - Update tickets directly to cib instead of through attrd
Medium: cib: Fix compilation of ACL code
High: PE: Implement referencing templates in constraints
Low: PE: Implement referencing templates in constraints (regression tests)
Medium: PE: A colocation constraint between two templates makes no sense
High: PE: Implement limit number of concurrent live migrations
Medium: crmd: Send out all of the meta parameters to lrmd for stop actions
Medium: RA: Fix the metadata of HealthSMART and SysInfo RAs
High: stonith: Fix the stonith timeout issue (cl#5009, bnc#727498)
Medium: stonith: Longer proposed timeout/interval of operations for stonith resources
Meidum: Core: Move several new fields since v1.1.6 to the end of the data types for ABI compliance
Low: stonith: Default to 20s timeout for start and monitor operations of stonith resources which is same as "default-action-timeout" (bnc#733337)
Medium: Tools: ptest - Fix "-X, --xml-text" option to handle a XML string as its argument
Medium: cib: Bug cl#5023 - Fix segfault caused by destroying an already freed hash table
High: ais: Bug cl#5027 - Conditionally use qb_log_from_external_source()
High: PE: Support of referencing resource templates in resource sets
Low: PE: Support of referencing resource templates in resource sets (regression tests)
High: stonith: A "start" operation of a stonith resource does a "monitor" on the device beyond registering it
Low: PE: Comments on the code for support of referencing resource templates in resource sets
Low: PE: Get XML document via getDocPtr()
Low: PE: More comments on the code for support of referencing resource templates in resource sets
Low: PE: Drop the unnecessary re-setting of variables
Low: PE: More validations on configurations related to resource templates
Low: PE: Logging tweaks
Medium: xml: Add rsc_ticket support into "pacemaker-1.2" schema
Low: doc: Documentation on "Multi-site Clusters and Tickets"
Medium: xml: Add resource template support into "pacemaker-1.2" schema
Low: doc: Documentation on "Resource Templates"
Low: doc: Minor revisions of documentation
Low: doc: Document the "migration-limit" cluster option
Low: doc: Revise of "Multi-site Clusters and Tickets" chapter
Low: doc: A minor revision of "Utilization and Placement Strategy" chapter
Low: doc: Fix a mail address
Low: doc: More description in "Multi-site Clusters and Tickets" chapter
Medium: Tools: crm_simulate - Implement the support of granting/revoking tickets for crm_simulate
Medium: Tools: crm_ticket - Granting/revoking tickets need user to confirm unless "--force" option is used
High: PE: Move the ticket state section to "/cib/status/tickets/ticket_state"
High: PE: Support to make tickets standby for relinquishing tickets gracefully
High: Tools: Implement a new crm_ticket
Low: Tools: Update CLI regression tests for the new crm_ticket
Low: Tools: Add CLI regression tests for ticket standby state
Low: PE: Update regression tests for the new ticket state section
Low: PE: Support to make tickets standby for relinquishing tickets gracefully (regression tests)
Low: Tools: Add crm_ticket to be profiled by valgrind
Medium: Tools: crm_ticket - Display the tickets that are referenced in rsc_ticket constraints but have never been granted yet.
High: Tools: crm_mon - Support to display tickets (based on Yuusuke Iida's work)
High: Tools: crm_simulate - Support to grant/revoke/standby/activate tickets from the new ticket state section
High: Resolve coverity RESOURCE_LEAK, REVERSE_INULL and UNREACHABLE defects
Guochun Shi (36):
add instructions on how to configure syslog-ng to work for cts
add more instructions for syslog-ng
fixed a typo ---
implemented logging daemon
step number was differed by one in syslog-ng description
use "su -" instead of "su" to get rid of a warning message
In FastDetect test, connection loss to heartbeat is not considered an error
fixed a compiling complain in IA64
if then followed by an empty statement is syntax error
change should_send_blocking to should_send_block 'cause it's a better name
changed cl_log_send_to_logging_daemon() to cl_log_set_uselogd()
add a parameter (int flag) to msgfromIPC()
bug 433
bug 442:
define the variable at the beginning of a funciton
filename should be initialized to NULL
MAXDATASIZE is already defined in ipc.h
need import CTS
reverse the last change
make the function declaration agrees with the one in c file
fixed a typo found by Roberto Nibali <ratz at drugphish.ch>
add the section on how to set up HostbasedAuthentication
add pils library to the link option for all binaries because clplumbing will depend on pils library
add compression capability
add man page for hb_standby and hb_takeover
make "make rpm" happy
include ha_logd.cf in Makefile
I believe it is trying to print an integer :)
add getpid() to configure.in and disable the pid match check for a client
change cl_get_struct(const char* msg, ...) to cl_get_struct(char* msg, ...)
modify pidtest according David Lee's suggestion
add config directory
bug 949: this bcast message is caused by not compressing the message
add man page for hb_addnode and hb_delnode
bug 338: Make quorum architecture pluggable
remove extra "
Holger Teutsch (4):
Low: Clusters from Scratch: fix parameter consensus in corosync example
Medium: core: Don't try to print unprintable option values in crm_help
Low: Shell: Rename variable set_obj_verify to set_obj_all as it always contains all objects
Medium: Shell: Check for violations of uniqueness for instance parameters during commit
Huang Zhen (170):
fix some bugs
add get metadata
add get metadata
fix a memory leak bug
make some terms consistent with lrmd.
