[Users] Wrong Master domain or its version

[Michael Kublin]

----- Original Message -----
> From: "Jonas Israelsson" <jonas.israelsson at elementary.se>
> To: "Michael Kublin" <mkublin at redhat.com>
> Cc: users at ovirt.org
> Sent: Tuesday, April 23, 2013 3:04:17 PM
> Subject: Re: [Users] Wrong Master domain or its version
> 
> > Hi, I suppose you have one host, if you will take a look to
> > rhev-engine logs you will see
> > kind of the following error: Cannot connect to Pool wrong master
> > domain or it version.
> 
> One host (only), yes.
> 
> > The only advice is to access to DB and update msater domain version
> > inside storage_pool table.
> > The query should be "update storage_pool set
> > master_domain_version=110"
> > after that please try to activate a host again
> 
> Indeed, that seem to do the trick (thanks!). Setting the version in the
> database to match the one from the log does make me able to bring the
> host back online
> 
> While the host came back online as well as all other storage domain the
> master domain did not. I remains in status locked. Part from not being
> able to access the vm-data from that domain it also gives some funny
> side effects. For instance I could quite happily put the export-domain
> to maintenance mode but not re-attach the same since the master domain
> is not active.
> 
If domains are marked as Locked in GUI they be also unlocked in DB.
My advice is to put host in maintainence, after that
please run the following query :  update storage_pool_iso_map set status = 0
(The following table also contains storage_pool_id and storage_id, so you can 
make query more specific, info about domains is located inside storage_domain_static
table)
Activate a host, after that host should try to connect to all storages and to pool again.

> Not sure if related, but when trying to also take the host into
> maintenance mode it refuses due to ongoing tasks.
> 
> [root at fat-tony admin]# vdsClient -s 0 getAllTasks
> 4ca8e2f9-b8a6-425d-baba-b3cb9a6a154f :
>          verb = createVolume
>          code = 0
>          state = failed
>          tag = spm
>          result =
>          message = Task is initializing
>          id = 4ca8e2f9-b8a6-425d-baba-b3cb9a6a154f
> 
> I did manage to get rid of that with 'vdsClient clearTask' but it makes
> no difference putting the host in and out of maintenance mode and or
> rebooting it. The lock remains.
> 
Task are left in host and not located in engine it is a known issue, and I think that engine 
should handle such problem if domains was not in status Locked 
> I fail to find any information where this lock is set. Is is done by the
> vdsm-layer and therefore to be controlled by the...
> 
> 'vdsClient aquire|releaseDomainLock' .. ?
> 
> That however seem to fail as well
> 
> vdsClient -s 0 releaseDomainLock dae1ba5e-548a-473e-94b6-7d08d3a6229a
> ea5643f7-f41f-4c44-a27f-f31240cd0f19
> Traceback (most recent call last):
>    File "/usr/share/vdsm/vdsClient.py", line 2411, in <module>
>    File "/usr/share/vdsm/vdsClient.py", line 1381, in releaseDomainLock
>    File "/usr/lib64/python2.7/xmlrpclib.py", line 1224, in __call__
>    File "/usr/lib64/python2.7/xmlrpclib.py", line 1578, in __request
>    File "/usr/lib64/python2.7/xmlrpclib.py", line 1264, in request
>    File "/usr/lib64/python2.7/xmlrpclib.py", line 1297, in single_request
>    File "/usr/lib64/python2.7/xmlrpclib.py", line 1473, in parse_response
>    File "/usr/lib64/python2.7/xmlrpclib.py", line 793, in close
> Fault: <Fault 1: '<type \'exceptions.Exception\'>:method
> "releaseDomainLock" is not supported'>
> 
> 
> Could someone please point me in the right direction ?
> 
> Rgds Jonas
>