Is it enough ? # tail -n500 -f /var/log/ceph/osd.28.log | grep -A5 -B5 'found clone without head' 2013-05-22 15:43:09.308352 7f707dd64700 0 log [INF] : 9.105 scrub ok 2013-05-22 15:44:21.054893 7f707dd64700 0 log [INF] : 9.451 scrub ok 2013-05-22 15:44:52.898784 7f707cd62700 0 log [INF] : 9.784 scrub ok 2013-05-22 15:47:43.148515 7f707cd62700 0 log [INF] : 9.3c3 scrub ok 2013-05-22 15:47:45.717085 7f707dd64700 0 log [INF] : 9.3d0 scrub ok 2013-05-22 15:52:14.573815 7f707dd64700 0 log [ERR] : scrub 3.6b ade3c16b/rb.0.15c26.238e1f29.000000009221/12d7//3 found clone without head 2013-05-22 15:55:07.230114 7f707d563700 0 log [ERR] : scrub 3.6b 261cc0eb/rb.0.15c26.238e1f29.000000003671/12d7//3 found clone without head 2013-05-22 15:56:56.456242 7f707d563700 0 log [ERR] : scrub 3.6b b10deaeb/rb.0.15c26.238e1f29.0000000086a2/12d7//3 found clone without head 2013-05-22 15:57:51.667085 7f707dd64700 0 log [ERR] : 3.6b scrub 3 errors 2013-05-22 15:57:55.241224 7f707dd64700 0 log [INF] : 9.450 scrub ok 2013-05-22 15:57:59.800383 7f707cd62700 0 log [INF] : 9.465 scrub ok 2013-05-22 15:59:55.024065 7f707661a700 0 -- 192.168.42.3:6803/12142 >> 192.168.42.5:6828/31490 pipe(0x2a689000 sd=108 :6803 s=2 pgs=200652 cs=73 l=0).fault with nothing to send, going to standby 2013-05-22 16:01:45.542579 7f7022770700 0 -- 192.168.42.3:6803/12142 >> 192.168.42.5:6828/31490 pipe(0x2a689280 sd=99 :6803 s=0 pgs=0 cs=0 l=0).accept connect_seq 74 vs existing 73 state standby -- 2013-05-22 16:29:49.544310 7f707dd64700 0 log [INF] : 9.4eb scrub ok 2013-05-22 16:29:53.190233 7f707dd64700 0 log [INF] : 9.4f4 scrub ok 2013-05-22 16:29:59.478736 7f707dd64700 0 log [INF] : 8.6bb scrub ok 2013-05-22 16:35:12.240246 7f7022770700 0 -- 192.168.42.3:6803/12142 >> 192.168.42.5:6828/31490 pipe(0x2a689280 sd=99 :6803 s=2 pgs=200667 cs=75 l=0).fault with nothing to send, going to standby 2013-05-22 16:35:19.519019 7f707d563700 0 log [INF] : 8.700 scrub ok 2013-05-22 16:39:15.422532 7f707dd64700 0 log [ERR] : scrub 3.1 b1869301/rb.0.15c26.238e1f29.000000000836/12d7//3 found clone without head 2013-05-22 16:40:04.995256 7f707cd62700 0 log [ERR] : scrub 3.1 bccad701/rb.0.15c26.238e1f29.000000009a00/12d7//3 found clone without head 2013-05-22 16:41:07.008717 7f707d563700 0 log [ERR] : scrub 3.1 8a9bec01/rb.0.15c26.238e1f29.000000009820/12d7//3 found clone without head 2013-05-22 16:41:42.460280 7f707c561700 0 log [ERR] : 3.1 scrub 3 errors 2013-05-22 16:46:12.385678 7f7077735700 0 -- 192.168.42.3:6803/12142 >> 192.168.42.5:6828/31490 pipe(0x2a689c80 sd=137 :6803 s=0 pgs=0 cs=0 l=0).accept connect_seq 76 vs existing 75 state standby 2013-05-22 16:58:36.079010 7f707661a700 0 -- 192.168.42.3:6803/12142 >> 192.168.42.3:6801/11745 pipe(0x2a689a00 sd=44 :6803 s=0 pgs=0 cs=0 l=0).accept connect_seq 40 vs existing 39 state standby 2013-05-22 16:58:36.798038 7f707d563700 0 log [INF] : 9.50c scrub ok 2013-05-22 16:58:40.104159 7f707c561700 0 log [INF] : 9.526 scrub ok Note : I have 8 scrub errors like that, on 4 impacted PG, and all impacted objects are about the same RBD image (rb.0.15c26.238e1f29). Le mercredi 22 mai 2013 à 11:01 -0700, Samuel Just a écrit : > Can you post your ceph.log with the period including all of these errors? > -Sam > > On Wed, May 22, 2013 at 5:39 AM, Dzianis Kahanovich > <mahatma@xxxxxxxxxxxxxx> wrote: > > Olivier Bonvalet пишет: > >> > >> Le lundi 20 mai 2013 à 00:06 +0200, Olivier Bonvalet a écrit : > >>> Le mardi 07 mai 2013 à 15:51 +0300, Dzianis Kahanovich a écrit : > >>>> I have 4 scrub errors (3 PGs - "found clone without head"), on one OSD. Not > >>>> repairing. How to repair it exclude re-creating of OSD? > >>>> > >>>> Now it "easy" to clean+create OSD, but in theory - in case there are multiple > >>>> OSDs - it may cause data lost. > >>> > >>> I have same problem : 8 objects (4 PG) with error "found clone without > >>> head". How can I fix that ? > >> since "pg repair" doesn't handle that kind of errors, is there a way to > >> manually fix that ? (it's a production cluster) > > > > Trying to fix manually I cause assertions in trimming process (died OSD). And > > many others troubles. So, if you want to keep cluster running, wait for > > developers answer. IMHO. > > > > About manual repair attempt: see issue #4937. Also similar results - in subject > > "Inconsistent PG's, repair ineffective". > > > > -- > > WBR, Dzianis Kahanovich AKA Denis Kaganovich, http://mahatma.bspu.unibel.by/ > > _______________________________________________ > > ceph-users mailing list > > ceph-users@xxxxxxxxxxxxxx > > http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com > _______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
 |