[sheepdog-users] [ANNOUNCE] Sheepdog 0.5.1 released

MORITA Kazutaka morita.kazutaka at lab.ntt.co.jp
Thu Oct 4 22:42:05 CEST 2012

At Thu, 04 Oct 2012 12:05:15 +0200,
Bastian Scholz wrote:
> Hi Kazutaka,
> > What is the output of 'collie cluster info' on the node you started
> > sheep successfully?
> >
> > Can you share your config and epoch/* in your store directory?  It
> > would help us to know what's going on.
> I can give you only the config-/epoch-files at the
> moment, because I cant access my testcluster now...
> If it is not enough, I can give you the output of
> collie node info at the weekend.

Thanks, I pushed the fix to the devel branch.

Your store is corrupted because of migration code in 0.5.1 (sorry for
that).  Can you try the following procedures to recover it?

 1. backup node2/1/epoch/00000001 (I guess this contains the correct
    node info.)

 2. get sheepdog from the devel branch
    (https://github.com/collie/sheepdog/zipball/devel) and run sheep
    with a '-u' option.

 3. kill all sheeps and copy the backuped epoch file to epoch/00000001
    on all nodes.

 4. run sheeps again



More information about the sheepdog-users mailing list