On Thu, Jun 2, 2022 at 4:39 AM Magnus HAGDORN <Magnus.Hagdorn@xxxxxxxx> wrote: > > at this stage we are not so worried about recovery since we moved to > our new pacific cluster. The problem arose during one of the nightly > syncs of the old cluster to the new cluster. However, we are quite keen > to use this as a learning opportunity to see what we can do to bring > this filesystem back to life. > > On Wed, 2022-06-01 at 20:11 -0400, Ramana Venkatesh Raja wrote: > > Can you temporarily turn up the MDS debug log level (debug_mds) to > > > > check what's happening to this MDS during replay? > > > > ceph config set mds debug_mds 10 > > > > > > > 2022-06-02 09:32:36.814 7faca6d16700 5 mds.beacon.store06 Sending > beacon up:replay seq 195662 > 2022-06-02 09:32:36.814 7faca6d16700 1 -- > [v2:192.168.34.113:6800/3361270776,v1:192.168.34.113:6801/3361270776] > --> [v2:192.168.34.179:3300/0,v1:192.168.34.179:6789/0] -- > mdsbeacon(196066899/store06 up:replay seq 195662 v200622) v7 -- > 0x5603d846d200 con 0x560185920c00 > 2022-06-02 09:32:36.814 7facab51f700 1 -- > [v2:192.168.34.113:6800/3361270776,v1:192.168.34.113:6801/3361270776] > <== mon.0 v2:192.168.34.179:3300/0 230794 ==== > mdsbeacon(196066899/store06 up:replay seq 195662 v200622) v7 ==== > 132+0+0 (crc 0 0 0) 0x5603d846d200 con 0x560185920c00 > 2022-06-02 09:32:36.814 7facab51f700 5 mds.beacon.store06 received > beacon reply up:replay seq 195662 rtt 0 > 2022-06-02 09:32:37.090 7faca4d12700 2 mds.0.cache Memory > usage: total 22446592, rss 18448072, heap 332040, baseline 307464, 0 / > 6982189 inodes have caps, 0 caps, 0 caps per inode > 2022-06-02 09:32:37.090 7faca4d12700 10 mds.0.cache cache not ready for > trimming > 2022-06-02 09:32:38.091 7faca4d12700 2 mds.0.cache Memory > usage: total 22446592, rss 18448072, heap 332040, baseline 307464, 0 / > 6982189 inodes have caps, 0 caps, 0 caps per inode > 2022-06-02 09:32:38.091 7faca4d12700 10 mds.0.cache cache not ready for > trimming > 2022-06-02 09:32:38.320 7faca6515700 1 -- > [v2:192.168.34.113:6800/3361270776,v1:192.168.34.113:6801/3361270776] > --> [v2:192.168.34.124:6805/1445500,v1:192.168.34.124:6807/1445500] -- > mgrreport(unknown.store06 +0-0 packed 1414) v8 -- 0x56018651ae00 con > 0x5601869cb400 > 2022-06-02 09:32:39.092 7faca4d12700 2 mds.0.cache Memory > usage: total 22446592, rss 18448072, heap 332040, baseline 307464, 0 / > 6982189 inodes have caps, 0 caps, 0 caps per inode > 2022-06-02 09:32:39.092 7faca4d12700 10 mds.0.cache cache not ready for > trimming > 2022-06-02 09:32:40.094 7faca4d12700 2 mds.0.cache Memory > usage: total 22446592, rss 18448072, heap 332040, baseline 307464, 0 / > 6982189 inodes have caps, 0 caps, 0 caps per inode > 2022-06-02 09:32:40.094 7faca4d12700 10 mds.0.cache cache not ready for > trimming > 2022-06-02 09:32:40.813 7faca6d16700 5 mds.beacon.store06 Sending > beacon up:replay seq 195663 > 2022-06-02 09:32:40.813 7faca6d16700 1 -- > [v2:192.168.34.113:6800/3361270776,v1:192.168.34.113:6801/3361270776] > --> [v2:192.168.34.179:3300/0,v1:192.168.34.179:6789/0] -- > mdsbeacon(196066899/store06 up:replay seq 195663 v200622) v7 -- > 0x5603d846d500 con 0x560185920c00 > 2022-06-02 09:32:40.813 7facab51f700 1 -- > [v2:192.168.34.113:6800/3361270776,v1:192.168.34.113:6801/3361270776] > <== mon.0 