move the RA path from code to configure
change some log levels
add provider for resource
move common functions of RA here
fix a bug
add a '/' at the end of path
fix a bug
remove unuse function
change the multi-providers implement
fix a bug
change the types of some fields in lrm_op_t from const char* to char* so client can release the memory
fix a bug : deal with the situation that the output of metadata operation is empty
add LinuxHA version2 dependent modules
add -v2 option for LinuxHA version2
add CM_LinuxHAv2 for LinuxHA version2
rename the stop_op() to cancel_op()
modify cts for CRM
add the command string of DelFileCommCmd
Remove the OnlyForTesting file without reloading heartbeat when we setup the test
fix a bug: op_type should be ra_type in get_provider_list()
fix a bug: op_type should be ra_type in get_provider_list()
Add the code for audit resource and improve the code of StartaCM
Add the draft readme section for running cts on release 2.*(1.99.*)
fix a mistype
Andrew's patch:add a function to translate the return code of operation to string
let lrmd pass timeout to RA.
let lrmd pass timeout to RA.
merge two parttens to one
Make it work on old verson of automake
Fix a bug, add define of apigid
change core dump path
put the cib.xml into HAdir on monitor node
add ERROR and CRIT to must not list
raise an error when user does not have right to write in HAdir
convert tabs to spaces
fix some memory issues
replace tab with 8 spaces
fixed a bug.
testcase Split-brain2: support more then 2 nodes split
merge two similar patterns
rollback to previse version, as Andrew said, 'node (%s).*: is dead' is not relyable
If a node is stonithed or be splited, we need use 'node (%s).*: is dead' to determine it's dead.
to make StonithTest can be run on both 1.X and 2.X version
fixed a bug, missing define the start function
add the missing assignment of Resources
add a new testcase, ComponentFail, which will kill one component (lrmd, crmd, stonithd, heartbeat) on one node, and watch whether the cluster can recover
fix a mistype
XML_LRM_ATTR_TARGET is 'on_node' instead of 'target', and XML_LRM_ATTR_TARGET_UUID has same problem.
fix a bug
change the return value of RemoteExec.cp() to exit code of scp, which zero means ok
make RemoteExec.cp() and RemoteExec.noBlock() try three times and return the exitcode
1. move the audit classes to CTSaudits.py;
add fencing enable code to cib
allow user run CTS with only one indicated testcase, fix Bug#370
fix a bug: the cib_fencing missing define without -r option.
make usage() work.
make cts catch error messages in final stage, fix Bug#362
check the exit code of scp, related to Bug#342
When string.find() finds a pattern, it returns the position of the pattern instead of 0
make the idle longer after stonith to let the cluster stable; add a check to ResourceRecover
the strdup() causes core dump because it does not match the crm_free().
make the resources start with monitors
remove the parameter value with non-integer key, may not the final solution
Make the message better in ResourceRecover
move the local copy of cib.xml to tmp directory, fix bug 427
add the root@ back, fix bug 427
append the pid to the local copy of cib
The cib is not installed at this monent. Use CIBResource to determing whether there are resources in cluster.
add the provider of RA to cib.
use cib to test whether there have resources
change spaces in print string to tab so it looks better.
missing second as unit of time
1, change the resource name to make the log more clear; 2, remove the temp file after use it; 3,replace some heading tab with space.
change the log pattern for ResourceRecover
change the name of resource in audit
add --resource-can-stop option to make cts do audit more precisely, fix bug#461
increase the time for lrm start resource
add HasQuorum implement for v2 Cluster Manager
using HasQuorum() in HAResourceAudit
set the no quorum policy to stop
add the interval parameter back
Make ResourceRecover can not choose DcIPaddr temporarily, refer to #460 for details
add xmit and recv loss rate to CTS
add reduce and restore the reliability of communications commands
add PacketLoss scenario component
move the communicate commands to CM_hb.py
modify the split-brain tests to fit the PacketLoss scenario component
add SaveCommCmd command
add some communication control functions to CM
using the new communication control functions in split-brain tests
add check before fix the communications
Change back so ResourceRecover can select DcIPaddr
Fix a bug, the HBConfig did not use the directory indicated by -D option.
add LRM_OP_NOTEXEC for op_status, used for the op had not been executed yet
change LRM_OP_NOTEXEC to LRM_OP_PENDING for English way, assign -1 to LRM_OP_PENDING.
add OCF IPaddr/IPaddr2 regression test
remove one log
add a new testcase : NearQuorumPointTest(bug#554)
change the stonith resource config
fixed a log bug.
revert the last change. make stonitch resource running on each nodes
dont return the resorces in incarnations
make CTS can deal with incarnations
get incarnation_max from instance_attributes instead of attriubtes. add some fields for audit incarnations
add some code for incarnation resource audit
fix a bug in log
only two paramters are needed.
Stonith resources need to be monitored for ResourceRecover test
add Stonithd as a new ResetMechanism for cts -- by James Pan
add new testcase StonithdTest by James Pan
the timeout unit is ms. so 10000 is 10 ten seconds.
