On Jun 23, 2011, at 9:17 PM, Mark Nigh wrote:
> I would like some assistance in troubleshooting and fixes the issue of a MDS in replay.
>
> root@ceph001:/var/log/ceph# ceph -s
> 2011-06-23 23:07:43.945503 pg v2802: 396 pgs: 328 active+clean, 50 down+degraded+peering, 18 crashed+down+degraded+peering; 1155 MB data, 2575 MB used, 11115 GB / 11118 GB avail; 8/5958 degraded (0.134%)
> 2011-06-23 23:07:43.946151 mds e1458: 1/1/1 up {0=0=up:replay}
> 2011-06-23 23:07:43.946170 osd e808: 4 osds: 4 up, 4 in
> 2011-06-23 23:07:43.946205 log 2011-06-23 23:06:34.284905 osd2 10.6.1.81:6800/3611 1 : [WRN] map e803 wrongly marked me down or wrong addr
> 2011-06-23 23:07:43.946245 mon e1: 1 mons at {0=10.6.1.80:6789/0}
> root@ceph001:/var/log/ceph# ceph -v
> ceph version 0.29.1-466-g86b41ff (commit:86b41ff96e0f6a9efb9553790efbb3c89a5ab080)
>
> I wish I could tell you how this happen but I did open another post earlier in the week that was solved with consistent version of ceph throughout my 2 node cluster. I have one (1) mds and one (1) mon located on the 1st node. The four (4) osd are spread across the two (2) nodes evenly. The version are consistent.
>
> Please let me know what I can provide to fix this problem. Thank you.
I don't think this is an MDS issue -- you've got a lot of PGs which aren't currently active and the MDS is probably blocked on reading from one of them.
Have your PGs sorted themselves out? If not, are they at least making progress? (Not that it should take them this long...)
-Greg--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
