[Sheepdog] Cluster doesn't come up correctly after reboot

[MORITA Kazutaka]

On 2010/04/19 17:09, Wido den Hollander wrote:
>> Let me clarify a few things. Did you run `shepherd shutdown' before
>> stopping collie processes? Do problems occur only when rebooting
>> sheepdog? Clean startup doesn't cause problems always?
> 
> No, i did not run "shepherd shutdown" before stopping the collie
> processes, but what i just noticed:
> 
> I ran "shepherd shutdown" on one of my 5 nodes and collie stopped on all
> 5 nodes, it that the expected behaviour?
> 
Yes. If you want to stop only one of them, kill the collie process directly.
Remained collies should work as 4 nodes cluster.

> Restarting the collie processes on all 5 nodes brought the cluster back
> up as it should, so it seems that calling a shutdown is mandatory before
> shutting down a node.
> 
Yes, shutdown command is neccessary when stopping all the nodes.
I think collie should show more friendly information about cluster status,
and provide the means to recover from unexpected shutdown or total failure.

Thanks,

Kazutaka Morita