Bug#659762: lvm2: LVM commands freeze after snapshot delete fails

[C. R. Meloche - CREDIL]

On 10/07/2012 07:33 AM, Paul LeoNerd Evans wrote:
> On Tue, Feb 14, 2012 at 06:20:57PM +0000, Paul LeoNerd Evans wrote:
>> On Mon, Feb 13, 2012 at 05:40:34PM +0000, Alasdair G Kergon wrote:
>>> dmsetup info -c
>>> dmsetup table
>>> dmsetup status
>>>
>>> will show the current device-mapper state
> It appears to have happened again:
>
> root at cel:~
> # lvremove --force vg_cel/backups-20120117
>    /dev/vg_cel/backups-20120117: read failed after 0 of 4096 at 161061208064: Input/output error
>    /dev/vg_cel/backups-20120117: read failed after 0 of 4096 at 161061265408: Input/output error
>    /dev/vg_cel/backups-20120117: read failed after 0 of 4096 at 0: Input/output error
>    /dev/vg_cel/backups-20120117: read failed after 0 of 4096 at 4096: Input/output error
>    Unable to deactivate open vg_cel-backups--20120117-cow (254:31)
>    Failed to resume backups-20120117.
> libdevmapper exiting with 3 device(s) still suspended.
I have the same problem as well. After reboot I would be able to finally 
delete the device. This device would no longer be a snapshot by then. I 
noted that if the device is mirrored (lvm mirror) I get the problem 
every time. However, if the device is not mirrored, for instance I tried 
with swap files that are not, then I do not get the problem. So I 
mirrored the entire server and I was able to do my backups. Note that I 
was unable to get out of this using the dmsetup resume idea as described 
before, my device do not seem to get suspended.

I have a freshly installed Ubuntu server in operation, I'll get to test 
on Tuesday to see if I get the same problem. If I do not I'll report the 
different versions of software & kernel used then. Do we know if the 
problem is still there in sid?   In the mean time I may have to 
re-install my server using md mirrored raid and lvm on top instead.


>
> root at cel:~
> # lvs
>
> [at this point the process is hard dead, no SIGTERM etc.. can wake it]
>
> [on another terminal...]
>
> root at cel:~
> # ps -lp 26009
> F S   UID   PID  PPID  C PRI  NI ADDR SZ WCHAN  TTY          TIME CMD
> 4 D     0 26009 25789  0  80   0 -  6384 ?      pts/9    00:00:00 lvs
>
> root at cel:~
> # kill -9 26009
>
> root at cel:~
> # ps -lp 26009
> F S   UID   PID  PPID  C PRI  NI ADDR SZ WCHAN  TTY          TIME CMD
> 4 D     0 26009 25789  0  80   0 -  6384 -      pts/9    00:00:00 lvs
>
>
>
>>> dmsetup info -c
>>> dmsetup table
>>> dmsetup status
> Fortunately even in this state these three commands still work OK. Find
> attached current outputs from them.
>