Tuesday, 2010-07-20

[2010/07/20 00:18:02] @ Log started by gepetto
[2010/07/20 00:18:02] @ jmccune joined channel #puppet-dev
[2010/07/20 00:25:48] @ Quit: jmccune: Ping timeout: 240 seconds
[2010/07/20 02:05:58] @ plathrop is now known as plathrop-away
[2010/07/20 05:13:17] @ Quit: londo_: Read error: Connection reset by peer
[2010/07/20 05:13:48] @ londo_ joined channel #puppet-dev
[2010/07/20 05:40:26] @ beata- joined channel #puppet-dev
[2010/07/20 05:42:10] @ Quit: beata: Ping timeout: 252 seconds
[2010/07/20 06:11:33] @ mla_ joined channel #puppet-dev
[2010/07/20 06:23:38] @ Quit: mla_: Quit: Page closed
[2010/07/20 06:48:54] @ nanliu joined channel #puppet-dev
[2010/07/20 06:54:25] @ notbrien joined channel #puppet-dev
[2010/07/20 07:24:57] @ jmccune joined channel #puppet-dev
[2010/07/20 07:27:49] @ Quit: nanliu: Quit: nanliu
[2010/07/20 07:31:01] @ Quit: jmccune: Ping timeout: 258 seconds
[2010/07/20 08:04:39] @ rcrowley joined channel #puppet-dev
[2010/07/20 08:05:35] @ nanliu joined channel #puppet-dev
[2010/07/20 08:07:31] @ Quit: tmz: Ping timeout: 265 seconds
[2010/07/20 08:09:09] @ jmccune joined channel #puppet-dev
[2010/07/20 08:50:35] <duritong> so I'm having another problem with 2.6
[2010/07/20 08:50:50] @ lak joined channel #puppet-dev
[2010/07/20 08:51:06] <duritong> sometimes compilation takes only 0.01s and the 2.6 client seems to get a valid catalog but which does nothing
[2010/07/20 08:51:09] @ Quit: lak: Client Quit
[2010/07/20 08:51:13] <duritong> it looks like an empty catalog
[2010/07/20 08:51:45] <masterzen> duritong: I got this issue before the manual import was in the same critical section as the automated import
[2010/07/20 08:51:57] @ lak joined channel #puppet-dev
[2010/07/20 08:52:01] <duritong> because it doesn't adjust anything and finishes very fast
[2010/07/20 08:52:10] <duritong> masterzen: I didn't get it with webrick
[2010/07/20 08:52:22] <duritong> and also not with a 25 client
[2010/07/20 08:52:23] <masterzen> duritong: yes, webrick doesn't multithread?
[2010/07/20 08:52:35] <duritong> but with mongrel and 2.6 clients I got the problem
[2010/07/20 08:52:52] <masterzen> duritong: did the final 2.6 resolve the speed issue for you? I didn't had time to check for my testcase
[2010/07/20 08:53:09] <masterzen> duritong: is it with a freshly started master?
[2010/07/20 08:53:57] <duritong> masterzen: no, it resolved the "auto-import again" issue, but not the speed issue that the first compile is 6 times slower compaired to 0.25
[2010/07/20 08:54:19] <duritong> masterzen: it usually starts appearing after a while
[2010/07/20 08:54:52] <duritong> masterzen: and as far as I could tell only with clients that already got once a proper catalog
[2010/07/20 08:54:53] <masterzen> duritong: too bad for the speed issue.
[2010/07/20 08:55:43] <masterzen> duritong: did you dump the transmitted json to see how it looks like
[2010/07/20 08:56:06] <masterzen> duritong: looks like we need a 2.6.1 :)
[2010/07/20 08:56:30] <masterzen> duritong: and the subsequent compilation, are they as fast as 0.25?
[2010/07/20 08:56:30] <duritong> masterzen: no, I couldn't figure out that and it was starting to be 3AM here, so I got to sleep
[2010/07/20 08:56:56] <masterzen> in my case the subsequent compilation is still 3 to 4x slower :(
[2010/07/20 08:57:02] <duritong> masterzen: I think so, but I'm not really sure, at least they are faster
[2010/07/20 08:57:37] <duritong> masterzen: it's a bit difficult, because I had multiple issues and had to sort out all the different symptoms of them to get a proper picture
[2010/07/20 08:57:50] <masterzen> and do your manifests change?
[2010/07/20 08:58:02] <masterzen> I mean can you relate change to the empty catalog behavior?
[2010/07/20 08:58:28] <duritong> hmm, I pushed some things meanwhile, yeah. But I'm not sure I can relate it to the change
[2010/07/20 08:58:37] <duritong> but that could be a reason
[2010/07/20 09:37:54] @ tmz joined channel #puppet-dev
[2010/07/20 09:42:45] @ beata joined channel #puppet-dev
[2010/07/20 09:46:04] <lak> duritong: so it's just the first compile that's slower, not subsequent compiles?
