[sheepdog-users] Crash of a node (maybe due to cache)

[Liu Yuan]

On 06/06/2013 07:56 PM, Valerio Pachera wrote:
> Oh yes. But not just before it.
> I turned off the guest and created a snapshot at about17:00, so 2/3
> hours before the backups started writing data.

What is QEMU and Sheep version for this cluster?

What command did you use to snapshot the vdi?

If you turned off the guest, the cache should be released. then we
shouldn't never get this error. Anyway, I've sent a patch to fix a vdi
snapshot bug  for the case that there is dirty objects in the cache
while taking snapshots.

Thanks,
Yuan