As you can see, two (2) of the OSDs are down and out. They weren't when I first wrote the post. root@ceph001:~# ceph osd dump -o - 2011-06-21 19:20:52.025776 mon <- [osd,dump] 2011-06-21 19:20:52.026506 mon0 -> 'dumped osdmap epoch 612' (0) epoch 612 fsid f1082578-eb4d-cdd7-4da2-5e6293396315 created 2011-06-20 16:05:04.711991 modifed 2011-06-21 18:15:33.498931 flags pg_pool 0 'data' pg_pool(rep pg_size 2 crush_ruleset 0 object_hash rjenkins pg_num 128 pgp_num 128 lpg_num 2 lpgp_num 2 last_change 1 owner 0) pg_pool 1 'metadata' pg_pool(rep pg_size 2 crush_ruleset 1 object_hash rjenkins pg_num 128 pgp_num 128 lpg_num 2 lpgp_num 2 last_change 1 owner 0) pg_pool 2 'rbd' pg_pool(rep pg_size 2 crush_ruleset 2 object_hash rjenkins pg_num 128 pgp_num 128 lpg_num 2 lpgp_num 2 last_change 1 owner 0) max_osd 4 osd0 up in weight 1 up_from 597 up_thru 611 down_at 596 last_clean_interval 2-595 10.6.1.80:6801/1891 10.6.1.80:6802/1891 10.6.1.80:6803/1891 osd1 up in weight 1 up_from 597 up_thru 611 down_at 596 last_clean_interval 3-595 10.6.1.80:6804/1982 10.6.1.80:6805/1982 10.6.1.80:6806/1982 osd2 down out up_from 596 up_thru 597 down_at 605 last_clean_interval 590-594 osd3 down out up_from 596 up_thru 597 down_at 605 last_clean_interval 593-594 They start up fine. root@ceph002:~# /etc/init.d/ceph start osd2 === osd.2 === Mounting Btrfs on ceph002:/mnt/osd2 Scanning for Btrfs filesystems failed to read /dev/sr0 Starting Ceph osd.2 on ceph002... ** WARNING: Ceph is still under development. Any feedback can be directed ** ** at ceph-devel@xxxxxxxxxxxxxxx or http://ceph.newdream.net/. ** starting osd2 at 0.0.0.0:6800/8620 osd_data /mnt/osd2 /data/osd2/journal root@ceph002:~# /etc/init.d/ceph start osd3 === osd.3 === Mounting Btrfs on ceph002:/mnt/osd3 Scanning for Btrfs filesystems failed to read /dev/sr0 Starting Ceph osd.3 on ceph002... ** WARNING: Ceph is still under development. Any feedback can be directed ** ** at ceph-devel@xxxxxxxxxxxxxxx or http://ceph.newdream.net/. ** starting osd3 at 0.0.0.0:6803/8769 osd_data /mnt/osd3 /data/osd3/journal But I do get these messages in the osd logs 2011-06-21 19:24:27.937179 7f93df30d700 -- 10.6.1.81:6801/8620 >> 10.6.1.81:6806/8769 pipe(0x30f7a00 sd=19 pgs=0 cs=0 l=0).accept bad authorizer 2011-06-21 19:24:27.937254 7f93df30d700 AuthNoneAuthorizeHandle::verify_authorizer() failed to decode 2011-06-21 19:24:27.937270 7f93df30d700 -- 10.6.1.81:6801/8620 >> 10.6.1.81:6806/8769 pipe(0x30f7a00 sd=19 pgs=0 cs=0 l=0).accept bad authorizer Here are the rule output. root@ceph001:~# crushtool -i cm --test --rule 0 devices weights (hex): [10000,10000,10000,10000] rule 0 (data), x = 0..9999 device 0: 5013 device 1: 4987 device 2: 4963 device 3: 5037 num results 2: 10000 root@ceph001:~# crushtool -i cm --test --rule 1 devices weights (hex): [10000,10000,10000,10000] rule 1 (metadata), x = 0..9999 device 0: 5010 device 1: 5037 device 2: 4975 device 3: 4978 num results 2: 10000 Thanks for your assistance. Mark Nigh Systems Architect mnigh@xxxxxxxxxxxxxxx (p) 314.392.6926 -----Original Message----- From: Sage Weil [mailto:sage@xxxxxxxxxxxx] Sent: Tuesday, June 21, 2011 5:49 PM To: Mark Nigh Cc: ceph-devel@xxxxxxxxxxxxxxx Subject: Re: MDS Stuck In Replay On Tue, 21 Jun 2011, Mark Nigh wrote: > I am currently testing Ceph on Debian v6.0 (2.6.32) on the following system: > > 2 servers each with 2 HDD and 1 osd per HDD for a total of 4 OSD > The 1st server has a one (1) mds and mon. > > Ceph gets built and functioning correctly but I think it is when I > change my crushmap so that no data is stored on a single server that I > get my mds in replay. I also get some pgs in peering. See below. If there are PGs in peering it will block MDS replay, so that's the core issue. Can you include the 'ceph osd dump -o -' output so we can verify the object pools are using the data and metadata crush rules? You can also verify that the individual crush rules are working with something like $ ceph osd getcrushmap -o cm 2011-06-21 15:44:09.434742 mon <- [osd,getcrushmap] 2011-06-21 15:44:09.435168 mon0 -> 'got crush map from osdmap epoch 2' (0) 2011-06-21 15:44:09.435370 7f8fd5b22720 wrote 320 byte payload to cm $ crushtool -i cm --test --rule 0 devices weights (hex): [10000] rule 0 (data), x = 0..9999 device 0: 10000 num results 1: 10000 $ crushtool -i cm --test --rule 1 devices weights (hex): [10000] rule 1 (metadata), x = 0..9999 device 0: 10000 num results 1: 10000 In your case you should see the objects split between 4 OSDs, not the 1 I currently have running on my dev box. sage > > 2011-06-21 16:49:01.551129 pg v1497: 396 pgs: 272 active+clean, 124 peering; 374 MB data, 773 MB used, 11117 GB / 11118 GB avail > 2011-06-21 16:49:01.551817 mds e10: 1/1/1 up {0=0=up:replay} > 2011-06-21 16:49:01.551836 osd e602: 4 osds: 4 up, 4 in > 2011-06-21 16:49:01.551898 log 2011-06-21 16:41:46.966074 mon0 10.6.1.80:6789/0 1 : [INF] mon.0@0 won leader election with quorum 0 > 2011-06-21 16:49:01.551947 mon e1: 1 mons at {0=10.6.1.80:6789/0} > > My crushmap is as follows: > > # begin crush map > > # devices > device 0 device0 > device 1 device1 > device 2 device2 > device 3 device3 > > # types > type 0 device > type 1 host > type 2 root > > # buckets > host host0 { > id -1 # do not change unnecessarily > # weight 2.000 > alg straw > hash 0 # rjenkins1 > item device0 weight 1.000 > item device1 weight 1.000 > } > host host1 { > id -2 # do not change unnecessarily > # weight 2.000 > alg straw > hash 0 # rjenkins1 > item device2 weight 1.000 > item device3 weight 1.000 > } > root root { > id -3 # do not change unnecessarily > # weight 2.000 > alg straw > hash 0 # rjenkins1 > item host0 weight 1.000 > item host1 weight 1.000 > } > > # rules > rule data { > ruleset 0 > type replicated > min_size 1 > max_size 10 > step take root > step chooseleaf firstn 0 type host > step emit > } > rule metadata { > ruleset 1 > type replicated > min_size 1 > max_size 10 > step take root > step choose firstn 0 type device > step emit > } > > I tried to restart the mds daemon with no luck. Here are a few lines of the mds log. Let me know if there is anything else I can provide. > > 2011-06-21 16:50:44.047829 7f2fa8483700 -- 