Re: [ceph-users] crush chooseleaf vs. choose
'ceph osd crush tunables optimal' or adjust an offline map file via the crushtool command line (more annoying) and retest; I suspect that is the problem. http://ceph.com/docs/master/rados/operations/crush-map/#tunables That solves the bug with weight 0, thanks. But is still get the following distribution: device 0: 423 device 1: 453 device 2: 430 device 3: 455 device 4: 657 device 5: 654 Host with only one osd gets too much data. On Fri, 3 Jan 2014, Dietmar Maurer wrote: In both cases, you only get 2 replicas on the remaining 2 hosts. OK, I was able to reproduce this with crushtool. The difference is if you have 4 hosts with 2 osds. In the choose case, you have some fraction of the data that chose the down host in the first step (most of the attempts, actually!) and then couldn't find a usable osd, leaving you with only 2 This is also reproducible. replicas. With chooseleaf that doesn't happen. The other difference is if you have one of the two OSDs on the host marked out. In the choose case, the remaining OSD will get allocated 2x the data; in the chooseleaf case, usage will remain proportional with the rest of the cluster and the data from the out OSD will be distributed across other OSDs (at least when there are 3 hosts!). I see, but data distribution seems not optimal in that case. For example using this crush map: # types type 0 osd type 1 host type 2 rack type 3 row type 4 room type 5 datacenter type 6 root # buckets host prox-ceph-1 { id -2 # do not change unnecessarily # weight 7.260 alg straw hash 0 # rjenkins1 item osd.0 weight 3.630 item osd.1 weight 3.630 } host prox-ceph-2 { id -3 # do not change unnecessarily # weight 7.260 alg straw hash 0 # rjenkins1 item osd.2 weight 3.630 item osd.3 weight 3.630 } host prox-ceph-3 { id -4 # do not change unnecessarily # weight 3.630 alg straw hash 0 # rjenkins1 item osd.4 weight 3.630 } host prox-ceph-4 { id -5 # do not change unnecessarily # weight 3.630 alg straw hash 0 # rjenkins1 item osd.5 weight 3.630 } root default { id -1 # do not change unnecessarily # weight 21.780 alg straw hash 0 # rjenkins1 item prox-ceph-1 weight 7.260 # 2 OSDs item prox-ceph-2 weight 7.260 # 2 OSDs item prox-ceph-3 weight 3.630 # 1 OSD item prox-ceph-4 weight 3.630 # 1 OSD } # rules rule data { ruleset 0 type replicated min_size 1 max_size 10 step take default step chooseleaf firstn 0 type host step emit } # end crush map crushtool shows the following utilization: # crushtool --test -i my.map --rule 0 --num-rep 3 --show-utilization device 0: 423 device 1: 452 device 2: 429 device 3: 452 device 4: 661 device 5: 655 Any explanation for that? Maybe related to the small number of devices? ___ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
Re: [ceph-users] crush chooseleaf vs. choose
On Mon, 6 Jan 2014, Dietmar Maurer wrote: 'ceph osd crush tunables optimal' or adjust an offline map file via the crushtool command line (more annoying) and retest; I suspect that is the problem. http://ceph.com/docs/master/rados/operations/crush-map/#tunables That solves the bug with weight 0, thanks. But is still get the following distribution: device 0: 423 device 1: 453 device 2: 430 device 3: 455 device 4: 657 device 5: 654 Host with only one osd gets too much data. I think this is just fundamentally a problem with distributing 3 replicas over only 4 hosts. Every piece of data in the system needs to include either host 3 or 4 (and thus device 4 or 5) in order to have 3 replicas (on separate hosts). Add more hosts or disks and the distribution