Team Chat Logs

September 18, 2009

2009 8
Mo Tu We Th Fr Sa Su
  1 2 3 4 5 6
7 8 9 10 11 12 13
14 15 16 17 18 19 20
21 22 23 24 25 26 27
28 29 30        
<< <   > >>

[02:44:08.065576]<groolot>hi
[02:44:53.253768]<groolot>hi
[02:45:14.382154]<groolot>does somebody knows where are "user preferences" saved on the server side ?
[02:47:02.759316]<osimons>groolot: see session / session_attribute
[02:47:31.034993]<groolot>all right
[02:48:04.362811]<groolot>great
[02:48:09.481268]<groolot>it's on session_attribute
[02:48:16.301065]<groolot>really thank you
[02:48:18.590963]<groolot>++
[02:49:55.939135]<osimons>groolot: no probs. the auth_cookie, session and session_attribute combo makes the user/prefs handling, including indirect preferences like remembering timeline selections+++
[02:50:15.545767]<groolot>ok
[02:50:47.245255]<groolot>then if I modify some settings at the beginning of the project creation
[02:51:13.835707]<groolot>everything should appear fine in the frontend
[02:52:10.283217]<groolot>because I try to make some project completely up-to-play since first connection of the owner
[02:52:54.773561]<groolot>well sorry for that bad explanation
[02:53:00.003166]<groolot>everything's good
[02:53:04.909884]<groolot>thanx again
[04:57:36.585785]<scfe>teo ran out of postgres connections
[04:57:42.053808]<scfe>please restart
[05:01:04.703768]<evil_twin>t.e.o: Ticket #8680 (trac uses uncommon mimetype for javascript) created - <http://trac.edgewall.org/ticket/8680> - Felix Schwarz <felix.schwarz@…>
[05:01:09.706343]<evil_twin>t.e.o: Ticket #8673 (Reporter gets two emails each time his ticket is modified) updated - <http://trac.edgewall.org/ticket/8673#comment:4> - anonymous
[05:18:03.421742]<cboos>hello
[05:22:05.047123]<cboos>osimons/cmlenz: hi, it seems that t.e.o is in trouble again - maybe due to Bitten, as the /build page doesn't show up after more than 5 minutes
[05:29:18.061531]<jborg>cboos: there?
[05:29:23.621828]<cboos>yep
[05:29:52.694103]<cboos>seems like there's another occasion to debug the #7785 issue ;-)
[05:30:17.250308]<cboos>but maybe it's just a problem with the recent Bitten upgrade
[05:32:02.359169]<osimons>cboos: hmmf. i still don't have access either - jonas was waiting for your keys... :-)
[05:32:29.367056]<jborg>osimons: and I just got it 2 minutes ago
[05:32:31.475529]<cboos>yeah, finally sent him those; it's also that I've been extremely busy till now
[05:32:36.788189]<jborg>adding accounts now
[05:32:44.252403]<osimons>i have had the builds page up several times today already (as slaves started running again just recently)
[05:32:57.721911]<cboos>today I have some time, at least for a quick peak at the issue
[05:33:28.954654]<cboos>osimons: btw, it's quite nice to get notification emails about the builds
[05:33:34.782094]<cboos>thanks ;-)
[05:33:56.573264]<osimons>heh. thank matt - it has been there for a long time, but hasn't been working i think. i noticed them now too :-)
[05:35:07.719800]<cboos>I suspect Bitten to put quite some heavy load on the server... t.e.o. is not completely dead, it's just very slow and queries to timeline or wiki page go through
[05:35:23.563034]<cboos>but the query to /build I started say 10 minutes ago is still not finished
[05:35:46.121630]<osimons>see in irc log someone mentioned about 45 mins ago an error with running out of connections
[05:38:38.362993]<cboos>ah btw jonas, even if t.e.o must be dead slow, please try to not restart the server yet, as I'd like to use gdb ...
[05:38:49.634058]<jborg>cboos: can you login now? cboos@lynx.edgewall.com
[05:38:55.317148]<cboos>(on the problematic trac.cgi process if any)
[05:41:08.859996]<jborg>cboos: the bitten slaves are running in a separate low priority virtual server so they will grind to a halt when the main web server is under stress
[05:44:43.419010]<cboos>ok, so I'm logged in, and there are a four trac.fcgi processes taking all the cpu
[05:46:08.132800]<osimons>i'll be more or less around for a couple of hours, but won't be joining the actual debugging party. try to ansewer questions if i can though, if related to bitten
[05:47:30.674825]<cboos>np thanks
[05:48:40.268324]<jborg>cboos, osimons: I've started on a simple t.e.o admin cheat sheet that I'll give you access to as soon as it's not completely empty
[05:51:01.134823]<jborg>cboos: finding anything interesting?
