[Users] Node crash

Itamar Heim iheim at redhat.com
Tue Feb 7 22:58:10 UTC 2012


On 02/07/2012 10:58 PM, зоррыч wrote:
> Hi.
>
> If one of the node crash, virtual servers running on it is not possible
> to recover. oVirt does not permit to run a virtual server running on the
> faulty node repeatedly and asked to restore the node.
>
> What to do if a node crash in a very long time? It is possible to
> restore the virtual servers running on it?

the node should be fenced to release the resources by configuring power 
management (and ovirt engine will fence it).
if you can't/won't configure fencing, you can manually confirm the node 
has been shut down, so engine will know it is no longer running the VMs.

if you will not configure fencing correctly, or hit 'manual confirm' 
without actually shutting it down - it *will* corrupt any VM running on 
more than a single host.

>
> There are mechanisms backup virtual servers, so that when node crash the
> virtual server will automatically run on another node?

yes - fencing (edit host, power managment).
engine will also auto start VMs which are flagged as HA VMs.



More information about the Users mailing list