will even out. sage On Fri, 3 Jan 2014, Dietmar Maurer wrote: In both cases, you only get 2 replicas on the remaining 2 hosts. OK, I was able to reproduce this with crushtool. The difference is if you have 4 hosts with 2 osds. In the choose case, you have some fraction of the data that chose the down host in the first step (most of the attempts, actually!) and then couldn't find a usable osd, leaving you with only 2 This is also reproducible. replicas. With chooseleaf that doesn't happen. The other difference is if you have one of the two OSDs on the host marked out. In the choose case, the remaining OSD will get allocated 2x the data; in the chooseleaf case, usage will remain proportional with the rest of the cluster and the data from the out OSD will be distributed across other OSDs (at least when there are 3 hosts!). I see, but data distribution seems not optimal in that case. For example using this crush map: # types type 0 osd type 1 host type 2 rack type 3 row type 4 room type 5 datacenter type 6 root # buckets host prox-ceph-1 { id -2 # do not change unnecessarily # weight 7.260 alg straw hash 0 # rjenkins1 item osd.0 weight 3.630 item osd.1 weight 3.630 } host prox-ceph-2 { id -3 # do not change unnecessarily # weight 7.260 alg straw hash 0 # rjenkins1 item osd.2 weight 3.630 item osd.3 weight 3.630 } host prox-ceph-3 { id -4 # do not change unnecessarily # weight 3.630 alg straw hash 0 # rjenkins1 item osd.4 weight 3.630 } host prox-ceph-4 { id -5 # do not change unnecessarily # weight 3.630 alg straw hash 0 # rjenkins1 item osd.5 weight 3.630 } root default { id -1 # do not change unnecessarily # weight 21.780 alg straw hash 0 # rjenkins1 item prox-ceph-1 weight 7.260 # 2 OSDs item prox-ceph-2 weight 7.260 # 2 OSDs item prox-ceph-3 weight 3.630 # 1 OSD item prox-ceph-4 weight 3.630 # 1 OSD } # rules rule data { ruleset 0 type replicated min_size 1 max_size 10 step take default step chooseleaf firstn 0 type host step emit } # end crush map crushtool shows the following utilization: # crushtool --test -i my.map --rule 0 --num-rep 3 --show-utilization device 0: 423 device 1: 452 device 2: 429 device 3: 452 device 4: 661 device 5: 655 Any explanation for that? Maybe related to the small number of devices? ___ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
Re: [ceph-users] crush chooseleaf vs. choose
Host with only one osd gets too much data. I think this is just fundamentally a problem with distributing 3 replicas over only 4 hosts. Every piece of data in the system needs to include either host 3 or 4 (and thus device 4 or 5) in order to have 3 replicas (on separate hosts). Add more hosts or disks and the distribution will even out. I also thought that, but the same thing also happen with more hosts. ___ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
[ceph-users] Ceph@HOME: the domestication of a wild cephalopod
Hi, The Ceph User Committee is proud to present its first use case :-) http://ceph.com/use-cases/cephhome-the-domestication-of-a-wild-cephalopod/ Many thanks to Alexandre Oliva for this inspiring story, Nathan Regola and Aaron Ten Clay for editing and proofreading and Patrick McGarry for wordpress wizardry. If you know of a Ceph use case, please tell us about it. We will discuss over the phone or via email and write it down for publication at http://ceph.com/use-cases Cheers -- Loïc Dachary, Artisan Logiciel Libre signature.asc Description: OpenPGP digital signature ___ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
Re: [ceph-users] [Rados] How to get the scrub progressing ?