[05:54:56.145573]<rblank>The build slaves are all stalled on the /builds page here, too.
[05:55:17.546250]<cboos>no, not really on 18142, will pick another one
[05:55:59.578528]<lisppaste5>cboos pasted "untitled" at http://paste.lisp.org/display/87276
[05:57:01.267201]<cboos>don't know how to interpret that... and that was for thread 8, the only interesting one (the other threads where blocked in sem_wait, and the thread 0 in select)
[05:57:10.495016]<cboos>hi remy ;-)
[05:57:14.309528]<rblank>Hello!
[05:57:43.341677]<rblank>Your stack trace mostly shows that some modules were compiled without debug info ;-)
[05:58:21.249809]<rblank>Or maybe without frame pointer.
[06:00:05.675276]<cboos>but what about the address, e.g. 0x0000000000448b6b for frame 0
[06:00:21.277627]<cboos>that's close to the 0x0000000000491052 in PyEval_EvalFrameEx of frame 3
[06:00:51.617298]<cboos>doing an "info shared" show that all libraries are loaded at very different addresses
[06:01:01.681827]<cboos>(e.g. 0x00007fd59a69efe0 0x00007fd59a6a1a68 Yes /lib/libnss_dns.so.2)
[06:01:18.960925]<cboos>so isn't frame 0 right in the python interpreter?
[06:01:35.453344]<cboos>(in a static function, for example)
[06:03:58.415420]<rblank>It probably is. But if the code was compiled with --omit-frame-pointer, IIRC gdb sometimes has trouble displaying stack info.
[06:06:04.006975]<cboos>well, that python interpreter has no debugging symbols, so it makes debugging a bit hard...
[06:06:33.450719]<cboos>but there's no lock afaict, putting some break points show the program is doing something
[06:07:10.313728]<jborg>too bad
[06:07:50.489955]<jborg>we really need a better way to reproduce this. Waiting more that a week between incidents isn't very productive...
[06:08:08.517029]<cboos>we should run trac using a debug build
[06:08:54.406881]<cboos>but fortunately today there are several processes that are in trouble, so even now that I gdb killed one, I can look into the others ;-)
[06:10:01.435135]<jborg>true, but debugging this on the official production server in the first place isn't very optimal
[06:10:27.834199]<jborg>cboos: gdb killed it? You should be able to simply disconnect from the process if you want
[06:10:52.966092]<cboos>sure, but I must have done something the process didn't liked
[06:11:01.001106]<RomainD>hello
[06:11:08.950849]<RomainD>is the site down ?
[06:11:17.278274]<RomainD>seems there is a database timeout on it...
[06:11:26.638783]<lisppaste5>cboos pasted "untitled" at http://paste.lisp.org/display/87277
[06:11:58.723498]<cboos>this is an example session, for pid 18145
[06:12:07.062216]<cboos>18145 www-data 20 0 866m 327m 7312 S 47 10.9 93:30.98 trac.fcgi
[06:12:30.081342]<retracile>RomainD: there are ~3 devs debugging t.e.o in here.... ;)
[06:13:10.378085]<cboos>well, not our fault really, the server was hosed, /then/ we try to see what's going on ;-)
[06:13:16.355225]<cboos>not the reverse way round ;-)
[06:13:40.085858]<RomainD>retracile, cboos, thanks, and no problem
[06:13:56.878158]<RomainD>is there any trac sourcecode mirror somewhere ? or even just a repository ?