add Component and Process to ClusterManager, used for ComponentFail test
implement Components() in CM_LinuxHAv2
modify ComponentFail test according to the changes in CM
disable ComponentFail temporarily, enable it when it stable enough
we can set the base ip address for test by --test-ip-base XXX.XXX.XXX.XXX now
convert haresources file to cib. initial version
add haresources2cib.py
toprettyxml will use " instead of \'
add some crm configs
avoid find bad news recursive
Only audit the number of incarnation when we have quorum
make the default log file same as in the Env
add is_managed parameter to resource
make the file logger has the same format as syslog
fix bug, use rsc.rid in log
make the cluster do not restart the normal shutdown node
make suppress cib as an option of CTS
convert several heartbeat resources to ocf ones and add monitor to resources. fix #807
make ocf ras have different monitor values
the proxy for remote python call
add remote python call feature
make sure the cluster is healthy before we kill something
add list tests option
add printing status of fencing
1. add the support for the format: IPaddr2::192.168.1.11/24/eth0
add exception and signal process to CTS. it will print summary when CTS is interrupted or get 10 signal
add audit of resource groups
add ResourceGroups() to CM_LinuxHAv2
use ResourceGroups() of CM_LinuxHAv2 in audit
add ResourceGroups() so we don't need return pair
add management tool directories
move to tools/
move haresources2cib.py.in to tools/
move haresources2cib.py.in from cts/ to tools/
put the haresources2cib.py to heartbeat directory
add support for lsb RA
convert more heartbeat RAs to OCF RAs
add the stonith support.
replace the deprecated whrandom with random; disable the warnings of os.tmpnam()
fix a typo error and the bug with empty line in resource file. thanks Sam Tran found them and gave the patch
make lrmadmin return everytime to avoid block
sometime the build dir is not the source dir, so we need indicate the source dir
add reminder to user about the lcoation and permission of cib.xml
generate the cib in the final place with right permission
fix a bug. stat need full path as parameter
change hasbin_PYTHON to hasbin_SCRIPTS
change hasbin_PYTHON to hasbin_SCRIPTS
change hasbin_PYTHON to hasbin_SCRIPTS
pull from Andrew's repository
add new plugin set for quorumd
pull from hg.linux-ha.org/dev
dont build quorumd if gnutls is not available
LRM: use heartbeat as the default provider of lsb RA
OSDL # 1415 TOOLS: use the new style crm attribute names to replace old style name
Jia Ming Pan (42):
DO_STANDBY can never be reached before.
"name" and "value" are not used here, remove them.
Add readlines.
Add StandbyTest2
use xml parsing instead of regrex match in test case StandbyTest2
StonithdTest2 prints complaints message when stonithd operation fail
check return values for Stonithd
- check apitest's result to see whether stonithd operation succeed.
1. add exceptions
move ClusterManager dependent code to class LinuxHAv2
move ClusterManager dependent code to class LinuxHAv2, thx to zhenh
added StonithdTest requirements
standby2 reports failure and returns imediately after it fails to get node uuid
print more information for bug700
only need the nodes section
should not call self.CM.cluster_stable() before waiting the node to reboot
added None as the default value of timeout
added mini howtos for StonithdTest
added Copyright & License (bug# 668)
added license statements, bug # 668
added license statements, bug # 668
added license statements
fixed bug # 980
moved the CIB data from CM_LinuxHAv2.py.in into CIB.py.in
removed deprecated module whrandom
Added option seed for user specified random seed, bug # 1125
Make sure machine is up before ssh over it, bug #1109 & #1124
just made a mistake, rolled back to the previouse version
Added option seed for user specified random seed, bug # 1125
Should be SysLog
added 2 dummy RAs for CTS testing
the original RA (OCFDummy2) keeps resource instance status in a file,
install testing RAs to cts directory, and use OCFMSDummy instead of OCFDummy2
1,better indent. 2,added lsb&master-slave resources, in full pathname
use OCFMSDummy instead
initialize nvparis before use
removed option --resource-class
replaced base64 with binascii to avoid '\n' in encoded strings
we can't ensure the uid on the monitor machin euqals the uid on the test machin for the same user, bug #1147
avoid file operations; use ocf_log instead of built-in log
wait until node is booted and sshd is up
added options for TSA plugin
Jiaju Zhang (1):
Small fixes to the controld RA
Jonathan Süssemilch Poulain (1):
Medium: Tools: crm_mon - Enable specification of community for SNMP traps
Keisuke MORI (4):
High: Core: fix memory leaks exposed by valgrind
High: Tools: crm_mon - fix memory leaks exposed by valgrind
High: PE: Fix memory leaks exposed by valgrind
[mq]: glib2.patch
Lars Ellenberg (10):
fix warning 'integer from pointer without a cast'
make --enable-ansi work on linux
-pedantic-errors fixes take one:
-pedantic-errors fixes 2:
Checkin of the DRBD Outdate Peer Daemon plugin.
move drbd-peer-outdater into /usr/lib/heartbeat,
add configure option for the drbd outdate peer daemon
moved drbd peer outdater into contrib/
Build: allow build to succeed --without-corosync
Medium: after signing in with heartbeat, change the heartbeat side ipc send queue len
Lars Marowsky-Bree (187):
Initial check-in of the few CRM design thoughts and notes available, so
First version of the diagram showing the most significant data flows in
CRM now unstoppable as spectacular flowchart diagram wins the heart of
Some editorial updates only.
Initial CRM merge. Not compiled in by default, nothing changes if not
Text reformatted to be readable.
Text-formatting cleanup.
More text cleanups.
CRM cleanups by Andrew.
Type fix.
Cleanup skeleton functions to use the IPC code more consistently.
Pet peeve removal: _Id et al now gone, replaced with consistent Id header.
Merge SWIG/Perl5 bindings.
Initial commit of Perl based pengine, likely breaks tons of stuff ;-)
Sign-in at least and make crmd crash... ;)
Strip whitespace from XML we sent to make libxml2 more happy.
Don't immediately die. And send the hello message correctly.
Signs in, doesn't crash, receives hello message from crmd - SHIP!