On Mon, Dec 30, 2013 at 11:14 PM, Kuo Hugo tonyt...@gmail.com wrote: Hi all, I have several question about osd scrub. Does the scrub job run in the background automatically? Is it working periodically ? Yes, the OSDs will periodically scrub the PGs they host based on load and the min/max scrub intervals. Need I to trigger scrub or deep-scrub progress ? No. How to know the current scrub progressing ? I don't believe this is reported anywhere in a user-friendly way, but I think you can get info on it out of the osd admin socket perfcounters. How to estimate the done time of scrub ? Likewise. ^ Is there some reason you want to follow this instead of just letting it do its thing? -Greg Software Engineer #42 @ http://inktank.com | http://ceph.com ___ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
Re: [ceph-users] ceph osd perf question
On Fri, Jan 3, 2014 at 2:02 AM, Andrei Mikhailovsky and...@arhont.com wrote: Hi guys, Could someone explain what's the new perf stats show and if the numbers are reasonable on my cluster? I am concerned about the high fs_commit_latency, which seems to be above 150ms for all osds. I've tried to find the documentation on what this command actually shows, but couldn't find anything. I am using 3TB sas drives with 4 osd journals on each ssd. Are the numbers below reasonable for a fairly idle ceph cluster (osd utilisation below 10% on average)? This is a report about flushing data out to the backing-store disk, and fs_commit_latency is generally going to include a syncfs syscall, so 150-600 ms is not unreasonable. The fs_apply_latency (for applying updates to the in-memory filesystem) and the numbers on the journal are the ones to look at. -Greg Software Engineer #42 @ http://inktank.com | http://ceph.com # ceph osd perf osdid fs_commit_latency(ms) fs_apply_latency(ms) 0 1924 1 2654 2 1161 3 1252 4 1661 5 2093 6 1846 7 1422 8 2091 9 1661 10 2161 11 3083 12 1502 13 1251 14 1752 15 1422 16 1504 when the cluster get's a bit busy (osd utilisation below 50% on average) I see: # ceph osd perf osdid fs_commit_latency(ms) fs_apply_latency(ms) 0 551 11 1 284 25 2 517 41 3 492 14 4 625 13 5 309 26 6 6509 7 517 21 8 634 25 9 784 32 10 3927 11 5018 12 602 12 13 467 14 14 476 36 15 451 11 16 383 21 Thanks Andrei ___ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com ___ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
Re: [ceph-users] Ceph Command Prepending None to output on one node (only)
I have a vague memory of this being something that happened in an outdated version of the ceph tool. Are you running an older binary on the node in question? -Greg Software Engineer #42 @ http://inktank.com | http://ceph.com On Sat, Jan 4, 2014 at 4:34 PM, Zeb Palmer z...@zebpalmer.com wrote: I have a small ceph 0.72.2 cluster built with ceph-deploy and running on ubuntu 12.04, this cluster is used as primary storage for my home openstack sandbox. I'm running into an issue I haven't seen before and have had a heck of a time searching for similar issues as None doesn't exactly make a good keyword. On one node, when I run any ceph command that interacts with the cluster, I get the appropriate output, but None is prepended to it. root@os2:/etc/ceph# ceph health None HEALTH_OK root@os2:/etc/ceph# ceph None ceph Again, this only happens on one of the four ceph nodes. I've verified conf files, keys, perms, versions, etc. match on all nodes, no connectivity issues, etc. In fact the ceph cluster is still healthy and working great with only one exception. Cinder-Volume also runs on this node and since None is also getting prepended to json formatted output, Cinder-Volume errors out in _get_mon_addrs() when json decoder chokes on the response from ceph. (I'll probably throw a quick pre-decode band-aid on that method to get Cinder back online until I can correct this) here's my config sans radosgw... although it hasn't changed recently. [global] fsid = 02a4abf4-3659-4525-bfe8-f1f5ea024030 mon_initial_members = fs1,os1,cortex,os2 mon_host = 10.10.3.8,10.10.3.10,10.10.3.7,10.10.3.20 auth_supported = cephx osd_journal_size = 1024 filestore_xattr_use_omap = true public_network = 10.10.3.0/24 cluster_network = 10.10.150.0/24 I've tried everything I can think of, hoping someone here can point out what I'm missing. Thanks zeb ___ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com ___ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
