Bug#584145: multipath-tools: udevd-work complains about device name differences

Ferenc Wagner wferi at niif.hu
Tue Jun 1 20:01:59 UTC 2010 

Guido Günther <agx at sigxcpu.org> writes:

> Hi Ferenc,
> On Tue, Jun 01, 2010 at 08:02:29PM +0200, Ferenc Wagner wrote:
>> Package: multipath-tools
>> Version: 0.4.8+git0.761c66f-8
>> Severity: normal
>> 
>> Hi,
>> 
>> After upgrading a system to current unstable, during reboot I got a lot
>> of warnings like:
>> 
>> udevd-work[161]: kernel-provided name 'dm-0' and NAME= 'mapper/SIBM-ESXSST936701LC____F3LC065GL00007604NMST' disagree, please use SYMLINK+= or change the kernel to provide the proper name
>> udevd-work[162]: kernel-provided name 'dm-4' and NAME= 'mapper/pws_www' disagree, please use SYMLINK+= or change the kernel to provide the proper name
>> udevd-work[494]: kernel-provided name 'dm-6' and NAME= 'mapper/pws_sql' disagree, please use SYMLINK+= or change the kernel to provide the proper name
>> udevd-work[161]: kernel-provided name 'dm-1' and NAME= 'mapper/nfs' disagree, please use SYMLINK+= or change the kernel to provide the proper name
>> udevd-work[474]: kernel-provided name 'dm-5' and NAME= 'mapper/360060160f4512000de9d9b74090cdf11' disagree, please use SYMLINK+= or change the kernel to provide the proper name
>> udevd-work[161]: kernel-provided name 'dm-7' and NAME= 'mapper/360060160f4512000de9d9b74090cdf11-part1' disagree, please use SYMLINK+= or change the kernel to provide the proper name
>> udevd-work[560]: kernel-provided name 'dm-10' and NAME= 'mapper/nfs-part2' disagree, please use SYMLINK+= or change the kernel to provide the proper name
>> udevd-work[162]: kernel-provided name 'dm-8' and NAME= 'mapper/360060160f4512000de9d9b74090cdf11-part2' disagree, please use SYMLINK+= or change the kernel to provide the proper name
>> udevd-work[494]: kernel-provided name 'dm-9' and NAME= 'mapper/nfs-part1' disagree, please use SYMLINK+= or change the kernel to provide the proper name
>> udevd-work[162]: kernel-provided name 'dm-8' and NAME= 'mapper/360060160f4512000de9d9b74090cdf11-part2' disagree, please use SYMLINK+= or change the kernel to provide the proper name
>> udevd-work[161]: kernel-provided name 'dm-7' and NAME= 'mapper/360060160f4512000de9d9b74090cdf11-part1' disagree, please use SYMLINK+= or change the kernel to provide the proper name
>> udevd-work[161]: kernel-provided name 'dm-11' and NAME= 'mapper/nfs-root' disagree, please use SYMLINK+= or change the kernel to provide the proper name
>> udevd-work[161]: kernel-provided name 'dm-12' and NAME= 'mapper/nfs-swap' disagree, please use SYMLINK+= or change the kernel to provide the proper name
>> 
>> Part (or all) of this may be out of scope for multipath (eg.
>> /dev/mapper/nfs-{root,swap} are LVM2 logical volumes), but the
>> problem is present here as well.  The landscape is quite mixed,
>> the /dev/dm-[23] links are even broken because of some space vs.
>> underscore discrepancy:
>> 
>> $ ls -l /dev/mapper
>> total 0
>> brw-rw---- 1 root root 254,  5 Jun  1 19:17 360060160f4512000de9d9b74090cdf11
>> brw-rw---- 1 root disk 254,  7 Jun  1 19:19 360060160f4512000de9d9b74090cdf11-part1
>> brw-rw---- 1 root disk 254,  8 Jun  1 19:19 360060160f4512000de9d9b74090cdf11-part2
>> crw-rw---- 1 root root  10, 58 Jun  1 19:19 control
>> brw-rw---- 1 root root 254,  1 Jun  1 19:17 nfs
