#ceph IRC Log


IRC Log for 2011-01-19

Timestamps are in GMT/BST.

[0:19] * Yoric (~David@dau94-10-88-189-211-192.fbx.proxad.net) Quit (Quit: Yoric)
[0:25] <Tv|work> libvirt needs cow support soo bad :-/
[0:25] <Tv|work> the kind of cow where you don't need to promise to never mutate the base image
[0:31] <cmccabe> tv: the qcow2 image format supports copy-on-write, but there are other image formats that don't
[0:32] <cmccabe> tv: I guess if they did add an API it would only work for some configurations
[0:32] <cmccabe> tv: I'm not sure what their philsophy is about stuff like that
[0:49] <Tv|work> cmccabe: qcow2 still requires you make the base image holy; it was no provision for cow when the parent changes
[0:51] <cmccabe> tv: I'm a little bit embarassed to admit this, but in the past I've always just used the raw format so that I can loopback mount the image
[0:55] <cmccabe> tv: I suppose you could always use btrfs :)
[0:56] <cmccabe> tv: to do cow at the filesystem level
[1:31] * verwilst (~verwilst@dD576FAAE.access.telenet.be) Quit (Quit: Ex-Chat)
[1:42] * greglap (~Adium@ has joined #ceph
[2:17] <Tv|work> whee autotest stuff working
[2:17] <Tv|work> next up, pile more vms and see how it allocates them
[2:26] <cmccabe> :)
[2:31] * DJLee (82d8d198@ircip1.mibbit.com) has joined #ceph
[2:31] * greglap (~Adium@ Quit (Read error: Connection reset by peer)
[2:36] * fzylogic (~fzylogic@ Quit (Quit: DreamHost Web Hosting http://www.dreamhost.com)
[2:37] * cmccabe (~cmccabe@ has left #ceph
[2:41] * joshd (~joshd@ip-66-33-206-8.dreamhost.com) Quit (Quit: Leaving.)
[2:42] * Tv|work (~Tv|work@ip-66-33-206-8.dreamhost.com) Quit (Ping timeout: 480 seconds)
[3:10] * bchrisman (~Adium@70-35-37-146.static.wiline.com) Quit (Quit: Leaving.)
[5:53] * bchrisman (~Adium@c-24-130-226-22.hsd1.ca.comcast.net) has joined #ceph
[6:08] * ghaskins (~ghaskins@66-189-113-47.dhcp.oxfr.ma.charter.com) Quit (Quit: Leaving)
[6:31] * greglap (~Adium@cpe-76-90-239-202.socal.res.rr.com) has joined #ceph
[6:53] * ijuz__ (~ijuz@p4FFF607D.dip.t-dialin.net) Quit (Ping timeout: 480 seconds)
[7:02] * ijuz__ (~ijuz@p4FFF5887.dip.t-dialin.net) has joined #ceph
[7:54] * Yoric (~David@dau94-10-88-189-211-192.fbx.proxad.net) has joined #ceph
[8:27] * MarkN (~nathan@ Quit (Quit: Leaving.)
[9:13] * Yoric (~David@dau94-10-88-189-211-192.fbx.proxad.net) Quit (Quit: Yoric)
[9:37] * joshd (~jdurgin@adsl-75-28-69-238.dsl.irvnca.sbcglobal.net) has joined #ceph
[9:37] * joshd (~jdurgin@adsl-75-28-69-238.dsl.irvnca.sbcglobal.net) Quit ()
[10:08] * Yoric (~David@ has joined #ceph
[10:23] * verwilst (~verwilst@router.begen1.office.netnoc.eu) has joined #ceph
[10:33] * s15y (~s15y@sac91-2-88-163-166-69.fbx.proxad.net) Quit (Ping timeout: 480 seconds)
[11:01] * bchrisman (~Adium@c-24-130-226-22.hsd1.ca.comcast.net) Quit (Ping timeout: 480 seconds)
[11:10] <stingray> huh huh
[11:34] <jantje> hi !