[ceph-users] building librados static library librados.a
Hi ceph-users team, I'm a junior systems developer. I'm developing some applications using librados (librados only rather than the whole Ceph package) from Ceph and it turns out the work of building the librados-only package from the huge Ceph source code would be enormous. All I want is just a static library, librados.a. As far as I know, there is no options in configure script and Makefile to build the static lib only. As I know, I must have object files .o to build the librados.a. There are only four object files in the src/librados/ dir, which are librados.o RadosClient.o IoCtxImpl.o snap_set_diff.o but these 4 file is not enough ( I built it and it failed to work) It would be great if you guys could give me some directions here. I'm looking forward for your response. Thanks and have a nice day ! ___ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
Re: [ceph-users] Ceph Command Prepending None to output on one node (only)
I've (re)confirmed that all nodes are the same build. # ceph --version ceph version 0.72.2 (a913ded2ff138aefb8cb84d347d72164099cfd60) ubuntu package version 0.72.2-1precise I was discussing this with my engineers this morning and a couple of them vaguely recalled that we had run into this on an earlier version of ceph when we were testing it, but no one could recall the circumstances or resolution. In fact, they thought that I had fixed it. :) Since this is my home sandbox cluster I can easily rebuild that node if need be, but I wanted to see if anyone could point me toward a better solution so I don't run into this again. thanks. On Mon, Jan 6, 2014 at 10:07 AM, Gregory Farnum g...@inktank.com wrote: I have a vague memory of this being something that happened in an outdated version of the ceph tool. Are you running an older binary on the node in question? -Greg Software Engineer #42 @ http://inktank.com | http://ceph.com On Sat, Jan 4, 2014 at 4:34 PM, Zeb Palmer z...@zebpalmer.com wrote: I have a small ceph 0.72.2 cluster built with ceph-deploy and running on ubuntu 12.04, this cluster is used as primary storage for my home openstack sandbox. I'm running into an issue I haven't seen before and have had a heck of a time searching for similar issues as None doesn't exactly make a good keyword. On one node, when I run any ceph command that interacts with the cluster, I get the appropriate output, but None is prepended to it. root@os2:/etc/ceph# ceph health None HEALTH_OK root@os2:/etc/ceph# ceph None ceph Again, this only happens on one of the four ceph nodes. I've verified conf files, keys, perms, versions, etc. match on all nodes, no connectivity issues, etc. In fact the ceph cluster is still healthy and working great with only one exception. Cinder-Volume also runs on this node and since None is also getting prepended to json formatted output, Cinder-Volume errors out in _get_mon_addrs() when json decoder chokes on the response from ceph. (I'll probably throw a quick pre-decode band-aid on that method to get Cinder back online until I can correct this) here's my config sans radosgw... although it hasn't changed recently. [global] fsid = 02a4abf4-3659-4525-bfe8-f1f5ea024030 mon_initial_members = fs1,os1,cortex,os2 mon_host = 10.10.3.8,10.10.3.10,10.10.3.7,10.10.3.20 auth_supported = cephx osd_journal_size = 1024 filestore_xattr_use_omap = true public_network = 10.10.3.0/24 cluster_network = 10.10.150.0/24 I've tried everything I can think of, hoping someone here can point out what I'm missing. Thanks zeb ___ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com ___ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
[ceph-users] radosgw swift java jars
Hi, could find all necessary jars required to run the java program. Is there any place to get all jars for both swift and s3? Thanks. ___ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
