[sheepdog-users] re-start a cluster using md

Liu Yuan namei.unix at gmail.com
Thu May 9 10:03:34 CEST 2013


On 05/08/2013 11:32 PM, Valerio Pachera wrote:
> And it works.
> 
> I tried then:
> cluster created and started with a single devie
>   sheep /mnt/sheepdog
> stop the cluster and add a disk
> restart the cluster giving *both* devices
>   sheep /mnt/sheepdog,/mnt/sheepdog/obj,/mnt/sdb1
> 
> Cluster starts but running vdi check find missing pieces:
> fixed missing a34c670000079e
> fixed missing a34c67000007e0
> fixed missing a34c67000007e0
> fixed missing a34c670000088c
> fixed missing a34c670000088c
> fixed missing a34c67000008a0
> fixed missing a34c67000008a0
> fixed missing a34c67000008c0
> fixed missing a34c6700000900
> fixed missing a34c6700000900
> fixed missing a34c6700000920
> fixed missing a34c6700000920

Ah, yeah, this method doesn't work well because the data re-balance
isn't triggered at startup. We don't know the if we change the topology
of disks compared to status before shutdown or crashed. So this method
should be deprecated.

Thanks,
Yuan




More information about the sheepdog-users mailing list