10.6.1.80:6800/19997 >> 10.6.1.80:6789/0 pipe(0x281d280 sd=10 pgs=1 cs=1 l=1).reader got message 141 0x283e780 mdsbeacon(4221/0 up:replay seq 141 v10) v2 > 2011-06-21 16:50:44.047892 7f2fac6fe700 -- 10.6.1.80:6800/19997 >> 10.6.1.80:6789/0 pipe(0x281d280 sd=10 pgs=1 cs=1 l=1).writer: state = 2 policy.server=0 > 2011-06-21 16:50:44.047918 7f2fac6fe700 -- 10.6.1.80:6800/19997 >> 10.6.1.80:6789/0 pipe(0x281d280 sd=10 pgs=1 cs=1 l=1).write_ack 141 > 2011-06-21 16:50:44.047944 7f2fac6fe700 -- 10.6.1.80:6800/19997 >> 10.6.1.80:6789/0 pipe(0x281d280 sd=10 pgs=1 cs=1 l=1).writer: state = 2 policy.server=0 > 2011-06-21 16:50:44.047969 7f2fa9485700 -- 10.6.1.80:6800/19997 <== mon0 10.6.1.80:6789/0 141 ==== mdsbeacon(4221/0 up:replay seq 141 v10) v2 ==== 103+0+0 (2350252520 0 0) 0x283e780 con 0x283d140 > 2011-06-21 16:50:44.047986 7f2fa9485700 -- 10.6.1.80:6800/19997 dispatch_throttle_release 103 to dispatch throttler 103/104857600 > 2011-06-21 16:50:44.058654 7f2fa8382700 -- 10.6.1.80:6800/19997 --> osd2 10.6.1.81:6800/1826 -- ping v1 -- ?+0 0x2841300 > 2011-06-21 16:50:44.058677 7f2fa8382700 -- 10.6.1.80:6800/19997 --> osd0 10.6.1.80:6801/1891 -- ping v1 -- ?+0 0x2841180 > 2011-06-21 16:50:44.058701 7f2fa6d7a700 -- 10.6.1.80:6800/19997 >> 10.6.1.81:6800/1826 pipe(0x2836a00 sd=9 pgs=7 cs=1 l=1).writer: state = 2 policy.server=0 > 2011-06-21 16:50:44.058730 7f2fa6d7a700 -- 10.6.1.80:6800/19997 >> 10.6.1.81:6800/1826 pipe(0x2836a00 sd=9 pgs=7 cs=1 l=1).writer: state = 2 policy.server=0 > 2011-06-21 16:50:44.058765 7f2fa727f700 -- 10.6.1.80:6800/19997 >> 10.6.1.80:6801/1891 pipe(0x281dc80 sd=7 pgs=6 cs=1 l=1).writer: state = 2 policy.server=0 > 2011-06-21 16:50:44.058810 7f2fa727f700 -- 10.6.1.80:6800/19997 >> 10.6.1.80:6801/1891 pipe(0x281dc80 sd=7 pgs=6 cs=1 l=1).writer: state = 2 policy.server=0 > > Mark Nigh > Systems Architect > Netelligent Corporation > > > > This transmission and any attached files are privileged, confidential or otherwise the exclusive property of the intended recipient or Netelligent Corporation. If you are not the intended recipient, any disclosure, copying, distribution or use of any of the information contained in or attached to this transmission is strictly prohibited. If you have received this transmission in error, please contact us immediately by responding to this message or by telephone (314-392-6900) and promptly destroy the original transmission and its attachments. > -- > 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 > > This transmission and any attached files are privileged, confidential or otherwise the exclusive property of the intended recipient or Netelligent Corporation. If you are not the intended recipient, any disclosure, copying, distribution or use of any of the information contained in or attached to this transmission is strictly prohibited. If you have received this transmission in error, please contact us immediately by responding to this message or by telephone (314-392-6900) and promptly destroy the original transmission and its attachments. -- 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
