Team Chat Logs

September 2, 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        
<< <   > >>

[00:05:45.503155]<otaku42>moin
[00:25:36.943180]<Edwards>Anyone can help me with a python crashing with tracd problem?
[00:34:33.070052]<evil_twin>t.e.o: Ticket #8625 (Drop SQLite2 support) updated - <http://trac.edgewall.org/ticket/8625#comment:3> - cboos
[00:52:15.366924]<scfe>Edwards: Probably you should just try asking :-)
[01:00:21.946160]<rjd>Hi. Got my main subversion in dmz, and a copy (rsynced every x minutes) in LAN on the trac/agilo machine. Obviously the pre/post commit scripts don't work, and I rather not place the agilo/trac in dmz. Any work around? Or do I have to run the trac and subversion (master) on the same machine?
[01:32:54.954375]<Edwards>Hi, I just installed svn 1.6.5 python26 and trac on my mac osx. When I run tracd as standalone, I can load the first page fine until it starts loading images then python crashes
[01:33:11.941427]<Edwards>Last line seen in terminal is "Bus error"
[01:34:25.764028]<Edwards>Everything is fine with regards to trac showing contents of svn repo and stuff. Just that for each page I load python will crash n I'll have to restart tracd.
[01:35:29.615182]<Edwards>If anyone has a clue on this issue please let me know here, as I'll be checking irclogs too. Thanks so much in advance
[01:40:51.122533]<rsmarples>hi!
[01:41:08.010628]<rsmarples>I have apache serving static trac content via the htdocs_location option
[01:41:33.865835]<rsmarples>however, it's not serving plugin resources via apache, but instead via trac
[01:41:58.933862]<rsmarples>for example:
[01:42:01.363641]<rsmarples><link rel="stylesheet" href="/trac/css/trac.css" type="text/css" /><link rel="stylesheet" href="/projects/self/chrome/tracfullblog/css/fullblog.css" type="text/css" />
[01:42:11.385895]<rsmarples>how can I fix this?
[01:45:58.867454]<osimons>rsmarples: i use apache rewrite engine to change all chrome urls except /chrome/site
[01:46:27.510052]<osimons>RewriteEngine On
[01:46:43.670087]<osimons>RewriteCond %{REQUEST_URI} !^/project/chrome/site/.*
[01:46:57.234305]<osimons>RewriteRule ^/project/chrome/([^/]+)/(.*) /chrome/$1/$2 [PT]
[01:47:01.378603]<osimons>^^ rsmarples
[01:47:45.109265]<osimons>then use "trac-admin $project deploy" to extract all resources and serve them from /chrome (or whereever)
[01:54:34.337575]<evil_twin>t.e.o: ticket-edit-minor-fixes-r8561.diff attached to Ticket #454 - <http://trac.edgewall.org/attachment/ticket/454/ticket-edit-minor-fixes-r8561.diff> - cboos
[01:57:36.221151]<rsmarples>osimons: ok, thanks. I'll look into that
[01:59:55.401722]<osimons>rsmarples: you can of course include 'site' as well if that is static and you are running a single project - i run a ton of projects, so i have many 'site' dirs and that is why i exclude it (slightly different regexp for parent setup, but same principle)
[02:02:16.242890]<rsmarples>nah, i have a few sites
[02:02:36.343700]<rsmarples>hmmm, the screenshots plugin causes the deploy to crash :.
[02:04:06.905194]<rsmarples>someone needs to re-tool the depoly function as my projects use different plugins
[02:04:35.438541]<rsmarples>osimons: ah! you're the fullblog author ..... any good ways of making it faster?
[02:05:32.446494]<osimons>rsmarples: indexes.. on todo.
[02:09:33.634626]<rsmarples>osimons: a quick look already shows indexes where I would put them .... what other indexes did you have in mind?
[02:11:21.963004]<osimons>that's it really - i started without to see what was needed, but never got around to making the upgrade code adding the indexes...