Add -S: Read XML on stdin and output reply to stdout.
Fix oversight in Perl bindings: Export ch_status.
Examples for pengine input.
Handle quit message.
CIB DTD in more detail and with a few changes. Still in flux.
Commit pending global todos from editor too ;-)
Allow instance parameters to be overridden per-node/attribute.
cib-example-1.xml is an example configuration which validates against
Initial outline (with tons of gaps, so many I didn't even mention them
This is the combined DTD for all XML the CRM handles.
Cleanups and extensions to the Transition Graph, now defines the syntax
Make the example configuration validate against the DTD.
Files have been consolidated into crm-1.0.dtd
Part I: Lower-case hostnames whereever they are coming in. STONITH
Initial attempt / suggestion at how to merge the msging scheme into the
Some more thoughts.
Make verbose less verbose. Attribute names do not have to be globally
Remove health attribute. We need to have more thought here, but likely
Couldn't resist a useless-use-of-cat cleanup ;)
Add document type declaration to all testcases.
Typo fix.
Use correct port number in example.
Further cleanups.
Add some additional required mandatory OCF environment variables.
Path to OCF_ROOT fixed.
- Define PIL_PLUGIN_BOILERPLATE() as it used to be, which implies a
Fix minor oversight: A group contains one or more resources, not just
64bit pointer issue.
Janitorial work: Stray \n removal.
Remove screw-up.
gcc4:
nvpairs need an id tag.
libltdl.tar.gz -> libltdl.tar
Compiler warnings fixed: Comparing floats with != and ==.
Phase of the moon constraints.
Add phase of the moon to the dtd too
Fix lack of prototype.
Revert lrm.c to 1.141 prior to 2.0.2 release.
Fix CIB feature revision to match master/slave.
Tyop
Fix DTD.
transient_attributes does have an id attribute.
Create varlibdir for pengine.
Revert, wrong Makefile.
Minor correction to the help text.
Querying the health of the cluster isn't yet implemented, so don't
Requesting the name of all resources not implemented either (#163563).
Novell bz #163933: Help did not match the actual options in the code.
Compile-time warning about unused variable if CRM_DEPRECATED_SINCE_2_0_4
Tyop.
Assigned the same id as the resource to the attribute set; prefix with
Off by three orders of magnitude ;-)
Temporary work-around for #1278: Allow rsc_location w/o rules into the
Further relax DTD validation in an effort to work-around #1278.
Andrew is right, we shouldn't have this work-around in CVS.
Work-around bug #1287: GUI currently requires empty groups to be
The GUI creates meaningless placeholders first, which only in a second
Work-around bug #1287: Allow empty groups, but warn.
Small formatting fix.
Explain some more commands & options.
NodeStatus should use the RemoteExec class too, otherwise BSC might hang
Make #1162 reproducible again, so the memory leak can be debugged.
Fix BSC/CTS - CTSproxy.py needs to be a script.
Fix copyright notice.
Fix copyright notice.
Done.
resource_stickiness attribute on groups accepted not all possible values
Merge with crm-stable
Correct use of strncat(): Third parameter is remaining free space, not
Use CFLAGS when building ccdv to make a warning from the build system go
Merge crm-stable into dev.
OSDL 1452 addition: Fix DTD syntax.
Remove stray space from output, pointed out by Christian Froemmel.
Merge crm-stable with dev.
Typos in comments and debug logs. (Sorry, I saw these during review and
crmd: crm_str_ -> safe_str_
Remove uninitialized variable warning on line 116.
Remove a bunch of dead code (Perl/SWIG bindings).
Remove CVS artifacts. ($Id$ and $Log$ have no meaning to mercurial.)
crm_attribute/master/standby: Always default the uname to the current
CIB: Downgrade one log message (Deallocating the CIB.)
Part of the reversion of a previous changeset got stuck in here. Doh.
Eligable is not a word ;-)
ocf-tester: Continue if validate-all succeeds.
Build: Restore build dependencies. (Reverts http://hg.linux-ha.org/dev/rev/27394b5c56e3)
Merge dev changes into a single head again.
ocf-tester: Exit with appropriate exit code for success or failure.
ocf-tester: Various integers were compared as strings.
Correct a comment.
Merge dev with local changes.
Merge eDir88 changes with dev.
CTS: Advertise the option to add resources to the CIB.
Complete Andrew's removal of autoconf-2.53.diff
lrm: stonith: fix spelling in a log message and lower log level.
lrm: stonith: fix spelling in a log message and lower log level.
CTS: Fix a path reference to LSBDummy, so that CTS has a chance of working.
CTS: Pattern too restrictive. (LF 1725)
CTS: More ComponentFail patterns which were too restrictive for ProcTrack (LF 1725)
CTS: More patterns adjusted for exiting processes (LF 1725).
Low: Fix Makefile.am reference (ha_logd.cf -> logd.cf)
crm_attribute: Improve --help output; there is not "=" required for the -t option.
Merging local changes with stable-1.0
RA: controld: minor fix to comments & printed messages.
High: crm: Avoid infinite loop during crm configure edit (bnc#480327)
Medium: Tools: crm: Start/stop clones via target-role.
High: ais plugin: Fix quorum calculation (bnc#487003)
Low: crm_mon: make clear quorum is not tied to a particular node.
Low: crm: initial version of a regression test (not installed yet)
Merge stable-1.0 with local changeset.
Medium: RA: controld: Return proper exit code for stop op.
Low: pengine: always provide notify, ordered, and interleave values to RAs.