[11:37] * allsystemsarego (~allsystem@ has joined #ceph
[11:40] <stingray> jantje:
[11:40] <stingray> how's going
[11:44] * bchrisman (~Adium@c-98-207-207-62.hsd1.ca.comcast.net) has joined #ceph
[12:47] * Yoric (~David@ Quit (Quit: Yoric)
[12:47] * Yoric (~David@ has joined #ceph
[13:00] * Guest3437 (quasselcor@bas11-montreal02-1128535712.dsl.bell.ca) Quit (Remote host closed the connection)
[13:09] * Yoric (~David@ Quit (Quit: Yoric)
[13:15] * bbigras (quasselcor@ has joined #ceph
[13:16] * bbigras is now known as Guest798
[13:50] * ghaskins (~ghaskins@66-189-113-47.dhcp.oxfr.ma.charter.com) has joined #ceph
[14:34] * ghaskins (~ghaskins@66-189-113-47.dhcp.oxfr.ma.charter.com) Quit (Quit: Leaving)
[14:46] <jantje> stingray: good, i'm currently in the process of creating 5x2OSD nodes + 1xMDS+MON
[14:46] <jantje> (and 4 clients)
[15:55] * ghaskins (~ghaskins@66-189-113-47.dhcp.oxfr.ma.charter.com) has joined #ceph
[16:03] * ghaskins (~ghaskins@66-189-113-47.dhcp.oxfr.ma.charter.com) Quit (Quit: This computer has gone to sleep)
[16:11] * ghaskins (~ghaskins@66-189-113-47.dhcp.oxfr.ma.charter.com) has joined #ceph
[16:47] * Meths_ (rift@ has joined #ceph
[16:51] * yx (~yx@28IAAB6KH.tor-irc.dnsbl.oftc.net) has joined #ceph
[16:52] * Meths__ (rift@ has joined #ceph
[16:53] * Meths (rift@ Quit (Ping timeout: 480 seconds)
[16:53] * Meths__ is now known as Meths
[16:55] * Meths_ (rift@ Quit (Ping timeout: 480 seconds)
[17:00] * Yoric (~David@ has joined #ceph
[17:23] * greglap (~Adium@cpe-76-90-239-202.socal.res.rr.com) Quit (Quit: Leaving.)
[17:32] * verwilst (~verwilst@router.begen1.office.netnoc.eu) Quit (Quit: Ex-Chat)
[17:45] * bchrisman (~Adium@c-98-207-207-62.hsd1.ca.comcast.net) Quit (Quit: Leaving.)
[17:49] <wido> jantje: 5x20 OSD's? 5 machines with 20 OSD's? So 100 in total?
[17:53] * greglap1 (~Adium@ has joined #ceph
[17:59] <darkfader> i tihnk it was 5x2
[17:59] <darkfader> and the "O" from OSD
[18:08] * Yoric (~David@ Quit (Ping timeout: 480 seconds)
[18:12] * Yoric (~David@ has joined #ceph
[18:23] * Tv|work (~Tv|work@ip-66-33-206-8.dreamhost.com) has joined #ceph
[18:31] * Yoric (~David@ Quit (Quit: Yoric)
[18:32] <wido> darkfader: Oh, yes...
[18:36] * bchrisman (~Adium@70-35-37-146.static.wiline.com) has joined #ceph
[18:38] * cmccabe1 (~cmccabe@c-24-23-253-6.hsd1.ca.comcast.net) has joined #ceph
[18:40] * greglap1 (~Adium@ Quit (Quit: Leaving.)
[18:44] * joshd (~joshd@ip-66-33-206-8.dreamhost.com) has joined #ceph
[18:59] <darkfader> do you know how often the testing/unstable .debs are updated?
[18:59] <sagewk> when i randomly need one and do it.
[18:59] <darkfader> because i always got old stuff when i tried the apt sources
[18:59] <darkfader> ah
[18:59] <Tv|work> yet one thing on my todo list ;)
[18:59] <darkfader> sagewk: would you mind if i change the wiki text to something closer to that?