[02:12:43.946985]<osimons>there is even a ticket for it i think - created quite recently. patch welcome :-)
[02:14:34.538277]<evil_twin>t.e.o: Ticket #454 ([patch]Edit ticket comments) updated - <http://trac.edgewall.org/ticket/454#comment:167> - cboos
[02:14:39.537951]<evil_twin>t.e.o: 454-save-edit-history-r8561.2.patch attached to Ticket #454 - <http://trac.edgewall.org/attachment/ticket/454/454-save-edit-history-r8561.2.patch> - cboos
[02:14:44.538102]<evil_twin>t.e.o: Ticket #8627 (Unable to sarch Contacts) created - <http://trac.edgewall.org/ticket/8627> - anonymous
[02:21:48.470980]<rsmarples>osimons: that's just for the timeline, my speed issue is with the main page
[02:22:28.819191]<rsmarples>it could also be genshi just being slow or the css being served by the cgi
[02:22:36.753279]<rsmarples>(fullblog main page that is)
[02:24:34.574585]<evil_twin>t.e.o: Ticket #8627 (Unable to sarch Contacts) closed - <http://trac.edgewall.org/ticket/8627#comment:1> - eblot
[02:24:39.574307]<evil_twin>t.e.o: Ticket #454 ([patch]Edit ticket comments) updated - <http://trac.edgewall.org/ticket/454#comment:168> - cboos
[02:24:41.178423]<rsmarples>heh, you accepted my rss patch! I wonder why I didn't get the email ... anyway, I'll update the package in pkgsrc-wip
[02:25:50.565666]<snoopotic>do you have a link to a recommended guided how to "style" TRAC?
[02:25:55.789744]<snoopotic>-d
[02:30:19.677644]<rsmarples>snoopotic: http://genshi.edgewall.org/
[02:30:29.863872]<osimons>@wiki TracInterfaceCustomization
[02:30:29.875685]<evil_twin>http://trac.edgewall.org/wiki/TracInterfaceCustomization
[02:31:22.950336]<snoopotic>:/ there I am and I'm very confused ^^
[02:32:51.878421]<rsmarples>osimons: updating from svn rev 61xx to your latest seems to have given a speed bumb somehow
[02:35:36.435772]<osimons>rsmarples: beats me why. can't remember anything performance related, but then can't really remember much without checking logs. so much code everywhere...
[02:37:05.883513]<rsmarples>i know the feeling :)
[02:39:34.614033]<evil_twin>t.e.o: ticket-edit-minor-fixes-r8561.diff attached to Ticket #454 - <http://trac.edgewall.org/attachment/ticket/454/ticket-edit-minor-fixes-r8561.diff> - cboos
[02:44:34.633888]<evil_twin>t.e.o: Ticket #5474 (Translation of Trac to French/Français [fr_FR]) updated - <http://trac.edgewall.org/ticket/5474#comment:32> - princesseryane
[02:49:34.694197]<evil_twin>t.e.o: TracTermsFr edited - <http://trac.edgewall.org/wiki/TracTermsFr?version=47> - princesseryane
[02:54:34.725623]<evil_twin>t.e.o: Ticket #5474 (Translation of Trac to French/Français [fr_FR]) updated - <http://trac.edgewall.org/ticket/5474#comment:33> - eblot
[03:13:38.504222]<rsmarples>osimons: moving fullbog static content out via the re-write seems to have given a minor speed bumb also - thanks!
[03:13:59.149303]*rsmarples wonders if trac will ever be fast in general though
[03:14:08.230961]<rsmarples>i dislike waiting a second or two :)
[03:34:35.069691]<evil_twin>t.e.o: 0.12/TracInstall edited - <http://trac.edgewall.org/wiki/0.12/TracInstall?version=8> - cboos
[03:35:19.870447]<kirean>in a report, can I have changetime formatted as date and time ad named as something specified
[03:35:58.341982]<kirean>like select changetime as modified (but with modified I only get the date)
[03:36:42.169753]<kirean>select changetime as datetime gives me the right format, but an ugly label
[03:39:36.880697]<evil_twin>t.e.o: TracL10N edited - <http://trac.edgewall.org/wiki/TracL10N?version=74> - cboos
[03:49:35.102338]<evil_twin>t.e.o: Ticket #454 ([patch]Edit ticket comments) updated - <http://trac.edgewall.org/ticket/454#comment:169> - rblank
[03:59:37.051943]<evil_twin>t.e.o: Ticket #8598 (HEAD request to `/login` should not redirect) updated - <http://trac.edgewall.org/ticket/8598#comment:3> - osimons
[04:11:40.210977]<kirean>osimons: do you know if I can force a format of a field in a report?