Low: pengine: Revert 78c383c19453 (addition of notify, interleave, and ordered)
Med: xml: Mask the "symmetrical" attribute on rsc_colocation constraints (bnc#540672)
Low: PE: remove left-over debugging log (bnc#546353)
Merge local changes with upstream.
Merge local changes with upstream.
Low: controld RA: remove debugging log message.
Merge local changes with upstream.
High: shell: Add allow-migrate as allowed meta-attribute (bnc#539968)
Low: crm: Enhance "colocation" help to explain how it works.
Low: crm_mon: Mail subject needs space for \0 byte.
XML: Low: Do not warn if the configuration is fully forward-compatible.
Merge 1.1 to devel.
Medium: Shell: Repair "cib cibstatus op" functionality (bnc#585641)
Merge 1.1 into devel.
Build: fix manpage build.
Merge pcmk-1.1 into pcmk-devel
Low: crmd: fix non-fatal assert in lrmd.c when resources are manually failed (bnc#590924)
Merge pcmk-1.1 into devel
High: Shell: Default to using the status section from the live CIB (bnc#592762)
Merge local changes to 1.1
Merge 1.1 into devel.
Merge 1.1 into devel
Merge 1.1 into devel.
Low: Test: $USER may not be populated if not run from a login shell.
Merge devel with 1.1.
High: ais: fix list of active processes sent to clients (bnc#603685)
Low: CLI: report bad ISO8601 duration specifiers (bnc#618808)
Merge transplanted change into 1.1
High: Build: Require cluster-glue 1.0.6 for STONITH security fix (CVE-2010-2496)
Medium: Core: Enable coredumps properly for MCP and AIS plugin (lf#2463)
Low: stonith-ng: record exit code for failed port list query (bnc#625594)
Low: stonith-ng: log output for failed port list query (bnc#625594)
Merge local changes with 1.1
Medium: cib: Allow backup files to proceed after wrap-around (bnc#630322)
Low: Shell: add "clearstate" command for nodes
Low: Fix typo in node clearstate command.
Medium: ais: fix crm_strdup calls
Medium: ais: use ais_string_to_boolean instead of crm_is_true
Low: stonith: reduce some log messages and use crm_* macros
Build: Support gcov code coverage testing
Low: Build: gcov needs all inlining disabled to generate useful coverage data
Low: PE: Add regression tests for kind=(Mandatory|Optional)
Low: PE: remove dead function
Low: Build: Add support for gcov to specfile
Low: PE: Add valgrind-dhat support to regression tests
Medium: Update version to 1.1.5.
Medium: Build: Change default for ACL support to "no"
Low: Shell: Allow creation of an empty shadow CIB ("cib new ... empty")
Low: Shell: Allow cibstatus to create new node status entries.
Merge 1.1 into devel.
Merge 1.1 into devel
Merge devel into 1.1.
Low: Shell: Do not print XML for permission denied errors
High: PE: Ensure that fencing has completed for stop actions on stonith-dependent resources (lf#2551)
Medium: PE: Regression test updates (lf#2551)
Medium: Shell: node clearstate must flush LRM state too (bnc#653795)
High: ais: Handle IPC error before checking for NULL data (bnc#702907)
High: ais: check for retryable errors when dispatching AIS messages
Lon Hohberger (41):
Initial checkin of rgm2pcmk
Initial commit
Misc. changes
Fix ccs2cib script to use correct filename
Initial failoverdomain -> domain conversion
Add crm meta attributes, failover domain links
Add unit tests for:
Add resource-defaults to readme
Add utility to disable rgmanager.
Add -h option to disable_rgmanager
Add build target for disable_rgmanager
Update tests to include config_version
Add -h to ccs_flatten utility
Fix segfault on some architectures due to improper snprintf
Make ccs2cib a bit more friendly to use
Fix up output CIB to pass crm_verify
Add missing cleanup target for disable_rgmanager
Add -f option to ccs2cib
Clean up debugging messages
Add call to crm_verify prior to doing destructive things
Fix type attribute for resources
Use less ugly IDs for domain constraints
Add debug option, clean up script somewhat
Low: Doc: Fix formatting of fence_ipmilan output
Low: Doc: Minor formatting cleanups
Add basic STONITH support
Add -X and -J to help output of ccs2cib
Remove unnecessary <attribute> node
Fix nvpair for translated stonith devices
Add test6.conf
Add output config file
Add some more verbosity to what we are doing
Don't use a map for ports if the hostname is the port
Don't produce empty group resources; pacemaker doesn't like them
Make CIB output match Pacemaker's expectations
Fix bash warning if force_update is not specified
Quiet crm_verify's warning about quorum
Switch to autotools
Generate paths during 'make'.
Fix minor issues in automake config
Add source of the XML2 checks
Marcus Barrow (9):
low: doc: Scratch cleanup Introduction text
Low: doc: Scratch cleanup Installation text
Low: doc: Scratch cleanup Verification text
Low: doc: Scratch cleanup Active-Passive text
Low: doc: Scratch cleanup Active-Active text
Low: doc: Scratch cleanup Stonith text
Medium: fencing: Remove unnecessary arguments in the API
Medium: stonith: Remove Glib data types from the API arguments
Medium: stonith: Allow clients to avoid use of Glib mainloop.
Matthew Soffen (35):
Replacement function taken from FreeBSD implementation of daemon.
Replaced raexec_plugin_LDFLAGS with libs
Fixed to build on FreeBSD/Solaris
Fixed to compile on FreeBSD/Solaris(older GCC).
