[Users] single VM disappeared

[Roy Golan]

On 09/12/2013 05:20 PM, Markus Stockhausen wrote:
>> Von: Roy Golan [rgolan at redhat.com]
>> Gesendet: Donnerstag, 12. September 2013 15:58
>> An: Markus Stockhausen
>> Betreff: Re: [Users] single VM disappeared
>>
>> On Thu 12 Sep 2013 02:57:13 PM IDT, Markus Stockhausen wrote:
>>> Hello,
>>>
>>> continuing with my 3.3 error series. I just lost one of my VMs after
>>> restart of ovirt-engine. The things I remember are:
>>>
>>> - The machine was up and running.
>>> - I changed the cluster cpu from penryn to nehalem
here the emulated machine nullified.
>>> - Machine was stopped
>>> - Restart failed with error in ovirt-engine.log
>>>
>>> Failed creating vm win2k3std_de in vds =
>>> 30b9bd7e-ea05-4910-85a7-f014ba8a8bab :
>>> colovn1 error =
>>> org.ovirt.engine.core.vdsbroker.vdsbroker.VDSNetworkException:
>>> org.apache.xmlrpc.common.XmlRpcExtensionException:
>>> Null values aren't supported, if isEnabledForExtensions() == false
>>>
>>> - Without any idea how to solve the error I restarted ovirt-engine
>>> - Now Machine is missing
>>> - Error log reads
>>>
>>> Correlation ID: 2c29cebd, Call Stack: null, Custom
>>> Event ID: -1, Message: Failed to import Vm win2k3std_de to
>>> Data Center Collogia, Cluster Produktion
>>>
>>> I uploaded the engine.log tohttp://pastebin.com/sHspGTgy
>>>
>>> Markus
>>>
>>>
>>> _______________________________________________
>>> Users mailing list
>>> Users at ovirt.org
>>> http://lists.ovirt.org/mailman/listinfo/users
>> I think thats an emulated machine problem which I'm just takin care of.
>>
>> to check:
>>
>> go to cluster, select your cluster, got to the genral subtab and check
>> what is "emulated machine" value
>> I guess its null. to make it work,
>> 1.take one of your host which belongs to that cluster to maintenance
>> 2.activate it back.
>> 3. check the emulated machine value is set now
>>
>> cause:
>> Bughttps://bugzilla.redhat.com/1004695
>> Editing the cluster nullify the emulated machine value
>>
>> Roy
> Hello,
>
> thanks for helping me through that minefield. Ok here are the results
> when following your guide.
please provide the output of:

psql engine postgres -c "select * from async_tasks;"
>
> - Emulated machine: pc-1.0
right, you restart the engine so the host already was reactivated so 
this is actually expected.
> - maintenance/activate one node of the cluster:
> - Emulated machine is still: pc-1.0
>
> I guess that is nothing you wanted to see.  So I changed the CPU
> type of the cluster and was able to understand what you think of.
> The emulated machine will go blank. After reactivating a node
> the value goes back to pc-1.0
>
> I played with the cluster settings in between and cannot say if the
> value was empty when the error occured. Just in case the emulated
> machine value was empty during restart of ovirt-engine. Did I loose
> the machine because of that?
no this can't lead to that but the restart of ovirt can.

Yair, seems like importVM task is still in the DB after an engine 
restart during run VM fail.

> As I'm testing that is no problem but
> I want to make sure if I can recover from that situation in any way.
>
> Best regards.
>
> Markus
>
>