[04:34:35.630161]<evil_twin>t.e.o: PySqlite edited - <http://trac.edgewall.org/wiki/PySqlite?version=43> - cboos
[04:36:10.647870]<rsmarples>osimons: personal_blog setting doees not seem to work
[04:36:26.285772]<rsmarples>also, could we have a knob to disable the browse by time feature
[04:36:55.999555]<rsmarples>?
[04:37:27.783077]<rsmarples>and maybe one for not showing the "creat about post" link
[04:49:35.706043]<evil_twin>t.e.o: Ticket #8598 (HEAD request to `/login` should not redirect) updated - <http://trac.edgewall.org/ticket/8598#comment:4> - rblank
[05:22:42.464886]<osimons>rsmarples: personal_blog settings should really work (it removes authors display++), and as for "create about post" that should only appear if you have BLOG_ADMIN permission (or something like that)
[05:24:37.754110]<evil_twin>t.e.o: Ticket #8598 (HEAD request to `/login` should not redirect) updated - <http://trac.edgewall.org/ticket/8598#comment:5> - osimons
[05:40:43.799002]<rsmarples>osimons: now I feel dumb - I have the personal_blog knob in the wrong section - it was in fullblognotification by mistake - lol
[05:44:25.505402]<stewart_>hello I am trying to add my self as an admin by running the permission add command but the admin tab is not appearing. Even after I logout and login again. How can I debug whats happening here?
[06:10:23.085678]<dunkert`>!st pacopablo
[06:10:23.095091]<evil_twin>pacopablo is idle for 3 days 6 hours 59 minutes and 43 seconds
[06:17:04.043095]<osimons>stewart_: changes should happen on the very next request, so you've done something wrong... try trac-admin $dir permission list to see what is registered
[06:19:40.427671]<evil_twin>t.e.o: Ticket #8628 (Considering revision 0 as oldest revision in TracMercurial.) created - <http://trac.edgewall.org/ticket/8628> - olemis+trac@…
[06:34:39.230128]<evil_twin>t.e.o: Ticket #7062 (Make editing ticket text less accessible) closed - <http://trac.edgewall.org/ticket/7062#comment:2> - cboos
[06:34:44.230724]<evil_twin>t.e.o: Ticket #8295 (rfe: collapsed ticket description modification window) updated - <http://trac.edgewall.org/ticket/8295#comment:6> - cboos
[06:38:36.205616]<olistik1>hello everybody
[06:38:47.833166]<olistik1>I've just installed trac 0.11 and the mercurial plugin
[06:38:55.214328]<olistik1>everything works fine
[06:39:23.810495]<olistik1>I'd like to create a post commit hook, but I'm totally newbie in this
[06:39:58.565424]<olistik1>is there something configurable?
[06:43:58.862917]<olistik1>what I need to do is just executing a shell script for every commit
[06:49:39.338451]<evil_twin>t.e.o: Ticket #8295 (rfe: collapsed ticket description modification window) updated - <http://trac.edgewall.org/ticket/8295#comment:7> - cboos
[06:52:25.502383]*retracile sighs something untoward about mornings.
[06:54:40.330280]<evil_twin>t.e.o: Ticket #454 ([patch]Edit ticket comments) updated - <http://trac.edgewall.org/ticket/454#comment:170> - cboos
[07:12:38.787132]<scfe>rsmarples: what error do you see in trac-admin deploy? I submitted a patch for 0.11 just yesterday
[07:14:46.117316]<rsmarples>scfe: probably not deploy itself - the screenshots plugin wants to be installed twice
[07:14:53.966840]<rsmarples>and deploy won't overwrite
[07:15:32.123198]<scfe>yes, that was my fix yesterday.
[07:15:54.043266]<scfe>we had a similar situation (multiple ITemplateProviders with the same prefix for htdocs)
[07:16:05.682243]<scfe>http://trac.edgewall.org/ticket/8622
[07:16:19.826067]<scfe>I hope we can get it in still for 0.11
[07:16:25.469729]<scfe>it's fixed in 0.12 already
[07:17:35.126286]<rsmarples>nice!
[07:20:05.013615]<dunkert`>Genshi is released under the revised BSD license, which is a liberal open source license
[07:20:08.142231]<dunkert`>thnx dude
[07:20:11.427566]<dunkert`>dudes
[07:20:12.215216]<dunkert`>;]
[07:20:26.205843]<dunkert`>almost got scared about the genshi license
[07:20:42.299777]<dunkert`>but i'm happy to see both trac and genshi are bsd
[07:32:53.591998]<snoopotic>what is the database-call for the timeline? how is it rendered?