Created strndup function from code in glibc
Added portability.h include
Added portability.h include and moved variable declaration
Moved variable definition to start of functions.
Created packaging target for Solaris ( From David Lee)
Fixed to build out side of source tree
Fixed to compile on older GCCs - variable declarations at the start of functions
Changes for LRM and CRM to compile on FreeBSD/Solaris ( No inline definition of variables )
sys/param.h needs to be included before any project headers
sys/param.h needs to be included before any project headers
Moved difinition of file to beginning of function.
Changed from requiring portability.h to sys/param.h .
Minor issues with compiling on FreeBSD ( no inline variable initializations ).
Changed portability.h to sys/param.h - since that IS what is needed.
Removed inline var declaration
Replaced all // COMMENTs with /* COMMENT */
Fixed more comments from // to standard C comments
Fixes for compiling on FreeBSD
Changed to ansi comments
Moved where var is defined
Moved variable declarations to beginning of blocks to allow older compilers to work
Fixed make to work in a build tree separate from the source
Fixing to NOT have inline variable declaration.
Changes for building FreeBSD portfile.
Corrected files for building a port.
Fixed to compile on FreeBSD.
Added libltdl.tar to the am_config_clean target.
Replaced /usr/lib with HA_LIBDIR
Replaced /usr/lib with HA_LIBDIR
Added libtool and quiet_libtool_opts
Replaced /usr/bin/ssh with autoconf variable.
NAKAHIRA Kazutomo (1):
Low: tools: crm_node - Enable the -A parameter, the short form of --openais
Pan, Deng (1):
Make checkpointd a configurable module. Default is not configured
Philipp Marek (1):
Doc: Pacemaker Explained: Extensive style and formatting update
Sebastian Reitenbach (1):
[RA] patch to fix bashism in resources/OCF/pgsql
Simon Horman (33):
- Replaced numerous malloc + strcpy/strncpy invocations with strdup
CRM should compile now... I think I broke a few things yeserday
Distribute the .txt files in the tar ball so we don't have to rely on lynx to build releases any more
add debian package target
libtool.m4 and ltdl.m4 seem to be needed for make dist in the libltdl/ directory (sometimes). It seems harmless enough to create them
Having an all target that calls a command that is only present on Solaris breaks the build for every other OS. Changed to a target that will not be built by default. This is in keeping with how other packaging formats are handled
Fixed anomoly where cl_status was in the STABLE_1_2 branch but not the HEAD branch
I have gone over the whole tree and made it so
Fix automake 1.4-p6 build breakage
Fix automake 1.4-p6 build breakage
PIDFILE definition conflicts with definition in heartbeat.h. It doesn't seem to be used in any case
PIDFILE is defined in ../include/heartbeat.h so redefining it in Makefile.am can cause a build failure (On Debian at least). PIDFILE isn't used in the code any more so just remove it
the ccdv tool is omitted from normal cleaning,
There may be other instances of this throughout the tree,
Fix for CAN-2005-2231 temporary file vulnerabilities
Change how automake conditionals are handled to keep automake1.4 happy
Rationalis SNMP configure
Not needed since config/snmp_agent.m4 was removed
Include LGPL in tarball
Ensure that b/mgmt in make dist tarballs. Its seems that using DIST_DIRS is in conjunction with a Make conditional basically overrides the usual behaviour that directories handled by a conditional are included regardless of its state. This results in a tarball that doesn't build because configure.in expects to be able to creat lib/mgmt/Makefile.am, but lib/mgmt/ does not exist. This change does not affect weather the code in that directory is built by default or not
Make lines <= 80 characters wide
Conditional ldirectord build
* install-exec-local should probably be install-data-local
Add scripts back into EXTRA_DIST, they don't get included in the tarball othewise, by bad
Remove autoconf and friends on make distclean
[BUILD] don't ignore errors on install
[BUILD] don't remove tarbals in clean target
Low: Build: Allow build outside of top_srcdir
Low: Build: Revert crm.dtd portion of "Allow build outside of top_srcdir"
Merge with stable
low: unset CFLAGS in a dash-friendly way
Low: RA: SysInfo: use awk instead of bash to calculate memory and disk sizes
Low: remove various bashisms
Simone Gotti (2):
High: cman: Dequeue all the cman emitted events and not only the first one leaving the others in the event's queue.
High: cman: Don't call fenced_join and fenced_leave when notifying cman of a fencing event.
Sun Jiang Dong (125):
change those 'exit' at parent process to 'return'
Add the contents about how to use evlog to send test log message to the
update for new LRM implementing with new architecture
New added
make sub-directory tools be compiled
changes&fixes according to CRM's requirement
Add include <libgen.h> for compiling on OSX
Change default directory for lsb scripts
Fix the LRM compiling issue caused by Horm's change
fix compiling issues caused by Horm's change :), there should be better way to do it,pls polish it
temporarily change to avoid compiling error
add the SimulStop test case, which is referred to SimulStart
add log message
1. Fix the SimulStop test case
add missed change
polish
add RA filter
fix a bug of adding RA filter
change for moving OCF RAs location
add OCF prefix for enviroment parameters
polish the adding OCF prefix for enviroment parameters
fix a silly bug of memory leak
add a log message for tracking possible error
* Code polish, make code more portable, clear and stronger, including Lars' suggestion.
