[Sheepdog] Corrupted VDI repository after Electricity offline

MORITA Kazutaka morita.kazutaka at lab.ntt.co.jp
Fri May 21 10:24:41 CEST 2010


> The sheepdog project works fine with my 3 Debian/squeeze boxes.
> Until an unexpected electricity offline occurred.
> After I restarted sheepdog services, some VDIs were missing.
> I wonder if UPS and a clean shutdown helps.
> So, I create a new sheepdog repository and shutdown each box sequentially
> and concurrently
> to simulate in-order clean shutdown and emergency shutdown in parallel for
> all nodes.
> The result is terrible. All case, VDI repository corrupts.
> Is the sheepdog designed for a non-stop environment?
> or I missed something?

The following patch would solve the restart problem after clean


I know it would be better to have tolerance against total node
failure, but currently not supported yet.



More information about the sheepdog mailing list