Re: [ceph-users] building librados static library librados.a
The default configuration for a Ceph build should produce a static rados library. If you actually want to build _only_ librados, that might require a bit automake tweeks. nwatkins@kyoto:~$ ls -l projects/ceph_install/lib/ total 691396 -rw-r--r-- 1 nwatkins nwatkins 219465940 Jan 6 09:56 librados.a -rwxr-xr-x 1 nwatkins nwatkins 1067 Jan 6 09:56 librados.la lrwxrwxrwx 1 nwatkins nwatkins17 Jan 6 09:56 librados.so - librados.so.2.0.0 lrwxrwxrwx 1 nwatkins nwatkins17 Jan 6 09:56 librados.so.2 - librados.so.2.0.0 -rwxr-xr-x 1 nwatkins nwatkins 89043452 Jan 6 09:56 librados.so.2.0.0 On Mon, Jan 6, 2014 at 3:45 AM, david hong davidhong1...@gmail.com wrote: Hi ceph-users team, I'm a junior systems developer. I'm developing some applications using librados (librados only rather than the whole Ceph package) from Ceph and it turns out the work of building the librados-only package from the huge Ceph source code would be enormous. All I want is just a static library, librados.a. As far as I know, there is no options in configure script and Makefile to build the static lib only. As I know, I must have object files .o to build the librados.a. There are only four object files in the src/librados/ dir, which are librados.o RadosClient.o IoCtxImpl.o snap_set_diff.o but these 4 file is not enough ( I built it and it failed to work) It would be great if you guys could give me some directions here. I'm looking forward for your response. Thanks and have a nice day ! ___ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com ___ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
Re: [ceph-users] CephFS files not appearing in DF (or rados ls)
On Thu, Jan 2, 2014 at 2:18 PM, Alex Pearson a...@apics.co.uk wrote: Hi All, Victory! Found the issue, it was a mistake on my part, however it does raise another questions... The issue was: root@osh1:~# ceph --cluster apics auth list installed auth entries: SNIP client.cuckoo key: AQBjTblS4AFAARAAZyumzFyk2JS8d9AjutRoTQ== caps: [mon] allow r caps: [osd] allow rwx pool=staging, allow rwx pool=media2 When I recreated the pool I changed from 'media2' to 'media3' - so there wasn't any authorization to the pool. I've corrected this (see below), then REMOUNTED the filesystem on the client (it didn't work until I'd done this) root@osh1:~# ceph --cluster apics auth caps client.cuckoo osd 'allow rwx pool=media3, allow rwx pool=staging' mon 'allow r' Ref: http://www.sebastien-han.fr/blog/2013/07/26/ceph-update-cephx-keys/ The BIG QUESTION though... The data was being stored - I verified this by MD5 summing the data after it was written. But it wasn't being accounted for anywhere and the permissions system looks to have failed. This looks like a big security hole, surely a permissions denied error should have occurred here? Also the data was being stored, but didn't appear in any DF commands, and couldn't be seen using 'ceph --cluster name ls -p pool name'? You aren't doing quite what you think here. What's actually happened is that the data was locally buffered for writing out, and when you read it (for the md5 sum) it was looking at that in-memory state. The CephFS client was then trying to flush that dirty file data out to the OSDs, and getting EPERM back. This is a pretty tricky situation for us to handle, and our (lack of a) solution right now is definitely not great. Unfortunately it's also about all that the POSIX spec lets us do — if you try and do a flush or a sync you would get back an error code, but short of that we have no mechanism for communicating to the user that they can't write to the place they're trying to write to. We've toyed with some sort of pre-emptive check that the user can write to the location their file is stored in (and returning an error on open if they can't), but it's actually quite a hard problem and hasn't gotten any serious attention yet. -Greg Software Engineer #42 @ http://inktank.com | http://ceph.com ___ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
Re: [ceph-users] How can I set the warning level?