[2010/07/20 09:46:28] @ Quit: beata-: Ping timeout: 248 seconds
[2010/07/20 09:46:44] <lak> because if that's the case, then you can basically guarantee it's a parsing issue, not a compilation issue
[2010/07/20 09:46:50] * lak will be back online in a bit
[2010/07/20 09:47:00] @ Quit: lak: Quit: lak
[2010/07/20 10:10:24] <masterzen> markus: I still see a 2.5x slowdown on compilations with 2.6.0 compared to 0.25.5 :(
[2010/07/20 10:12:12] <duritong> masterzen: I think we should open a ticket to start collecting things about that and to document the issue
[2010/07/20 10:12:26] <masterzen> duritong: you're completely right
[2010/07/20 10:13:55] <duritong> masterzen: I'm currently adding an other bug report, you'd like to start?
[2010/07/20 10:14:36] <masterzen> duritong: I'll be AFK for a couple of hours. Can you log it?
[2010/07/20 10:14:56] <duritong> masterzen: ok I start something
[2010/07/20 10:36:00] <duritong> masterzen: #4302
[2010/07/20 10:36:01] <gepetto> duritong: masterzen: #4302 is http://projects.puppetlabs.com/issues/show/4302 "Puppet - Bug #4302: compile performance on 2.6 - Puppet Labs"
[2010/07/20 10:36:29] * duritong is off for a while as well
[2010/07/20 10:40:44] @ plathrop-away is now known as plathrop
[2010/07/20 10:49:21] @ beata_ joined channel #puppet-dev
[2010/07/20 10:54:43] <masterzen> markus: check #4302 and my findings when you'll have time :)
[2010/07/20 10:54:44] <gepetto> masterzen: markus: #4302 is http://projects.puppetlabs.com/issues/show/4302 "Puppet - Bug #4302: compile performance on 2.6 - Puppet Labs"
[2010/07/20 11:28:30] @ rcrowley_ joined channel #puppet-dev
[2010/07/20 11:30:20] @ Quit: rcrowley: Remote host closed the connection
[2010/07/20 11:34:33] @ Quit: rcrowley_: Ping timeout: 240 seconds
[2010/07/20 11:39:09] @ rcrowley joined channel #puppet-dev
[2010/07/20 11:45:10] @ Quit: beata_: Quit: Lost terminal
[2010/07/20 12:00:18] @ Quit: nanliu: Quit: nanliu
[2010/07/20 12:12:16] @ Quit: jmccune: Read error: Operation timed out
[2010/07/20 13:38:18] @ plathrop is now known as plathrop-away
[2010/07/20 13:43:08] @ plathrop-away is now known as plathrop
[2010/07/20 13:45:14] @ lak joined channel #puppet-dev
[2010/07/20 14:09:27] @ Quit: joe-mac: Write error: Connection reset by peer
[2010/07/20 14:12:10] @ jmccune joined channel #puppet-dev
[2010/07/20 14:57:28] @ plathrop is now known as plathrop-away
[2010/07/20 15:41:55] @ Quit: notbrien: Quit: notbrien
[2010/07/20 17:41:02] @ Quit: lak: Quit: lak
[2010/07/20 18:33:04] @ Quit: rcrowley: Quit: rcrowley
[2010/07/20 18:54:22] @ Quit: jmccune: Ping timeout: 246 seconds
[2010/07/20 19:40:48] @ jmccune joined channel #puppet-dev
[2010/07/20 21:59:01] @ Quit: jmccune: Ping timeout: 252 seconds
[2010/07/20 22:03:10] @ quit (okay bye)
[2010/07/20 22:03:32] @ Joined channel #puppet-dev
[2010/07/20 22:03:32] @ Topic is "http://projects.reductivelabs.com"
[2010/07/20 22:03:32] @ Topic set by ReinH!~reinh@li14-106.members.linode.com on Wed Mar 17 16:29:02 -0700 2010
[2010/07/20 22:03:35] @ Mode +cnt by kornbluth.freenode.net
[2010/07/20 22:08:51] @ jmccune joined channel #puppet-dev
[2010/07/20 22:15:46] @ Quit: jmccune: Ping timeout: 265 seconds
[2010/07/20 23:02:35] @ jmccune joined channel #puppet-dev
[2010/07/20 23:07:43] @ rcrowley joined channel #puppet-dev
[2010/07/20 23:37:58] @ Quit: jmccune: Ping timeout: 260 seconds
[2010/07/20 23:42:53] @ Quit: rcrowley: Quit: rcrowley

Generated by irclog2html.py 2.6 by Marius Gedminas - find it at mg.pov.lt!