v2:192.168.34.179:3300/0 230795 ==== > mdsbeacon(196066899/store06 up:replay seq 195663 v200622) v7 ==== > 132+0+0 (crc 0 0 0) 0x5603d846d500 con 0x560185920c00 > 2022-06-02 09:32:40.813 7facab51f700 5 mds.beacon.store06 received > beacon reply up:replay seq 195663 rtt 0 > 2022-06-02 09:32:41.095 7faca4d12700 2 mds.0.cache Memory > usage: total 22446592, rss 18448072, heap 332040, baseline 307464, 0 / > 6982189 inodes have caps, 0 caps, 0 caps per inode > > "cache is not ready for trimming" is logged when MDS is in replay state. It doesn't tell much about why it's stuck in replay state. The MDS is probably waiting on something from the OSDs. Maybe check the objecter request of the MDS? ceph tell mds.<id> objecter_requests If that's not helpful, then try setting `ceph config set mds debug_objecter 10`, restart the MDS, and check the objecter related logs in the MDS? The other MDS is in resolve state is waiting for the peer to also reach up:resolve. > > > > Is the health of the MDS host okay? Is it low on memory? > > > > > plenty > [root@store06 ~]# free > total used free shared buff/cache a > vailable > Mem: 131939604 75007512 2646656 3380 54285436 > 52944852 > Swap: 32930300 1800 32928500 > > > > > > > The cluster is healthy.> > > > > Can you share the output of the `ceph status` , `ceph fs status` and > > > > `ceph --version`? > [root@store06 ~]# ceph status > cluster: > id: ebaa4a8f-5f17-4d57-b83b-a10f0226efaa > health: HEALTH_WARN > 1 filesystem is degraded > > services: > mon: 3 daemons, quorum store09,store08,store07 (age 10d) > mgr: store08(active, since 15h), standbys: store09, store07 > mds: one:2/2 {0=store06=up:replay,1=store05=up:resolve} 3 > up:standby > osd: 116 osds: 116 up (since 10d), 116 in (since 4M) > > data: > pools: 3 pools, 5121 pgs > objects: 275.90M objects, 202 TiB > usage: 625 TiB used, 182 TiB / 807 TiB avail > pgs: 5115 active+clean > 6 active+clean+scrubbing+deep > > > [root@store06 ~]# ceph fs status > one - 741 clients > === > +------+---------+---------+----------+-------+-------+ > | Rank | State | MDS | Activity | dns | inos | > +------+---------+---------+----------+-------+-------+ > | 0 | replay | store06 | | 7012k | 6982k | > | 1 | resolve | store05 | | 82.9k | 78.4k | > +------+---------+---------+----------+-------+-------+ > +------------------+----------+-------+-------+ > | Pool | type | used | avail | > +------------------+----------+-------+-------+ > | weddell_metadata | metadata | 111G | 1963G | > | weddell_data | data | 622T | 44.0T | > +------------------+----------+-------+-------+ > +-------------+ > | Standby MDS | > +-------------+ > | store09 | > | store08 | > | store07 | > +-------------+ > MDS version: ceph version 14.2.22 > (ca74598065096e6fcbd8433c8779a2be0c889351) nautilus (stable) > > [root@store06 ~]# ceph --version > ceph version 14.2.22 (ca74598065096e6fcbd8433c8779a2be0c889351) > nautilus (stable) > The University of Edinburgh is a charitable body, registered in Scotland, with registration number SC005336. Is e buidheann carthannais a th’ ann an Oilthigh Dhùn Èideann, clàraichte an Alba, àireamh clàraidh SC005336. _______________________________________________ ceph-users mailing list -- ceph-users@xxxxxxx To unsubscribe send an email to ceph-users-leave@xxxxxxx