* Code polish, make it more clear, stronger
* Code polish, make it more clear, stronger
Polish code to make it more clear and stronger
polish: correct the log message or its using
polish: correct the log message
Change due to LRM internal API change, which is for passing rsc_id to OCF enviroment
* cvs ci -m "Change due to LRM internal API change, which is for passing
* Export OCF_RESOURCE_INSTANCE before executing an OCF RA
export a dummy value for the evironment variable OCF_RESOURCE_INSTANCE when try to get metadata
node fencing(stonithd) subsystem
node fencing(stonithd) subsystem
the node fencing subsystem, not stable yet
the node fencing subsystem, not stable yet
remove the redundant install hook
the node fencing subsystem, not stable yet
remove stonith related after submitting the new node fencing subsystem
the node fencing subsystem, not stable yet
update for the new node fencing subsystem
adjust the compiling order
should not return PIL_OOPS when cannot connect to stonithd
should not return PIL_OOPS when cannot connect to stonithd
Correct the mis-type from "copy-paste"
Correct the mis-type from "copy-paste"
remove the redundant install hook
donnot signon in plugin loading, just do signon/signoff in execra. Maybe a little less effecient, but stronger
donnot signon in plugin loading, just do signon/signoff in execra. Maybe a little less effecient, but stronger
Now ccdv can be auto-built but not auto-cleaned, even with old version automake. ;-)
Add the filter with the pair of patterns '### BEGIN INIT INFO' and '### END INIT INFO'
Change the mapping value of the return code of 'status' operation; add the support to 'monitor' operation via the simulation with 'status' operation
Support the 'status' operation via the simulation with 'monitor' operation
Add the return code validating and other minor polish
Add the return code validating and other minor polish
fix a bug: out of memory block border
generate meta data
Generate meta-data, but i think it is just a framework, need to discuss further
changes for returning correct RC for heartbeat RA by check its string output to stdout
changes for returning correct RC for heartbeat RA by check its string output to stdout
Bug 351: get the metadata segments from the stonith plugins
Bug 351: get the metadata segments from the stonith plugins.
Bug 351: get the metadata segments from the stonith plugins.
provide a debug info about parameters
Better log messages
open a commented log, someone need it ;-)
use a general supported operation instead
add resource instance id to the log message
Bug 475. To avoid lrmadmin to block forever
Bug 475. Increase the timeout value
comment detailed logs -- donnot need them for normal running
make a detailed log message controllable
log tweak
log tweak
Make GCC4 happy.
Use stat to make it more portable. The portbble issue found by David Lee
remove the commented code, donnot use it anymore
remove the commented code, donnot use it anymore
should set optype as a parameter even no additional parameters
bug668: license,copyright update
bug668: license update
bug668: license update
donnot fail, just ignore addtional parameters
remove printw prototype warning, done by Sun Xun <xunsun at cn.ibm.com>
configuration update for CIM providers
bug919: make the error log message clearer
update for bug834:Stonith daemon API needs per-operation user_data field
tweak log output
tweak log output
bug995: return EXECRA_NOT_RUNNING when no output
decrease the repeating of the same log
log tweak
log tweak
Replace the hardcode path with configurable ones
tweak log and correct a typo
tweak log and correct a typo
change the timeout value to be 20s, since the old value is too small sometimes
replace the hardcode path
Fix a pattern error
Bug 1222 fixing - a silly bug
1) Remove a preceded bland line in metadata
1) Remove a preceded bland line in metadata
Bug 1229: Add more metadata pieces for heartbeat resources, make it more helpful
add the line continuing symbol
Add more information to the metadata of a STONITH RA
Add more information to the metadata of a STONITH RA
format tweak
make heartbeat RAs' provider be 'heartbeat'
add a missing tag in metadat
add a missing tag in metadat
- Let the format string be a better data type
- Let the format string be a better data type
make the matching case insensitive for LSB comment keywords
metadata template tweak
metadata template tweak
relax the stickness to some LSB initscript standard; matching pattern name tweak
add more helpful information to metadata
bug1275: add signing off
bug1275: add signing off
log tweak; temporarily work around bug 1300
log tweak; temporarily work around bug 1300
bug#1300: adjust priority
bug#1300: adjust priority
fix a silly error
Tim Serong (18):
Low: tools: crm_simulate - fix small xpath memory leak in inject_node_state()
Merge with 1.1
Low: crm_simulate: Allow bringing new node online (bnc#591162)
Medium: PE: Ensure resource op timeout takes precedence over op_defaults (lf#2474)
Medium: crmd: Record completed migrate ops in CIB status section as migrate_to/migrate_from, not as stop/start (lf#2481)
Medium: PE: Update regression tests
High: Stonith: Fix two memory leaks
Medium: Shell: Let crm_resource handle node & resource iteration during resource cleanup (skips down nodes: bnc#647378)
Low: Shell: Update crm resource cleanup regression test
Low: Shell: Only import readline when run interactively (stops ".[?1034h" being emitted when $TERM=xterm)
High: Core: Repair reference to child-of-child after removal of xml_child_iter_filter from get_message_xml()
Medium: Update version to 1.1.5 (configure.ac).