On Wed, Dec 25, 2013 at 6:13 PM, vernon1...@126.com vernon1...@126.com wrote: Hello, my Mon's always HEALTH_WARN, and I run ceph health detail, it show me like this: HEALTH_WARN mon.2 addr 192.168.0.7:6789/0 has 30% avail disk space -- low disk space! I want to know how to set this warning level? I have to made it give out the alarm with the space remaining no more than 10%. There's not any option to disable specific ceph health warnings. You can change the threshold at which that warning occurs with the mon data avail [warn|crit] config options, though (default to 30 and 5). -Greg Software Engineer #42 @ http://inktank.com | http://ceph.com ___ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
Re: [ceph-users] cannot see recovery statistics + pgs stuck unclean
[Hrm, this email was in my spam folder.] At a quick glance, you're probably running into some issues because you've got two racks of very different weights. Things will probably get better if you enable the optimal crush tunables; check out the docs on that and see if you can switch to them. -Greg Software Engineer #42 @ http://inktank.com | http://ceph.com On Fri, Dec 27, 2013 at 3:58 AM, Sreejith Vijayendran sreejith.vijayend...@inmobi.com wrote: Hello, We have a 3 node cluster set up with OSDs created on all the 3 nodes. The replication was set to 2. 1 We are in testing phase and tried to bring down all the OSDs in a particular node and was testing the migration of PGs to other OSDs. But the PGs are not getting replicated on other OSDs and the status of the replication also was not clear at all. Below is the ceph status at that point: === sreejith@sb1001:/var/run/ceph$ sudo ceph status cluster 9b48b60c-bebe-4714-8a61-91ca5b388a17 health HEALTH_WARN 885 pgs degraded; 885 pgs stuck unclean; recovery 59/232 objects degraded (25.431%); 22/60 in osds are down monmap e2: 3 mons at {sb1001=10.2.4.90:6789/0,sb1002=10.2.4.202:6789/0,sb1004=10.2.4.203:6789/0}, election epoch 22, quorum 0,1,2 sb1001,sb1002,sb1004 osdmap e378: 68 osds: 38 up, 60 in pgmap v4490: 1564 pgs, 25 pools, 1320 MB data, 116 objects 8415 MB used, 109 TB / 109 TB avail 59/232 objects degraded (25.431%) 679 active+clean 862 active+degraded 23 active+degraded+remapped === We waited for around 4-5 hours and the status just increased marginally to (25.431%) from (27%) at the start. 2 we then tweaked some OSD values to speed up the recovery namely(osd_recovery_threads, osd_recovery_max_active, osd_recovery_max_chunk, osd_max_backfills, osd_backfill_retry_interval etc) as we were only concerned about getting the OSDs rebalanced as of now. But this didnt improve at all all over the night till the morning. 3 We then manually started all the OSDs in that specific node and the status came back up: But then we could see that there were 23PGs stuck unclean and in 'active+remapped state' = sreejith@sb1001:~$ sudo ceph status [sudo] password for sreejith: cluster 9b48b60c-bebe-4714-8a61-91ca5b388a17 health HEALTH_WARN 23 pgs stuck unclean monmap e2: 3 mons at {sb1001=10.2.4.90:6789/0,sb1002=10.2.4.202:6789/0,sb1004=10.2.4.203:6789/0}, election epoch 22, quorum 0,1,2 sb1001,sb1002,sb1004 osdmap e382: 68 osds: 61 up, 61 in pgmap v4931: 1564 pgs, 25 pools, 1320 MB data, 116 objects 7931 MB used, 110 TB / 110 TB avail 1541 active+clean 23 active+remapped = The 'pg dump_stuck unclean' was showing that all the PGs were on 4 OSDs and no other PG were on those same OSDs. SO: 4 we took those OSDs out of the cluster using 'ceph osd out {id}'. Then the unclean PG number increased to 52. Even after making the OSDs back 'IN', the situation didn't improve. = root@sb1001:/home/sreejith# ceph health detail HEALTH_WARN 52 pgs stuck unclean pg 9.63 is stuck unclean since forever, current state active+remapped, last acting [47,7] pg 11.61 is stuck unclean since forever, current state active+remapped, last acting [47,7] pg 10.62 is stuck unclean since forever, current state active+remapped, last acting [47,7] pg 13.5f is stuck unclean since forever, current state active+remapped, last acting [47,7] pg 15.5d is stuck unclean since forever, current state active+remapped, last acting [47,7] pg 14.5e is stuck unclean since forever, current state active+remapped, last acting [47,7] pg 9.47 is stuck unclean for 530.594604, current state active+remapped, last acting [66,43] pg 7.49 is stuck unclean for 530.594593, current state active+remapped, last acting [66,43] pg 5.4b is stuck unclean for 530.594481, current state active+remapped, last acting [66,43] pg 3.4d is stuck unclean for 530.594449, current state active+remapped, last acting [66,43] pg 11.45 is stuck unclean for 530.594635, current state active+remapped, last acting [66,43] pg 13.43 is stuck unclean for 530.594654, current state active+remapped, last acting [66,43] pg 15.41 is stuck unclean for 530.594695, current state active+remapped, last acting [66,43] pg 6.4a is stuck unclean for 530.594366, current state active+remapped, last acting [66,43] pg 10.46 is stuck unclean for 530.594387, current state active+remapped, last acting [66,43] pg 14.42 is stuck unclean for 530.594422, current state active+remapped, last acting [66,43] pg 4.4c is stuck unclean for 530.594341, current state active+remapped, last acting [66,43] pg 12.44 is stuck unclean for 530.594361, current state active+remapped, last acting [66,43] pg 8.48 is stuck unclean for 530.594294, current state
Re: [ceph-users] What's the status of feature: S3 object versioning?