[07:33:15.166764]<snoopotic>or where do I find it?
[07:36:01.829669]<rdohms>good morning
[07:36:17.577997]<rdohms>can i add a svn repository to an already existing trac install?
[07:36:22.650966]<snoopotic>or in other words: I have the goal to see in the timeline WHO has changed a comment in a ticket. Maybe someone has an Idea/plugin suggestion
[07:37:02.785506]<snoopotic>rdohms: yep you can :)
[07:37:27.998148]<rdohms>snoopotic: just add reposiroty_dir in trac. ini?
[07:38:02.615296]<snoopotic>lin in your trac env to that svn repository and do a "trac-admin /path/to/your/trac/env resync"
[07:38:10.689511]<snoopotic>yeah.
[07:38:31.299640]<snoopotic>but the repository must be a "local" dir no url.
[07:38:50.217809]<rdohms>yeah i saw that
[07:38:54.441314]<rdohms>let me try it
[07:39:14.729725]<snoopotic>also works with nfs devices :D
[07:40:11.898676]<rdohms>also.. if the repo needs authentication. where can i set this?
[07:43:40.271090]<snoopotic>rdohms: the usual svn-way (keyword: authz). I suggest you to use the same authentication for svn and trac. I don't know what happens if that is different, have never tried.
[07:43:53.347859]<stewart_>osimons: the permission seems to be there
[07:44:47.114608]<rdohms>ok let me try
[07:49:53.787560]<cboos>cmlenz: hello, t.e.o is hanging again ... genshi, bitten etc. are reachable but slow
[07:52:36.147118]<cmlenz>crap
[07:55:38.191609]<scfe>snoopotic: 'who changed a comment' - the timeline is made up by many different plugins
[07:55:50.506923]<scfe>ITimelineEventProvider
[07:57:17.058936]<scfe>snoopotic: check TicketModule.render_timeline_event (horrible code, but still...)
[07:59:26.241614]<snoopotic>heh thanks :|
[07:59:30.334142]<cboos>cmlenz: would be interesting to log the statements for a few days (log_statement='mod' in postgres.conf), and relate that to the stale connections
[08:03:20.131090]<cmlenz>ok all restarted, logging enabled as requested
[08:03:25.598413]<osimons>so much for our hypothesis cboos - though the graphs certainly weren't up to any good either...
[08:03:27.197909]<cmlenz>also pinged jonas
[08:03:57.834204]<rdohms>snoopotic: it owrked.. but when i access the trunk i get 404 Not Found (No handler matched request to %s)
[08:04:47.513890]<cboos>osimons: well this morning t.e.o was very responsive
[08:05:05.585137]<osimons>it was - i noticed that too...
[08:05:13.026898]<snoopotic>rdohms: I dunno what you did.
[08:05:53.627132]*rdohms added the repo and ran resync... the crowse page is ok and it shows latest commits.. but clicking on the trunk link gets me that error above
[08:06:10.451999]<cboos>and yesterday I also looked again at the pool and at #8443. Though I have some ideas how to improve things as noted there, I don't really understand how "IDLE in transaction" connections could accumulate...
[08:06:45.953668]<cboos>cmlenz: thanks!
[08:11:17.390960]<snoopotic>rdohms: what trunk-link? in the browse-window?
[08:11:56.593667]<rdohms>path: browser/trunk
[08:12:43.533273]<snoopotic>can you paste you trac.ini?
[08:12:47.534211]<snoopotic>+r
[08:13:33.186110]<rdohms>hold on.... everything else is working .. let me go over some things
[08:15:08.612196]<osimons>cboos: searched a bit, and someone suggested just doing a I did a search on pg_locks to see what locks the "<IDLE> in transaction" connection was holding onto - should perhaps give some idea as to what subsystem at least?
[08:16:25.070856]*osimons don't use postgres. should stop pretending he's got a clue....