[19:00] <sagewk> hope to set up nightly autobuilds of the packages on all branches of interest soon...
[19:00] <sagewk> yeah go for it
[19:00] <darkfader> sagewk: i would enjoy helping to set up the autobuilds
[19:00] <darkfader> at least one can do that in ksh
[19:00] <Tv|work> gitbuilder looks pretty good for any kind of "for all the branches and tags" logic, btw
[19:01] <darkfader> oooh sounds fancy
[19:01] <darkfader> but over my head probably ;)
[19:01] <darkfader> i also don't think it matters too much if the binaries are current, it was just misleading to read they are current if they're not
[19:01] <sagewk> currently i'm using the bash scripts in the root of ceph.git
[19:01] <darkfader> building isn't much of a problem
[19:02] <darkfader> so basically you just lack a cronjob? :)
[19:03] <Tv|work> if you give cron a job like that, now you have two problems
[19:04] <darkfader> Tv|work: you mean that there will be a broken distribution once it doesnt build? and it might incomplete?
[19:04] <sagewk> the main problem i have are with the sid pbuilder tarball getting out of sync (and sometimes being broken in the upstream repo)
[19:04] <Tv|work> cron has a tendency to cause things like runs in parallel
[19:04] <darkfader> well then let an admin look at it ;p
[19:05] <darkfader> if i may pun - coders handle servers the way admins code
[19:06] <Tv|work> well i have a background of both, so i'm as disgusted by either side being sloppy ;)
[19:06] <darkfader> hehehe
[19:06] <gregaf> I have to do admin stuff here sometimes
[19:06] <gregaf> it makes my head hurt
[19:06] <darkfader> we should save sage's time for something better anyway
[19:13] <darkfader> if i had vm's with i386+x64 and ubuntu+debian and 10.04-10.10 and lenny+squeeze would that be the needed combinations/
[19:13] <darkfader> i figure crosscompiling would not be too much fun
[19:14] <gregaf> darkfader: needed combinations for what?
[19:15] <darkfader> gregaf: for building all the funny .debs
[19:15] <darkfader> so that the apt sources from http://ceph.newdream.net/wiki/Debian would be current
[19:16] <gregaf> hmm, I'm not sure what .debs we actually build, but I think that would do it
[19:16] <gregaf> though we've had users running older versions of Ubuntu, and of course there's all the other packaging systems in use ;)
[19:17] <darkfader> <- oracle vm user, needing speciail oooooold versions
[19:17] <darkfader> err no, i wont really try running ceph on that :)
[19:19] <darkfader> Tv|work: did you already try gitbuilder?
[19:22] <stingray> does anyone here uses gerrit for something?
[19:22] <darkfader> i'll start trying to make a script that can tear up and down all the build vms. if i manage to do that i'll come back
[19:22] <stingray> s/something/anything/
[19:23] <gregaf> darkfader: sorry, tv's doing interviews now — we have QA candidates in today!
[19:23] <gregaf> I don't think he's tried it yet, though
[19:23] <darkfader> ah cool
[19:23] <darkfader> i read your discussion about the QA people one of the last days
[19:23] <darkfader> was interesting
[19:23] <gregaf> stingray: we've looked at using it but haven't put in the effort to do so yet
[19:23] <darkfader> ok i'll try to get a hold of him some time later. it would be best if i don't mess around with it on my own
[19:24] <gregaf> stingray: also, did Sage talk to you at all about your mds journal problem?
[19:24] <darkfader> bus factor and all that ;)
[19:34] <stingray> gregaf: no
[19:34] <stingray> gregaf: he didn't
[19:34] <gregaf> okay
[19:34] <gregaf> we spent some time looking at it yesterday
[19:35] <gregaf> and best we can tell the OSD lost some commits because there's a big hole where there shouldn't be :/
[19:35] <gregaf> we'd done some work in that area which apparently didn't get enough testing
[19:36] <stingray> gregaf: did it lost them out of order?
