[ovirt-users] Frequent vm migration failure

[Michal Skrivanek]

> On 17 Apr 2018, at 11:28, Stefano Stagnaro <stefanos at prismatelecomtesting.com> wrote:
> 
> On Thu, 2018-04-12 at 20:20 +0200, Michal Skrivanek wrote:
>> 
>> 
>>> On 12 Apr 2018, at 18:26, Stefano Stagnaro <stefanos at prismatelecomtesting.com <mailto:stefanos at prismatelecomtesting.com>> wrote:
>>> Hi,
>>> 
>>> I recently upgraded an oVirt deployment from 3.6 to 4.0 and then 4.1.9 (my actual release). Since then, when migrating many hosts simultaneously I always experience few migrations failure like 1 on 10 vms. The failure can occur on any host; moreover, after a couple of failure the destination host fall in Error status and I have to manually re-activate or wait 30 min.

Are they all in a 3.6 cluster? 

>>> 
>>> Tipical error found on vdsm log is (from the source host):
>>> 2018-04-12 17:01:32,097+0200 ERROR (migsrc/3192dfe7) [virt.vm] (vmId='3192dfe7-eeac-4626-8c86-e49facc9006f') migration destination error: Fatal error during migration (migration:287)
>>> 
>>> Please find the logs of source host (v15.ovirt), destination host (v14.ovirt) and engine here: https://www.dropbox.com/sh/xhf8ry4ih40poxd/AABxiFCIxDe14HSx2DqLE61ya?dl=0 <https://www.dropbox.com/sh/xhf8ry4ih40poxd/AABxiFCIxDe14HSx2DqLE61ya?dl=0>
>>> 
>>> Some of the vm affected from the migration failure are:
>>> svn	3192dfe7-eeac-4626-8c86-e49facc9006f
>>> wood	a8e83ff0-dfed-4074-b6b6-e947b8ebb952
>>> qnx66	5697c4a4-9e40-4dd6-aba2-c8ab9904a584
>> 
>> can you also include qemu log from /var/log/libvirt/qemu/<vmname>?
> 
> Hi Michal, I've added libvirt logs for relevant VMs on the previous Dropbox share.

I do not see anything wrong. It’s a bit too much data to go through, can you pinpoint the time and VM name when you see a failure?

Thanks,
michal
> 
>> 
>> btw you seem to be using the legacy migration policy throttling the speed significantly. Please read into the migration enhancements in 4.0
>> https://www.ovirt.org/develop/release-management/features/virt/migration-enhancements/ <https://www.ovirt.org/develop/release-management/features/virt/migration-enhancements/>
> I've already moved to Minimal Downtime and then to Post-copy with same results. VM migrations continue to fail randomly.
> 
>> 
>> Thanks,
>> michal
> 
> Thanks,
> Stefano.
> 
> 
>> 
>>> 
>>> Thank you very much for your help.
>>> 
>>> -- 
>>> Stefano Stagnaro
>>> 
>>> Prisma Telecom Testing S.r.l.
>>> Via Petrocchi, 4
>>> 20127 Milano – Italy
>>> 
>>> Tel. 02 26113507 int 339
>>> e-mail: stefanos at prismatelecomtesting.com <mailto:stefanos at prismatelecomtesting.com>
>>> skype: stefano.stagnaro
>>> _______________________________________________
>>> Users mailing list
>>> Users at ovirt.org
>>> http://lists.ovirt.org/mailman/listinfo/users
>> 
> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ovirt.org/pipermail/users/attachments/20180419/86fe9919/attachment.html>