[sheepdog-users] sheepdog recovery question

Liu Yuan namei.unix at gmail.com
Tue May 20 07:08:19 CEST 2014


On Sat, May 17, 2014 at 03:37:39PM +0800, 飞 wrote:
> hello, I have running a sheepdog testing cluster with 3 hosts, and every VDI have 3 copies,
> now , I am testing the recovery with following steps:
> 
> 1. I create 100 VDI , and every VDI about 5GB.
> 2. I shutdown the power of one host to Simulated host failure
> 3. 20 minutes later , I power on the host
> 
> than , I can see the host in recovery progress, but the host seems to delete all objects first, 
> and then , rebuild the objects from other copies . it's something wrong ?
> I think , the sheep shound be just update the Different objects, 
> Should NOT remove all the objects and then rebuild .

We actually didn't remove the objects, we move the objects into stale directory
and if not stale, we'll move it back. Practically we do what you described.

Thanks
Yuan



More information about the sheepdog-users mailing list