[ovirt-users] Unable to change compatibility version due to hosted engine

[Simone Tiraboschi]

On Thu, Dec 15, 2016 at 10:33 AM, Renout Gerrits <mail at renout.nl> wrote:

> Hi All,
>
> We have an environment which we want to upgrade to ovirt 4.0. This was
> initially installed at 3.5, then upgraded to 3.6.
> Problem we're facing is that for an upgrade to 4.0 a compatibility version
> of 3.6 is required.
> When changing the cluster compatibility version of the 'Default' cluster
> from 3.5 to 3.6 we get the error in the gui: "Cannot change cluster
> compatibility version when a VM is active. please shutdown all VMs in the
> cluster."
> Even when we shutdown all vm's, except for the Hosted Engine we get this
> error.
> On the hosts a 'vdsClient -s 0 list' is done which will return the HE.
> In the engine logs we have the following error: "2016-12-08 13:00:18,139
> WARN  [org.ovirt.engine.core.bll.storage.UpdateStoragePoolCommand]
> (default task-25) [77a50037] CanDoAction of action 'UpdateStoragePool'
> failed for user admin at internal. Reasons: VAR__TYPE__STORAGE__POOL,VAR__ACTION__UPDATE,$ClustersList
> Default,ERROR_CANNOT_UPDATE_STORAGE_POOL_COMPATIBILITY_
> VERSION_BIGGER_THAN_CLUSTERS"
>
> So problem would be that the HE is in the Default cluster. But how does
> one change the compatibility version when the HE is down?
> I've tried shutting down the engine, changing the version in the DB:
> "UPDATE vds_groups SET compatibility_version='3.6';" and starting the
> engine again.
>
> When I do that and try to start a VM:
> 2016-12-09T13:30:21.346740Z qemu-kvm: warning: CPU(s) not present in any
> NUMA nodes: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
> 2016-12-09T13:30:21.346883Z qemu-kvm: warning: All CPU(s) up to maxcpus
> should be described in NUMA config
> 2016-12-09T13:30:21.355699Z qemu-kvm: "-memory 'slots|maxmem'" is not
> supported by: rhel6.5.0
>
> So that change was rolled back to compatibilty 3.5. After that we we're
> able to start vm's again.
> Please note that all hosts and HE are EL7.
>
> To me this doesn't seem like a strange set-up or upgrade path. Would it be
> possible to start the HE in another cluster than Default or is there a way
> to bypass the vdsClient list check?
> What is the recommended way of upgrading the HE in this case?
>

Please take a look here:
https://bugzilla.redhat.com/show_bug.cgi?id=1364557


>
> Kind regards,
> Renout
>
> _______________________________________________
> 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/20161215/888af94f/attachment-0001.html>