[ovirt-users] hosted-engine migration problems on 3.6, 3.6.1, 3.6.2

Peter laniman1 at laniman.net
Fri Jan 29 19:51:37 UTC 2016


I cannot get the HE to run on node1.  I did 'hosted-engine -vm-shutdown'
on node2.  After it was down I did 'hosted-engine --vm-start' on node1
but did not even get a qemu process.  I did 'hosted-engine
--vm-poweroff' on node1 and 'hosted-engine --vm-start' on node2 and got
it up and running again.

How do I get the HE storage domain imported to the engine to fix all of
this?  Is it a database thing?

Thanks,
Peter

On 01/28/2016 11:02 AM, Peter wrote:
> Martin,
>
> The current state is HE on node2 and all other VMs on node1.  Node2 is
> in local maintenance as of last night and HE should have migrated but
> can't.  The requested logs for this situation are at 
> ftp://aftp.fsl.noaa.gov/divisions/its/peter/ovirt-logs-201601028a/.
>
> I will try shutting down HE on node2 and see if it will run on node1. 
> Last night it would not.
>
> Thanks.
> Peter
>
> On 01/28/2016 03:14 AM, Martin Sivak wrote:
>> Hi,
>>
>> we really need more logs. Preferably the full agent, vdsm and engine
>> log around the time the migration is attempted.
>>
>> The warn/error messages are all related to the fact that hosted engine
>> runs in the "3.5 mode" with the storage domain not being imported to
>> the engine yet.
>>
>> --
>> Martin Sivak
>> oVirt / SLA
>>
>>
>> On Thu, Jan 28, 2016 at 9:28 AM, Nir Soffer <nsoffer at redhat.com> wrote:
>>> Adding Martin
>>>
>>> On Thu, Jan 28, 2016 at 5:54 AM, Peter <laniman1 at laniman.net> wrote:
>>>> I am running oVirt 3.6.2 (original install was 3.6) hosted-engine on a couple of Centos 7.2 servers with SAS attached storage using the new FC support to connect to the LUNs.
>>>>
>>>> Neither the hosted-engine storage nor the hosted-engine VM show up in the GUI.  I know there have been a lot of bugs that have been fixed for this but I still can't import anything.  This may be the basis of my problem.
>>>>
>>>> The problem is that I can't reliably get the hosted-engine to migrate to the other head.  If all VMs and HE are running on node1 and I try to put it in maintenance the VMs will migrate but the HE will not and node1 will show 1 VM remaining on the node (even though the HE VM doesn't show up in the VM list) and will not go into maintenance.
>>>>
>>>> I try setting 'hosted-engine --set-maintenance --mode=local' but that doesn't always help.  The output of 'hosted-engine --vm-status' eventually changes to show the local node in maintenance with score 0 but often it doesn't move.  In /var/log/ovirt-hosted-engine-ha/agent.log I see
>>>>
>>>> MainThread::INFO::2016-01-28 03:15:42,951::config::205::ovirt_hosted_engine_ha.agent.hosted_engine.HostedEngine.config::(refresh_local_conf_file) Trying to get a fresher copy of vm configuration from the OVF_STORE
>>>> MainThread::WARNING::2016-01-28 03:15:43,296::ovf_store::105::ovirt_hosted_engine_ha.lib.ovf.ovf_store.OVFStore::(scan) Unable to find OVF_STORE
>>>> MainThread::ERROR::2016-01-28 03:15:43,297::config::234::ovirt_hosted_engine_ha.agent.hosted_engine.HostedEngine.config::(refresh_local_conf_file) Unable to get vm.conf from OVF_STORE, falling back to initial vm.conf
>>>> MainThread::WARNING::2016-01-28 03:15:43,316::hosted_engine::767::ovirt_hosted_engine_ha.agent.hosted_engine.HostedEngine::(_stop_domain_monitor_if_possible) The VM is running locally or we have no data, keeping the domain monitor.
>>>>
>>>> I've restarted servers and/or processes but I'm not getting anywhere.  I appreciate any help.
>>>>
>>>> _______________________________________________
>>>> Users mailing list
>>>> Users at ovirt.org
>>>> http://lists.ovirt.org/mailman/listinfo/users





More information about the Users mailing list