General

Profile

David Burley

Issues

open closed Total
Assigned issues 0 0 0
Reported issues 0 1 1

Activity

07/27/2015

02:10 PM Ceph Bug #11586: RBD Data Corruption with Logging on Giant
Samuel Just wrote:
> #12465 might be a plausible explanation. Can you reproduce on a disposable cluster? For #1246...
David Burley

07/25/2015

12:51 AM Ceph Bug #11586: RBD Data Corruption with Logging on Giant
Samuel Just wrote:
> Do you mean that the rbd object file on that osd wound up larger than 4MB?
Yes, it was 4.7...
David Burley

07/21/2015

11:33 PM Ceph Bug #11586: RBD Data Corruption with Logging on Giant
I just looked for one, and the script stored them in /tmp, which has been cleared since. If you really need an exampl... David Burley
11:08 PM Ceph Bug #11586: RBD Data Corruption with Logging on Giant
None in an unfixed state. I can probably hunt down a copy of an inconsistent pg object with the corruption though, gi... David Burley
10:48 PM Ceph Bug #11586: RBD Data Corruption with Logging on Giant
We did initially, since that's what the docs said to do:
http://ceph.com/docs/master/rados/troubleshooting/troublesh...
David Burley
10:36 PM Ceph Bug #11586: RBD Data Corruption with Logging on Giant
Both. We had some ext4 or XFS filesystems direct mounted on database hosts that were also corrupt, for instance. Thos... David Burley
10:26 PM Ceph Bug #11586: RBD Data Corruption with Logging on Giant
Pacemaker is managing the resource and ensuring the rbd is only present on one node at a time. I am not sure how this... David Burley
10:14 PM Ceph Bug #11586: RBD Data Corruption with Logging on Giant
Its implemented as described here (excepting this doc is old and the commands needed updated to function on the newer... David Burley
10:05 PM Ceph Bug #11586: RBD Data Corruption with Logging on Giant
We have a mix of use cases for the RBDs, and all of the use cases were impacted if the volume was created prior to th... David Burley
09:55 PM Ceph Bug #11586: RBD Data Corruption with Logging on Giant
Several mon processes were OOM killed across the cluster, causing the cluster mons to lose a quorum. The NFS cluster ... David Burley

Also available in: Atom