Re: Internal error during live migration saving

[Nathan March <nathan@gt.net>]

On 9/14/2011 10:53 AM, Shriram Rajagopalan wrote:
> On Tue, Sep 13, 2011 at 2:01 PM, Nathan March<nathan@gt.net>  wrote:
>> Just wondering if this is a known bug?
>>
>> Trying to migrate the VM off to a diff dom0 results in the below error.
>> Other VMs migrated off fine (started at around the same time as this vm) and
>> I've tried a few different target servers, all resulting in the same thing.
>>
> Were other domains linux 3.0.3 as well ?
All the dom0's are 3.0.3 and all the domU's are 2.6.32.27 (w/ grsec).

I did a cold reboot of the VM and now it migrates properly.

>> [2011-09-13 13:48:24 3996] DEBUG (XendCheckpoint:124) [xc_save]:
>> /usr/lib/xen/bin/xc_save 29 77 0 0 1
>> [2011-09-13 13:48:24 3996] INFO (XendCheckpoint:423) xc_save: failed to get
>> the suspend evtchn port
>> [2011-09-13 13:48:24 3996] INFO (XendCheckpoint:423)
>> [2011-09-13 13:49:03 3996] DEBUG (XendCheckpoint:394) suspend
>> [2011-09-13 13:49:03 3996] DEBUG (XendCheckpoint:127) In saveInputHandler
>> suspend
>> [2011-09-13 13:49:03 3996] DEBUG (XendCheckpoint:129) Suspending 77 ...
>> [2011-09-13 13:49:03 3996] DEBUG (XendDomainInfo:524)
>> XendDomainInfo.shutdown(suspend)
>> [2011-09-13 13:49:03 3996] DEBUG (XendDomainInfo:1881)
>> XendDomainInfo.handleShutdownWatch
>> [2011-09-13 13:50:06 3996] DEBUG (XendDomainInfo:1881)
>> XendDomainInfo.handleShutdownWatch
>> [2011-09-13 13:50:06 3996] INFO (XendCheckpoint:423) xc: error: Suspend
>> request failed: Internal error
>> [2011-09-13 13:50:06 3996] INFO (XendCheckpoint:423) xc: error: Domain
>> appears not to have suspended: Internal error
>> [2011-09-13 13:50:06 3996] ERROR (XendCheckpoint:185) Save failed on domain
>> globish (77) - resuming.
>> Traceback (most recent call last):
>>   File "/usr/lib64/python2.6/site-packages/xen/xend/XendCheckpoint.py", line
>> 146, in save
>>     forkHelper(cmd, fd, saveInputHandler, False)
>>   File "/usr/lib64/python2.6/site-packages/xen/xend/XendCheckpoint.py", line
>> 395, in forkHelper
>>     inputHandler(line, child.tochild)
>>   File "/usr/lib64/python2.6/site-packages/xen/xend/XendCheckpoint.py", line
>> 131, in saveInputHandler
>>     dominfo.waitForSuspend()
>>   File "/usr/lib64/python2.6/site-packages/xen/xend/XendDomainInfo.py", line
>> 2998, in waitForSuspend
>>     raise XendError(msg)
>> XendError: Timeout waiting for domain 77 to suspend
>> [2011-09-13 13:50:06 3996] DEBUG (XendDomainInfo:3135)
>> XendDomainInfo.resumeDomain(77)
>>
>> xend-debug.log and the target dom0 logs don't show anything of value.
>>
>> This is xen 4.1.1 on linux 3.0.3
>>
> Did you try xm save -c (or the xl equivalent) ? This should be
> activating the same
> code path where this error seems to appear.
>
> Also, make sure you have CONFIG_XEN_SAVE_RESTORE enabled.
Unfortunately I didn't think to try it. I do have that set on both dom0 
and domu.

- Nathan