[sheepdog-users] Question on cluster recovery when power failure?

MORITA Kazutaka morita.kazutaka at lab.ntt.co.jp
Tue Oct 2 19:34:55 CEST 2012


At Tue, 2 Oct 2012 17:04:19 +0700,
icez network wrote:
> 
> Hello
> 
> Last week I found the power failure at the test cluster that take down my
> whole sheepdog cluster. The cluster restore process is smooth as I started
> back all the sheepdog nodes. All data are in place with no any data loss.
> 
> Then I think that, as far as I know, when whole cluster
> shutdown, sheepdog requires all nodes in the same cluster to be started to
> resume running again. (Please correct me if I'm wrong.) And If there's node
> in the cluster that failed to start (e.g.: storage hard disk failure caused
> from power failure, system crash with no way to recovery) so the whole
> cluster cannot resume the running state back as there's missing node.
> 
> May I ask that if the situation above occurs, how can I recover the cluster
> back. Does 'collie cluster recovery' do the magic that resume the cluster
> state? Or which command I have to use to recover the cluster back.

In such case, 'collie cluster recover force' fixes your cluster state.

Thanks,

Kazutaka



More information about the sheepdog-users mailing list