[08:16:52.613485]<rdohms>snoopotic: got it... th downloadable paths was wrong.. trunk did not have * after it
[08:17:18.215551]<snoopotic>yeah just wanted to tell you :D
[08:17:31.289792]<snoopotic>:)
[08:17:32.750013]<snoopotic>default it looks like: downloadable_paths = /trunk, /branches/*, /tags/*
[08:18:35.682128]<rdohms>yeah that was it
[08:18:42.039246]<felixdv>hi there rdohms
[08:18:46.358938]<felixdv>can i help? :)
[08:19:49.174932]<rdohms>felixdv: nah.. got it sorted out i guess
[08:20:52.342972]<snoopotic>:)
[08:20:53.294959]<rdohms>yep... looks sweet
[08:20:58.394150]<rdohms>thanks guys
[08:28:15.722038]<cboos>osimons: well according to Shane, the IDLE thing happens when we don't call ROLLBACK - but we should always do, when connections are put back in the pool. So if this doesn't happen, it just mean we /don't/ put those connections back to the pool, and this might happen if we have long running requests. The server setup can play a role as well (number of connections allowed by process * number...
[08:28:17.214182]<cboos>...of server processes), though I don't know the specifics of *.e.o here.
[08:29:56.686358]<osimons>cboos: how about the combination with fcgi that disconnects abruptly at unknown stages of the request? could that perhaps not get propagated properly to actually return the connection - or will the db wait for timeout if it doesn't 'hear' anything?
[08:36:52.483139]<cboos>ah, t.e.o is using fcgi
[08:40:50.198923]<peeps[work]>is there a way to test tickets on a system and then permanently delete them?
[08:41:54.322612]<peeps[work]>i'm wanting to test out a different workflow
[08:42:25.511555]<peeps[work]>i want a resolved step before things get closed, so the bug opener has a chance to verify
[08:44:36.735079]<snoopotic>delete?
[08:44:54.053547]<snoopotic>I do not recommend to delete tickets. close them is the better way
[08:46:03.199392]<snoopotic>And I suggest you to twrite an UML or something in a graphical way that shows the different ticket states and the transitions (from state to the other state), then you can easily implement that in you trac-configuration
[08:46:20.041924]<peeps[work]>i'd just rather not have a bunch of meaningless test tickets in my system
[08:46:49.433178]<peeps[work]>snoopotic, what sort of tool would you recommend for that?
[08:47:00.153681]<peeps[work]>i don't really know UML
[08:48:05.826868]<snoopotic>peeps[work]: I mean graphics like that on the workflow-wiki-page: http://trac.edgewall.org/wiki/TracWorkflow
[08:48:23.845848]<snoopotic>you can use a pen and paper :D
[08:50:44.279331]<santagada>I changed my python interpreter from 2.5 to 2.6 and not my plugins don't work anymore on trac 0.11.5
[08:51:20.838509]<santagada>I was using python setup.py develop so it should work regardless of python interpreter
[08:51:48.243210]<santagada>but even with bdist_egg and the egg on the plugins directory it does not work
[08:51:59.914118]<santagada>anyone have a guess on why this is happening?
[08:52:35.234693]<peeps[work]>santagada, i think this is what I want: http://i27.tinypic.com/2m6k1s1.jpg
[08:52:45.866433]<peeps[work]>oops, i mean snoopotic
[08:53:05.441924]<santagada>peeps[work]: no I just remembered, this problem is because the plugin depends on json
[08:55:14.072715]<doki_pen>santagada: are you sure it should work regardless of interpreter?
[08:55:24.976436]<snoopotic>peeps[work]: yeah if you have such a graphic its qiet simple :D
[08:55:31.933819]<doki_pen>santagada: setup.py develop creates links in /usr/local/lib/python2.x/site-packages
[08:55:34.771063]<snoopotic>peeps[work]: you may use http://trac-hacks.org/wiki/TypedTicketWorkflowPlugin too
[08:55:51.885173]<doki_pen>santagada: so you will need to reinstall all plugins after upgrade
[08:56:21.767155]<snoopotic>peeps[work]: maybe http://trac-hacks.org/wiki/AdvancedTicketWorkflowPlugin too but you don't need the triage-option if you use typedticket.
[08:56:23.521197]<santagada>doki_pen: I mena setup.py develop -m -d trac_env/plugins
[08:58:33.286925]<peeps[work]>snoopotic, is there any special qualities about the closed state?
[08:58:34.577205]<doki_pen>santagada: not sure, but I'm guessing you are missing some dependency in python
[08:58:39.563542]<doki_pen>santagada: maybe setuptools?
[08:59:04.429730]<santagada>doki_pen: no the only problem is that my plugin depends on simplejson and it is named json on python 2.6
[08:59:10.473127]<santagada>now it is part of the stdlib
[09:00:06.590114]<doki_pen>santagada: ah
[09:00:08.199066]<snoopotic>peeps[work]: Well if the ticket is closed your developer know its closed :D
[09:00:24.065855]<snoopotic>so usual there is no need anymore to touch it :D
[09:00:35.158159]<snoopotic>but I gotta go now. maybe see you later.
