#ceph IRC Log

Index

IRC Log for 2010-10-15

Timestamps are in GMT/BST.

[1:38] * greglap1 (~Adium@ip-66-33-206-8.dreamhost.com) has joined #ceph
[1:38] * greglap (~Adium@ip-66-33-206-8.dreamhost.com) Quit (Read error: Connection reset by peer)
[2:49] * greglap (~Adium@ip-66-33-206-8.dreamhost.com) has joined #ceph
[2:49] * greglap1 (~Adium@ip-66-33-206-8.dreamhost.com) Quit (Read error: Connection reset by peer)
[2:51] * MarkN (~nathan@59.167.240.178) Quit (Ping timeout: 480 seconds)
[3:01] * MarkN (~nathan@59.167.240.178) has joined #ceph
[3:04] * greglap1 (~Adium@ip-66-33-206-8.dreamhost.com) has joined #ceph
[3:04] * greglap (~Adium@ip-66-33-206-8.dreamhost.com) Quit (Read error: Connection reset by peer)
[5:24] * greglap1 (~Adium@ip-66-33-206-8.dreamhost.com) Quit (Quit: Leaving.)
[5:52] * greglap (~Adium@166.205.138.118) has joined #ceph
[6:46] * s15y (~s15y@sac91-2-88-163-166-69.fbx.proxad.net) Quit (Quit: WeeChat 0.2.6)
[6:53] * f4m8_ is now known as f4m8
[6:59] * s15y (~s15y@sac91-2-88-163-166-69.fbx.proxad.net) has joined #ceph
[6:59] * greglap (~Adium@166.205.138.118) Quit (Read error: Connection reset by peer)
[7:22] * greglap (~Adium@cpe-76-90-74-194.socal.res.rr.com) has joined #ceph
[8:03] * Yoric (~David@dau94-10-88-189-211-192.fbx.proxad.net) has joined #ceph
[8:22] * lidongyang (~lidongyan@222.126.194.154) Quit (Remote host closed the connection)
[8:28] * lidongyang (~lidongyan@222.126.194.154) has joined #ceph
[9:18] * Yoric (~David@dau94-10-88-189-211-192.fbx.proxad.net) Quit (Quit: Yoric)
[9:18] * Yoric (~David@dau94-10-88-189-211-192.fbx.proxad.net) has joined #ceph
[9:27] * Yoric (~David@dau94-10-88-189-211-192.fbx.proxad.net) Quit (Quit: Yoric)
[9:28] * allsystemsarego (~allsystem@188.27.167.113) has joined #ceph
[9:36] * Yoric (~David@dau94-10-88-189-211-192.fbx.proxad.net) has joined #ceph
[10:13] * johnl (~johnl@cpc3-brad19-2-0-cust563.barn.cable.virginmedia.com) has joined #ceph
[11:36] * Yoric (~David@dau94-10-88-189-211-192.fbx.proxad.net) Quit (Quit: Yoric)
[12:18] * julienhuang (~julienhua@pasteur.dedibox.netavenir.com) has joined #ceph
[12:24] * julienhuang (~julienhua@pasteur.dedibox.netavenir.com) Quit (Quit: julienhuang)
[12:27] * Yoric (~David@213.144.210.93) has joined #ceph
[15:46] * johnl (~johnl@cpc3-brad19-2-0-cust563.barn.cable.virginmedia.com) Quit (Quit: bye)
[15:48] * alexxy[home] (~alexxy@79.173.82.178) Quit (Read error: Connection reset by peer)
[15:49] * f4m8 is now known as f4m8_
[15:49] * alexxy (~alexxy@79.173.82.178) has joined #ceph
[17:41] * greglap (~Adium@cpe-76-90-74-194.socal.res.rr.com) Quit (Quit: Leaving.)
[17:51] * greglap (~Adium@166.205.139.90) has joined #ceph
[18:11] * greglap (~Adium@166.205.139.90) Quit (Read error: Connection reset by peer)
[19:04] * joshd (~joshd@ip-66-33-206-8.dreamhost.com) has joined #ceph
[19:20] * Yoric (~David@213.144.210.93) Quit (Quit: Yoric)
[19:44] <yehudasa> wido: are you there?
[21:31] <NoahWatkins> Hey guys..Are empty file creations in Ceph MDS-only operations, or is an initial empty object created on OSD?
[21:31] <sagewk> mds-only
[21:31] <NoahWatkins> Excellent, thx sage
[21:31] <sagewk> np :)
[21:32] <sagewk> hey will you be in santa cruz next week?
[21:32] <NoahWatkins> Yup
[21:33] <NoahWatkins> Your speaking?
[21:34] <sagewk> yeah
[21:34] <sagewk> oh right, looks like you are too
[21:38] <NoahWatkins> Oh yeh =)
[21:41] <sagewk> well i'll see you then.. i'm interested in hearing what everyone is working on these days
[21:43] <NoahWatkins> Yeh cool. It will be fun. See you then!
