Ah, this appears to be http://tracker.ceph.com/issues/6087.
If you are able and care to install dev packages then the fix is now
in the dumpling branch and will be in the next release.
-Greg
Software Engineer #42 @ http://inktank.com | http://ceph.com
On Wed, Aug 21, 2013 at 2:06 AM, Damien Churc
I've uploaded the complete log[0]. It's about 70MB just as a warning.
[0] damoxc.net/ceph-mds.ceph2.log.1.gz
On 21 August 2013 07:07, Gregory Farnum wrote:
> Do you have full logs from the beginning of replay? I believe you
> should only see this when a client is reconnecting to the MDS with
> f
Do you have full logs from the beginning of replay? I believe you
should only see this when a client is reconnecting to the MDS with
files that the MDS doesn't know about already, which shouldn't happen
at all in a single-MDS system. Although that "pool -1" also looks
suspicious and makes me wonder
Hi,
After upgrading to dumpling I appear unable to get the mds cluster
running. The active server just sits in the rejoin state spinning and
causing lots of i/o on the osds. Looking at the logs it appears to be
going through checking a vast number of missing inodes.
2013-08-20 13:50:29.129624 7fd