[19:37] <gregaf> no, they actually got lost — it isn't something I'm familiar with myself but sage did some refactoring and seems to have introduced a race
[19:37] <stingray> hm
[19:37] <gregaf> yeah :/
[19:37] <gregaf> anyway, we can hack together something to make your cluster come back up if you like
[19:38] <gregaf> just as a one-off, to make it skip the zeroed section
[19:38] <stingray> well, recovery from journal corruption could be nice in any case, no?
[19:39] <gregaf> yeah, and it's on our to-do list but it's not going to happen very quickly
[19:52] <bchrisman> How does ceph support sparse files/thin provisioning? In some basic testing, a sparse file on ceph reports the final offset of a file as its size (via du). I tested by creating a file, seeking to 1GB, and writing 1MB of roughly-random data. The actual storage used by ceph is small (definitely not 1GB), but a 'du' reports the file length rather than the size of the actual contents of the file.
[19:53] <Tv|work> darkfader: i have a gitbuilder for ceph running, will install it on a proper server in the coming days
[20:02] * Yoric (~David@dau94-10-88-189-211-192.fbx.proxad.net) has joined #ceph
[20:03] <stingray> gregaf: I don't know. We can try skipping the damaged part and then see if anything else will appear, if that's useful for debugging
[20:03] <stingray> if it's too much work I'll just regenerate the cluster
[20:04] <gregaf> stingray: I think I've whipped up a simple fix but I need sage to get out of an interview as I'm not certain about one of the interfaces
[20:04] <gregaf> there is useful data after the zeroed portion though I'm not sure if it'll work or not
[20:04] <gregaf> though if you can regenerate the cluster that'll probably be safest in the long run
[20:15] * Meths_ (rift@ has joined #ceph
[20:18] <stingray> ok
[20:18] <stingray> I'll do it tomorrow, I guess
[20:21] * Meths (rift@ Quit (Ping timeout: 480 seconds)
[20:22] * Meths_ is now known as Meths
[20:47] <gregaf> stingray: if you want to try recovery, I think the patch at http://pastebin.com/qkwLyGSY should work for that
[22:02] * tjikkun (~tjikkun@195-240-122-237.ip.telfort.nl) Quit (Remote host closed the connection)
[22:02] * tjikkun (~tjikkun@2001:7b8:356:0:204:bff:fe80:8080) has joined #ceph
[22:02] <wido> hi
[22:03] <wido> the btrfs bug I noticed (#715), should I report this at #btrfs?
[22:09] * MarkN (~nathan@ has joined #ceph
[22:29] <yehudasa> wido: about #727, can you try the patch in http://ceph.newdream.net/git/?p=ceph-client.git;a=patch;h=766fc43973b16f9becb6b7402b3e052dbb84adee?
[22:29] <yehudasa> wido: but you'll have to reboot that client machine
[22:42] * DJLee (82d8d198@ircip1.mibbit.com) Quit (Quit: http://www.mibbit.com ajax IRC Client)
[22:49] * s15y (~s15y@sac91-2-88-163-166-69.fbx.proxad.net) has joined #ceph
[23:40] <bchrisman> Can't tell from the mailing lists… have subdirectory/folder quotas been implemented? And if so, are they supposed to be obeying the user.quota extended attribute such as setfattr -n user.quota -v 30 /fs0/quota_test_dir?
[23:41] <bchrisman> (with fs0 being a mounted ceph filesystem)
[23:41] <sagewk> no, and no.
[23:41] <sagewk> the recursive stats give you the accounting half, but not enforcement.
[23:41] <sagewk> the tcloud guys have some patches to add enforcement, but they're not upstream
[23:42] <bchrisman> ahh ok… thanks
[23:46] * Yoric (~David@dau94-10-88-189-211-192.fbx.proxad.net) Quit (Quit: Yoric)

These logs were automatically created by CephLogBot on irc.oftc.net using the Java IRC LogBot.