[23:13] * cmccabe2 (~cmccabe@dsl081-243-128.sfo1.dsl.speakeasy.net) Quit (Quit: Leaving.)
[23:18] <wido> yehudasa: here now
[23:25] <yehudasa> wido: about bug#473, can you verify that it still happens on latest unstable?
[23:26] <yehudasa> wido: (which will be pushed in a minutet)
[23:27] <wido> I haven't seen the bug, but my FS has been down due to the MDS bug (#483)
[23:27] <wido> So I can't test it right now
[23:30] <sagewk> wido: you mean #478?
[23:30] * allsystemsarego (~allsystem@188.27.167.113) Quit (Quit: Leaving)
[23:35] <wido> uh, let me check
[23:36] <wido> yes, sorry, 478
[23:36] <sagewk> ok cool. looking at it now
[23:37] <wido> yehudasa: The RBD doesn't consume that much memory anymore, but right now i'm seeing a I/O stall inside my VM. The rsync of kernel.org has been running for about a few hours now, just got home and checked it, all I/O is blocking
[23:38] <wido> all OSD's are up, no messages in the libvirt qemu log from librados or something
[23:41] <wido> when doing "rados df" I see the RBD pool still growing, and that VM is the only thing online right now
[23:41] <yehudasa> wido: maybe it's just being slow?
[23:42] <wido> well, right now my "sync" is blocking, kernel starts to spit out messages that it is blocking for 120 seconds
[23:43] <wido> it seems really slow indeed, not sure why, but i'll dig a bit further, never mind for now
[23:43] <wido> might be a slow OSD
[23:44] <yehudasa> I think latest qemu rbd fixes make it run slower
[23:46] <wido> Ah, ok. I've been trying to follow the messages on the ml, but didn't really get it
[23:46] <yehudasa> I don't think it was mentioned there
[23:47] <yehudasa> but generally, for every rados io completion we need to signal a qemu thread that can run some cleanup
[23:47] <yehudasa> before we did it from the rados thread context, which apparently is a no-no
[23:48] <wido> yes, I read that, didn't know it would have such an impact
[23:49] <wido> I have been seeing some strange write speeds, one moment I would get 114MB/sec write, the next only 69MB/sec
[23:49] <yehudasa> if it tuns out that this is a major performance hog, we might be able to have some other workaround that will fix that
[23:49] <yehudasa> how much was it before?
[23:49] <wido> stable around 80MB/sec
[23:50] <yehudasa> it can be also because of different qemu version that you're using
[23:50] <wido> yes, that was with 0.12.3, running .5 now
[23:50] <wido> btw, yes, my RBD was really slow, my "sync" just finished, took 20 minutes :)
[23:51] <wido> but i'm not sure about the Ubunt version, there is no qemu-error.h, thus no error_report function and brvd_file_open is called brvd_open
[23:52] <yehudasa> qemu-error is a new thing
[23:52] <yehudasa> it wasn't in the 0.12.x
[23:52] <wido> Yes, but the qemu-kvm at the repo is 0.12.5 too, right?
[23:52] <yehudasa> I think it's 0.13
[23:52] <wido> I checked the tree (rbd branch) and it says 0.12.5 in the VERSION file
[23:52] <wido> so I assumed 0.12.5
[23:54] <yehudasa> actually, it's not 0.13.. it's just a snapshot from the qemu-kvm tree from a while back, but it's not 0.12
[23:55] * sjust (~sam@ip-66-33-206-8.dreamhost.com) has joined #ceph
[23:56] <wido> to get RBD in the kclient I still need the RBD branch? I really don't know anymore which branch I had then I filed #473, since I had libceph, rbd and ceph
[23:56] <sagewk> wido: the unstable has it now.
[23:57] <sagewk> will be pushing upstream when the merge window opens
[23:58] <wido> ok, tnx

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