On Thu, Jan 2, 2014 at 12:40 AM, Ray Lv ra...@yahoo-inc.com wrote: Hi there, Noted that there is a Blueprint item about S3 object versioning in radosgw for Firefly at http://wiki.ceph.com/Planning/Blueprints/Firefly/rgw%3A_object_versioning And Sage has announced v0.74 release for Firefly. Do you guys know the status of this feature? It was a community blueprint and got some discussion (http://wiki.ceph.com/Planning/CDS/CDS_Firefly links to http://www.youtube.com/watch?v=DWK5RrNRhHUfeature=sharet=1h30m00s and http://pad.ceph.com/p/cdsfirefly-object-versioning), but nobody was able to contribute developers to it. -Greg Software Engineer #42 @ http://inktank.com | http://ceph.com ___ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
Re: [ceph-users] radosgw package - missing deps on Ubuntu 13.04
On 1/2/14, 1:42 PM, Sage Weil s...@inktank.com wrote: The precise version has a few annoying (though rare) bugs, and more importantly does not support caching properly. For clusters of any size this can become a performance problem, particularly when the cluster is stressed (lots of OSDs catching up on OSDMaps). What¹s this about poor ceph performance on ubuntu 12.04? As a LTS release, I was leaning towards using it for a production cluster. Thanks, JL ___ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
Re: [ceph-users] repair incosistent pg using emperor
Did the inconsistent flag eventually get cleared? It might have been you didn’t wait long enough for the repair to get through the pg. David Zafman Senior Developer http://www.inktank.com On Dec 28, 2013, at 12:29 PM, Corin Langosch corin.lango...@netskin.com wrote: Hi Sage, Am 28.12.2013 19:18, schrieb Sage Weil: ceph pg scrub 6.29f ...and see if it comes back with errors or not. If it doesn't, you can What do you mean with comnes back with error or not? ~# ceph pg scrub 6.29f instructing pg 6.29f on osd.8 to scrub But the logs don't show any scrubbing. In fact the command doesn't see to do anything at all ceph pg repair 6.29f to clear the inconsistent flag. ~# ceph pg repair 6.29f instructing pg 6.29f on osd.8 to repair Again, nothing in the logs. It seems the commands are completely ignored? I already restarted osd 8 two times and tried again, no change... Corin ___ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com ___ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
Re: [ceph-users] Current state of OpenStack/Ceph rbd live migration?