>> brw-rw---- 1 root disk 254, 17 Jun  1 19:19 nfs-export
>> brw-rw---- 1 root disk 254, 16 Jun  1 19:19 nfs-home
>> brw-rw---- 1 root disk 254,  9 Jun  1 19:19 nfs-part1
>> brw-rw---- 1 root disk 254, 10 Jun  1 19:19 nfs-part2
>> brw-rw---- 1 root root 254, 11 Jun  1 19:17 nfs-root
>> brw-rw---- 1 root root 254, 12 Jun  1 19:17 nfs-swap
>> brw-rw---- 1 root disk 254, 13 Jun  1 19:19 nfs-tmp
>> brw-rw---- 1 root disk 254, 14 Jun  1 19:19 nfs-usr
>> brw-rw---- 1 root disk 254, 15 Jun  1 19:19 nfs-var
>> brw-rw---- 1 root root 254,  6 Jun  1 19:17 pws_sql
>> brw-rw---- 1 root root 254,  4 Jun  1 19:17 pws_www
>> brw-rw---- 1 root disk 254,  0 Jun  1 19:17 SIBM-ESXSST936701LC    F3LC065GL00007604NMST
>> brw-rw---- 1 root root 254,  2 Jun  1 19:17 SIBM-ESXSST936701LC    F3LC065GL00007604NMST-part1
>> brw-rw---- 1 root root 254,  3 Jun  1 19:17 SIBM-ESXSST936701LC    F3LC065GL00007604NMST-part2
>> brw-rw---- 1 root disk 254, 18 Jun  1 19:19 VG_sql-sql
>> brw-rw---- 1 root disk 254, 19 Jun  1 19:19 VG_www-www
>> $ ls -l /dev/dm*
>> brw-rw---- 1 root disk 254,  0 Jun  1 19:19 /dev/dm-0
>> brw-rw---- 1 root disk 254,  1 Jun  1 19:19 /dev/dm-1
>> lrwxrwxrwx 1 root root      16 Jun  1 19:19 /dev/dm-10 -> mapper/nfs-part2
>> brw-rw---- 1 root disk 254, 11 Jun  1 19:19 /dev/dm-11
>> brw-rw---- 1 root disk 254, 12 Jun  1 19:19 /dev/dm-12
>> lrwxrwxrwx 1 root root      14 Jun  1 19:19 /dev/dm-13 -> mapper/nfs-tmp
>> lrwxrwxrwx 1 root root      14 Jun  1 19:19 /dev/dm-14 -> mapper/nfs-usr
>> lrwxrwxrwx 1 root root      14 Jun  1 19:19 /dev/dm-15 -> mapper/nfs-var
>> lrwxrwxrwx 1 root root      15 Jun  1 19:19 /dev/dm-16 -> mapper/nfs-home
>> lrwxrwxrwx 1 root root      17 Jun  1 19:19 /dev/dm-17 -> mapper/nfs-export
>> lrwxrwxrwx 1 root root      17 Jun  1 19:19 /dev/dm-18 -> mapper/VG_sql-sql
>> lrwxrwxrwx 1 root root      17 Jun  1 19:19 /dev/dm-19 -> mapper/VG_www-www
>> lrwxrwxrwx 1 root root      57 Jun  1 19:19 /dev/dm-2 -> mapper/SIBM-ESXSST936701LC____F3LC065GL00007604NMST-part1
>> lrwxrwxrwx 1 root root      57 Jun  1 19:19 /dev/dm-3 -> mapper/SIBM-ESXSST936701LC____F3LC065GL00007604NMST-part2
>> brw-rw---- 1 root disk 254,  4 Jun  1 19:19 /dev/dm-4
>> brw-rw---- 1 root disk 254,  5 Jun  1 19:19 /dev/dm-5
>> brw-rw---- 1 root disk 254,  6 Jun  1 19:19 /dev/dm-6
>> lrwxrwxrwx 1 root root      46 Jun  1 19:19 /dev/dm-7 -> mapper/360060160f4512000de9d9b74090cdf11-part1
>> lrwxrwxrwx 1 root root      46 Jun  1 19:19 /dev/dm-8 -> mapper/360060160f4512000de9d9b74090cdf11-part2
>> lrwxrwxrwx 1 root root      16 Jun  1 19:19 /dev/dm-9 -> mapper/nfs-part1
>> 
> [..snip..] 
>
>> Something is wrong or at least transitioning here, I don't see the full
>> picture.  Basically, the system seems to work (I don't use the local
>> disk), though.
>
> This is broken due to changes in udev:
>
> http://lwn.net/Articles/387384/
>
> the warnings are therefore harmless. cc'ing the udev maintainer. Marco,
> what's the way forward with this? Shall the /dev/mapper/ names become
> symlinks to /dev/dm-X?

That probably means LVM is affected as well (Cc added).  I wonder if
reversing the links will play nice with root=/dev/mapper/vg-rootlv...

Anyway, the spaces should really go, and the partition devices should
probably have the same ownership as the corresponding whole disk device.
-- 
Cheers,
Feri.

More information about the pkg-lvm-maintainers mailing list