[ovirt-users] SPM in case of Failure

[Maor Lipchuk]

Hi arsene,

See my comments inline

On Mon, Jun 12, 2017 at 1:02 PM, Arsène Gschwind
<arsene.gschwind at unibas.ch> wrote:
> Hi
>
> Our setup looks like:
>
> - 2 clusters in 2 different site connected with 10GBit LAN
> - Storage based on FC SAN replicated on both site and available for both
> site (The LUNs are available over 4 pathes, 2 from each site)
>
> My observation:
>
> In case one site goes down and this site owned SPM is it not possible to
> move or force SPM on the second site.

It could be a sanlock issue.
The SPM uses sanlock on the storage domain, so once the SPM host will
be rebooted and sanlock will be released from the storage domain (IINM
after 80 seconds) another Host can obtain a lock on that storage
domain and become the new SPM.
What is the message in the logs that you get when you try to do that?


> On the site which is down it's possible to reset all VMs that crashed using
> the "Confirm Host rebooted" menu on the oVirt Host but this does not reset
> SPM.
> The only solution I found was to bring the Host which owned SPM up again to
> be able to move it to the other site and then reactivate the storage
> domains.

I would try to attach the storage domain ( detach it first if it is
already attached) so you could register any VMs/Templates/Disks that
were added in the original env.

>
> Is this a normal behavior?
> Is there any way to force SPM reelection ?
>
> Thanks for your help or idea...
>
> Regards,
> Arsène
>
> --
>
> Arsène Gschwind
> Fa. Sapify AG im Auftrag der Universität Basel
> IT Services
> Klingelbergstr. 70 |  CH-4056 Basel  |  Switzerland
> Tel. +41 79 449 25 63  |  http://its.unibas.ch
> ITS-ServiceDesk: support-its at unibas.ch | +41 61 267 14 11
>
>
> _______________________________________________
> Users mailing list
> Users at ovirt.org
> http://lists.ovirt.org/mailman/listinfo/users
>