[06:14:28.697781]<lisppaste5>cboos annotated #87277 "continuing ..." at http://paste.lisp.org/display/87277#1
[06:15:12.476457]<jborg>RomainD: The svn repo should still be available: https://svn.edgewall.org/repos/trac/
[06:16:07.652906]<evil_twin>t.e.o: Ticket #8680 (trac uses uncommon mimetype for javascript) updated - <http://trac.edgewall.org/ticket/8680#comment:1> - cboos
[06:17:09.571247]<cboos>oh, it looked like the thread 18 I was watching completed, was just a normal request... strange
[06:29:15.522429]<RomainD>jborg, thanks
[06:29:44.317804]<cboos>ok, so nothing "unusual" in the trac.fcgi processes, they're all having normally running threads, but we have a high number of "idle in transaction" connections
[06:30:34.361884]<cboos>(40)
[06:31:15.007116]<cboos>75 total connections
[06:34:15.698854]*retracile -> work
[06:42:40.839173]<jborg>I guess a broken/buggy connection pool would display the same symptoms
[06:50:33.217162]*retracile grumbles something derogatory about mornings.
[07:08:07.356624]<boonekamp>hola! seems trac.edgewall.org is very slow
[07:09:50.843585]<cboos>yep, looking into it
[07:09:54.349147]<boonekamp>actually, it's a database connection error)
[07:10:19.841279]<boonekamp>ok!
[07:30:17.034240]<cboos>osimons: there was some *heavy* bitten related activity around 1:00 PM today, that seems to have brought the server to the current state
[07:33:25.111238]<osimons>cboos: hmmf. could try disabling the remaining graphs for now - they seem to be re-enabled after upgrade
[07:34:44.170904]<cboos>no, it has something to do with scheduling builds
[07:34:53.493982]<cboos>for all revisions, for all configurations, then failing
[07:35:00.691459]<cboos>then trying to unschedule everything...
[07:35:06.869771]<osimons>anyway, they are at bitten.report.testing.TestResultsChartGenerator and bitten.report.coverage.TestCoverageChartGenerator (lint should already be disabled)
[07:35:18.349810]<cboos>I've made the relevant logs reachable from ftp.edgewall.org
[07:35:20.466156]<cboos>under tmp
[07:35:30.913335]<osimons>oh. i see - the code in bitten.queue
[07:35:50.985939]<cboos>maybe yes ;-)
[07:36:05.058558]<osimons>populate_builds() is key there
[07:40:46.939226]<evil_twin>t.e.o: Ticket #8680 (trac uses uncommon mimetype for javascript) updated - <http://trac.edgewall.org/ticket/8680#comment:2> - Felix Schwarz <felix.schwarz@…>
[07:42:40.185831]<osimons>i have no time now unfortunately, cboos - packing kids & bags into car for weekend travel. will be sporadically online next couple of days, but no promises... :-)
[07:43:23.914209]<jborg>yui, I've disabled bitten on t.e.o for now since trac was pegging the cpu for some reason
[07:45:46.196729]<evil_twin>t.e.o: Ticket #8293 (Reporter not Receiving Email notifications Until modifies own ticket) updated - <http://trac.edgewall.org/ticket/8293#comment:15> - anonymous
[07:46:14.773309]<cboos>I guess we'll have to live without bitten for some time, unless I find the time to dig into the code this w.e. - unlikely as well :
[07:49:55.455062]<jborg>perhaps but since it has worked before it should probably be safe to leave enabled but since I will go offline in a few minutes it's probably best to disable it for now
[07:49:59.977421]<osimons>not sure what has changed on not either, as i don't think i've touched much of db-related code during my 2 month tenure with the project. the previous running bitten on t.e.o was a revision from sometime in spring (~r575 something)
[07:50:53.818416]<jborg>yeah, I think the only thing that has changed is the amount of data in the database
[07:51:01.518284]<jborg>and perhaps slightly more traffic
[07:51:03.661999]<osimons>could well be
[07:52:16.545156]<osimons>leaving. catch you all later.
[07:53:49.006749]<cboos>bye, don't forget to download the files, if you need something interesting to read during your long week end ;-)
[07:55:13.734648]<ebel>I'm trying to install a plugin, but it's not showing up on the plugin list.
[07:55:35.322134]<ebel>I compiled it locally on python 2.4.6, but the server is running 2.4.3, could that be a problem?
[07:56:07.602847]<jborg>got to run, bbl
[07:58:21.554088]<ebel>ah wait, in compatible trac versions, never mind :P
[07:59:10.949130]<cboos>bye jonas, and thanks again!
[08:35:43.546338]<ebel>Anyone know anything about the tracgantt plugin?
