[Sheepdog] support object recovery

[MORITA Kazutaka]

On Fri, Jan 22, 2010 at 11:01 AM, Piavlo <piavka at cs.bgu.ac.il> wrote:
> some more questions regarding current recovery implementation:
> With current recovery what is supposed to happen then one of the  nodes
> goes down with --copies=N redundancy:
> 1) Are the vm blocks get replicated to other nodes so that there are
> still N copies of each vm block in active cluster?

Yes.

> If yes does this happen in the background, it the recovery replication
> triggered automatically?

Yes, it is triggered automatically in the background.

> 2)Same questions regarding the vdi meta directories.

Yes, too.
One of sheepdog goals is `autonomously managed storage system'..

>> Yes, vdi directory is like metadata, which we calld `super object' before.
>> Currently, the redundancy of vdi directory is same as objects redundancy.
>> Perhaps, we should change it because vdi directory is more important than
>> data objects.
>>
> Since this vdi metadata overhead is very small , it is probably
> reasonable to store the vdi metadata for all vm images in the sheepdog
> cluster
> on each storage node. This would not require any metadata recovery if
> some node goes offline.

Probably, the cost of updating vdi directory on every nodes is not so cheap
if the number of nodes is large, because each node must return the ack
after updating local vdi directory.
However, no need for metadata recovery is very good.
We should consider more about it.

Thanks,

Kazutaka Morita