High: PE: Correct target_rc for probes of inactive resources (fix regression introduced by cs:ac3f03006e95)
Low: PE: Add regression test for reprobe of inactive resources
Medium: Shell: Support "enable-acl" property
Med: Use generic USE_GHASH_COMPAT macro instead of autoconf's HAVE_LIBGLIB_2_0
Low: doc: Add ACL documentation
Medium: Core: Use pre-glib2-2.28 version of g_str_hash to ensure consistent score calculation for clone instances
Vladislav Bogdanov (1):
Low: agents: Add ifspeed resource agent
Xun Sun (1):
popular "the the" typo
Yan Gao (54):
Dev: PE: Implement more resource placement strategies: utilization, minimal and balanced
Low: PE: Add regression tests for the new placement strategies
Low: PE: Repair the file name extensions for several test scores files
Low: cib: Check also whether an user's primary group is matched
Low: PE: Replace g_hash_table_iter_*() with g_hash_table_foreach() (lf#2351)
Dev: Tools: Add utilization support for crm_attribute and crm_resource
Dev: PE: (bnc#589873) Show utilization/capacity information
Low: ais: Convert endian for all of the integer fields in messages (bnc#472014)
Low: PE: Use crm_free() for destroying temporary text
Low: Tools: Do not print help on executing "crm_node -i"
Low: Core: Prevent cl_log() from overriding the errno set in crm_int_helper() under a chroot environment (lf#2407)
Medium: PE: Before starting utilization-using resources on a node, take off the load which will no longer run there (bnc#594782)
Low: PE: (Regression tests) Before starting utilization-using resources on a node, take off the load which will no longer run there (bnc#594782)
Low: PE: Update utilization enabled tests
High: crmd: Prevent segmentation fault
Medium: PE: Support clones in resource sets (bnc#586613, lf#2404, lf#2409)
Low: PE: Update regression tests for resource sets support
Low: PE: Support clones in resource sets (bnc#586613, lf#2404, lf#2409) (regression tests)
Low: Build: Define AIS_COROSYNC and SUPPORT_AIS with an existing variable
Dev: cib: Implement multi-level ACLs for the CIB
Hg: Merge with 1.1
Low: cib: Determine a shadow directory according to the uid
Low: cib: HOME takes precedence over the password entry on determining a shadow directory
High: cib: Copy the cib_user field for ACL support on forwarding cib requests to DC
Low: Build: Add conditional compilation for CIB ACL support
High: cib: Allow requests without cib_user field
Low: xml: Move the schema for ACL support to pacemaker-1.1
Low: ACL: Convert uid directly with strtol() for avoiding unnecessary error logs
High: cib: Generate the original format of cib diff for ACL processing
Dev: cib: Add "enable-acl" property for CIB
Dev: cib: Read CIB properties on startup and update them whenever changes
Low: cib: Move initialization works to the back of the option processing
Low: cib: Revert b01804905892 and simply change the arguments of crm_log_init()
Low: cib: Change the function name "crm_config_changed()" to "cib_internal_config_changed()"
Dev: cib: Support multiple role references in ACLs
Medium: cib: ACL: Grant the write permission to an object that only has "id" attribute and whose children in diff are allowed
Dev: cib: ACL: Support delegated access
Low: Build: Conditionally compile determine_request_user()
Medium: cib: Add the delegated API for shadow cib (bnc#664985)
Medium: xml: Add ACL support into pacemaker-1.2 schema
Medium: ACL: Remove the support of numeric UIDs
Low: xml: Change acl_user's "id" to "ID" type
Dev: ACL: Support access control on "cleanup"; Support "cib_dryrun" call option
Medium: crmd: ACL: Prohibit non-privileged users to "fail" resources
Low: Build: Conditionally compile the previous commit
Dev: cib: Record and display the originator details of CIB changes
Medium: cib: Update "cib-last-written" for live CIB on a config change
Low: cib: Filter out unnecessary "update-*" fields on digest calculating
High: cib: Check the validation version before adding the originator details of a CIB change
Hg: Merge 1.1 into devel
Medium: PE: Optimize the placement strategy (bnc#669686)
Low: PE: Update regression tests (bnc#669686)
Low: PE: Optimize the placement strategy (regression tests) (bnc#669686)
Medium: PE: Indicate resource failures even if on-fail="ignore"
Yuusuke IIDA (2):
Medium: Shell: allow adding force parameter to resource migrate
Medium: Tools: crm_simulate: really use the operation outcome value supplied with -i
Yuusuke Iida (3):
Low: mcp: Changed an environment variable. To the thing which is really used.
Low: PE: changed a pre-fix to refer to a right value.
High: crmd: Correctly determine when a node can suicide with fencing
Zhao, Forrest (1):
check in the AIS event service code
Zhao, Kai (2):
add class ResourceRecover() testcase by Zhao Kai and add RunningNodes() function by Huang Zhen
move ResourceRecover class from CM_linuxHAv2.py.in to CTStests.py.in
Zhu Yi (2):
telecom depends on membership
cms (ais message service) code initial checkin
Zou Yixiong (4):
Disable building eventd for now.
Adding a --enable-eventd option to configure.in. The Eventd is not
a fix for "make rpm" error caused by empty directories.
Fix the "make rpm" issue for cms.
dvossel (2):
fixes build error
Removes error in doc Makefile.am
misch at multinet.de (1):
Low: extras: HealthSMART: metadata should always work (lf#2269)
nozawat (1):
Correction to exact sed scripts
renayama19661014 at ybb.ne.jp (1):
Low: Shell: error on unknown method in configure load
root (1):
Low: cts: Log badnews when rsyslog rate limiting occurs.
xwhu at novell.com (2):
Low: Shell: templates: add cmirror to ocfs2 and clvm
Low: Shell: templates: add cmirror to ocfs2 and clvm
-----------------------------------------------------------------------
No new revisions were added by this update.
--
Alioth's /usr/local/bin/git-commit-notice on /srv/git.debian.org/git/debian-ha/pacemaker.git
More information about the Debian-HA-Commits
mailing list