[08:36:27.298025]<ebel>Specifically if I have a trac with lots of tickets already entered, and I've install the tracgantt plugin, I get lots of errors when trying to look at the gantt charts
[08:36:55.093258]<ebel>I think it's cause the trac database has tickets that don't have the information needed for the gantt plugin. hence it causes an error
[08:43:13.755187]<retracile>ebel: yeah, that sounds about right.
[08:43:24.534644]<retracile>ebel: last I looked at that plugin, it didn't work very well.
[08:43:49.167886]<retracile>ebel: it does seem that a lot of people are interested in functionality like what it promises to offer though.
[08:44:21.450872]<ebel>Ah super. :)
[08:44:38.990400]<ebel>retracile: so is there some way to get it to update the existing tickets so that they have the required fields?
[08:45:18.758449]<ebel>It's a shame that the original author has abandoned it, we had to patch it to get it to work with out 0.10 install
[08:45:50.170009]<retracile>ebel: well, you have two options: fix the plugin the right way so that it automatically upgrades the db when installed, OR write a bit of sql to add the fields to all existing tickets
[08:46:04.450519]<retracile>(backups highly recommended)
[08:46:08.579213]<ebel>;)
[08:46:24.417897]<ebel>but no-one has done this already, no?
[08:46:56.506690]<retracile>ebel: not that I know of, but I haven't been looking in the past... ohh... year or so.
[08:47:27.934718]<rblank>cboos: Still around?
[08:47:52.238947]<ebel>retracile: brilliant :)
[11:38:38.683282]<jhammel>when faking a req, does anyone know how to get the permission system on the request?
[11:38:43.701331]<jhammel>req.perm
[12:13:49.206597]<dr_tunes>setting up WikiExportPlugin -- how does one set the default language for this plugin? The default label showing up is "Exportar PDF", I'd prefer either "Export PDF" or simply "PDF"
[12:40:04.797850]<scfe>jhammel: you're writing unit tests? You want to inject some fake permissions?
[12:40:22.713584]<jhammel>scfe: no, for the RepositoryHookSystem
[12:40:30.918112]<jhammel>scfe: figured it out though
[12:40:37.798968]<scfe>jhammel: great :-)
[12:41:25.321705]<jhammel>scfe: if you're curious, r6564 on th...turns out to be pretty simple
[12:42:10.734018]<scfe>jhammel: in which svn?
[12:42:27.542941]<jhammel>trac-hacks.org
[15:35:56.358179]<evil_twin>t.e.o: MediaWiki edited - <http://trac.edgewall.org/wiki/MediaWiki?version=3> - Khopesh
[15:40:56.449908]<evil_twin>t.e.o: Ticket #6062 (MediaWiki-style alternative wiki syntax) updated - <http://trac.edgewall.org/ticket/6062#comment:10> - Khopesh
[15:45:58.815191]<evil_twin>t.e.o: WorkFlow/Examples edited - <http://trac.edgewall.org/wiki/WorkFlow/Examples?version=6> - jimenez@…
[17:59:39.962940]<james_g>In a multi-project Trac environment, are the tickets shared? :)
[18:03:13.980975]<retracile>james_g: if you have one trac environment for multiple projects, you'll probably make the 'components' field correspond to the various projects... or am I misunderstanding your question?
[18:03:35.942418]<retracile>(If you are using tracforge or something, then that changes things considerably...)
[18:04:20.826470]<james_g>Nope.. just a single standalone trac environment.. the ideal situation would be shared tickets
[18:04:27.441972]<james_g>Not sure if the capability is there
[18:04:39.358318]<james_g>Separate SVN
[18:05:39.262213]<retracile>Ah... 0.11 you can only have a single svn repo, but there's multirepo support somewhere...
[18:05:56.691285]<retracile>... might be in 0.12dev, but I'm not sure; I wasn't paying that close of attention to it.
[18:06:03.978722]<james_g>k :) Thanks for the help
[21:50:28.953459]<bracken>Is the only way to change the default CSS for a trac install to change it on the file system?
[22:06:05.895047]<evil_twin>t.e.o: Ticket #5487 (Translation of Trac to Swedish/Svenska [sv_SE]) updated - <http://trac.edgewall.org/ticket/5487#comment:10> - wise@…
[22:21:03.986522]<evil_twin>t.e.o: sv_SE_messages.po.diff attached to Ticket #5487 - <http://trac.edgewall.org/attachment/ticket/5487/sv_SE_messages.po.diff> - wise@…