On Tue, Jan 7, 2014 at 6:13 AM, Jeff Bachtel jbach...@bericotechnologies.com wrote: I just wanted to get a quick sanity check (and ammunition for updating from Grizzly to Havana). Per https://blueprints.launchpad.net/nova/+spec/bring-rbd-support-libvirt-images-type it seems that explicit support for rbd image types has been brought into OpenStack/Havana. Does this correspond to live-migration working properly yet in Nova? It doesn't support live-migration now. But if you want a quick fix, you can look at https://review.openstack.org/#/c/56527/. It's provided by my team and use it for months. The ideal live-migration patch merging to master maybe late, we are struggled with refactor tasks now. For background, the nova libvirt driver in Grizzly did not grok how to live migrate for rbd (specifically the need to copy instance folders around, see http://lists.ceph.com/pipermail/ceph-users-ceph.com/2013-March/000536.html ). I'm just curious if this situation is rectified? Thanks, Jeff ___ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com -- Best Regards, Wheat ___ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
[ceph-users] How to deploy ceph with a Debian version other than stable (Hello James Page ^o^)
Hello, I previously created a test cluster using the Argonaut packages available in Debian testing aka Jessie (atm). Since it was pointed out to me that I ought to play with something more recent, I bumped the machines to sid, which has 0.72.2 packages natively. The sid packages do not include ceph-deploy, so I tried mkcephfs, which I was familiar with from the older version. It warned about being deprecated, but seemed to create all the correct config data, keyrings, etc. However when starting ceph after the install it fired up the monitor OK but failed at the first OSD with: --- === osd.2 === Mounting xfs on irt03:/var/lib/ceph/osd/ceph-2 Error ENOENT: osd.2 does not exist. create it before updating the crush map failed: 'timeout 10 /usr/bin/ceph --name=osd.2 --keyring=/etc/ceph/keyring.osd.2osd crush create-or-move-- 2 0.36 root=default host=irt03 ' --- Guess something wasn't created properly after all or not at the place where it was expected to be. So I grabbed the latest ceph-deploy, ceph-deploy_1.3.4-1~bpo70+1_all.deb, and tried that, but as I feared from its name this is expecting nothing newer than Wheezy: --- [ceph_deploy.cli][INFO ] Invoked (1.3.4): /usr/bin/ceph-deploy mon create irt03 [ceph_deploy.mon][DEBUG ] Deploying mon, cluster ceph hosts irt03 [ceph_deploy.mon][DEBUG ] detecting platform for host irt03 ... [irt03][DEBUG ] connected to host: irt03 [irt03][DEBUG ] detect platform information from remote host [ceph_deploy][ERROR ] UnsupportedPlatform: Platform is not supported: debian jessie/sid --- Other than a manual deploy (which I tried before and failed, probably because the documentation for that is not quite in touch with version reality/variety and at points makes assumptions/omits steps), what's left to try? Is this the expected state of affairs, as in: 1. The current Debian Sid package can't create a new cluster by itself 2. ceph-deploy from ceph.com isn't available for Jessie or sid Regards, Christian -- Christian BalzerNetwork/Systems Engineer ch...@gol.com Global OnLine Japan/Fusion Communications http://www.gol.com/ ___ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
Re: [ceph-users] crush chooseleaf vs. choose
I think this is just fundamentally a problem with distributing 3 replicas over only 4 hosts. Every piece of data in the system needs to include either host 3 or 4 (and thus device 4 or 5) in order to have 3 replicas (on separate hosts). Add more hosts or disks and the distribution will even out. I also thought that, but the same thing also happen with more hosts. OK, I have done more test now, and it seems that distribution is quite good if num_hosts = 2*num_repl ___ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
Re: [ceph-users] radosgw swift java jars
I meant could not find required jar files to run java swift program. On Mon, Jan 6, 2014 at 11:35 PM, raj kumar rajkumar600...@gmail.com wrote: Hi, could find all necessary jars required to run the java program. Is there any place to get all jars for both swift and s3? Thanks. ___ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
Re: [ceph-users] radosgw swift java jars
On 01/07/2014 08:15 AM, raj kumar wrote: I meant could not find required jar files to run java swift program. I don't think nobody has a clue of what you mean. Ceph is completely written in C++, so there are no Java JARs. The only piece of Java is the CephFS JNI integration and the RADOS Java bindings. If you have a problem with a Swift tool that requires Java JARs I recommend you ask the maintainers of that project. Wido On Mon, Jan 6, 2014 at 11:35 PM, raj kumar rajkumar600...@gmail.com mailto:rajkumar600...@gmail.com wrote: Hi, could find all necessary jars required to run the java program. Is there any place to get all jars for both swift and s3? Thanks. ___ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com -- Wido den Hollander 42on B.V. Phone: +31 (0)20 700 9902 Skype: contact42on ___ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com