[09:00:59.734325]<santagada>wow t.e.o is sloooow
[09:02:55.802928]<santagada>okay, t.e.o is unusable right now... I just wanted to submit a bug report
[09:03:33.900665]<osimons>santagada: don't - known issue. see mailing list. cboos is also looking into it.
[09:04:09.929877]<santagada>osimons: I saw it, but didn't think it was this bad
[09:04:12.178990]<cboos>well, cboos is mainly perplexed by the problem and has no access to t.e.o., so don't expect a miracle ;-)
[09:04:26.886641]<peeps[work]>is there a list i can see of all the built in workflow operations?
[09:04:57.995109]<santagada>peeps[work]: there is one on t.e.o wiki... not that I can get there anymore
[09:05:18.603131]<osimons>cboos: perhaps asking cmlenz really nicely? :-)
[09:05:19.619127]<santagada>osimons: who has access to t.e.o, can I help?
[09:05:53.739022]<peeps[work]>damn, it was just working a minute ago
[09:05:56.106534]<osimons>cboos: i also noticed a blog mentioning that your suggested postgres logs may fill gigabytes of data in a very short time for larger sites... :-)
[09:07:04.614147]<cboos>osimons: I imagine ... but as it seems we'll get another hang soon, maybe we can turn it off afterwards
[09:07:59.749380]<cboos>as for the access on t.e.o, I've already asked jonas in the past, and he was negative.
[09:08:00.653818]<santagada>peeps[work]: http://trac.edgewall.org/wiki/TracWorkflow#BasicTicketWorkflowCustomization
[09:08:20.214175]<santagada>peeps[work]: if you can't access it I can paste the text somewhere
[09:11:18.179131]<osimons>oki, cboos - if that is the way of things. his call after all. i'd still hope there could be some cooperation with access for those that could possibly help getting these matters solved
[09:12:16.639179]<osimons>jborg, where art though? :-)
[09:12:50.302993]<cboos>cmlenz: could it be simply the following: nb trac server processes x 10 > max_connections ?
[09:13:37.615823]<cmlenz>13 trac.fcgi processes right now
[09:14:02.209679]<cmlenz>where's max_connections?
[09:14:19.886196]<cboos>10 is the pool size and max_connections is defined in postgres.conf
[09:14:20.086414]<cmlenz>and why "x 10" ?
[09:14:25.122611]<cmlenz>ah
[09:15:01.819156]<evil_twin>t.e.o: Ticket #3446 (Rate of `database is locked` errors is too high) updated - <http://trac.edgewall.org/ticket/3446#comment:35> - cboos
[09:15:08.335184]<cmlenz>max_connections = 100
[09:15:13.075378]<cmlenz>so 130 > 100
[09:15:13.345502]<cboos>ah ha
[09:16:00.133365]<cboos>knowing that the new pool code could eventually lead to more active (transient) connections than the pool size...
[09:16:04.295328]<cboos>(IIRC)
[09:16:37.665235]<cboos>so you could try to tweak max_connections and/or the TRAC_DB_POOL_SIZE environment variable
[09:17:02.439133]<cmlenz>I can also reduce the max_procs for the lighty fcgi handler
[09:17:14.985165]<cmlenz>but that's already at a lowly 4
[09:17:15.959219]<cboos>or that. But if memory consumption is currently fine, I'd suggest increasing the max_connections.
[09:18:00.434432]<cmlenz>memory consumption looks pretty high to me
[09:18:05.645942]<cmlenz>Mem: 3088528k total, 2739928k used, 348600k free, 17000k buffers
[09:18:17.690640]<cmlenz>:/
[09:18:32.018041]<cmlenz>one trac.fcgi process is using 37.8%
[09:18:40.474013]<cboos>ah ! great !
[09:18:44.810716]<cmlenz>(%MEM)
[09:18:49.137986]<cboos>catch this one with gdb!
[09:19:04.929467]<cmlenz>gimme instructions sir :)
[09:19:26.241218]<cboos>well, it's only worth it if that process stays at this level for some time
[09:19:50.113885]<cboos>otherwise it's likely just another Genshi spike ;-)
[09:20:11.909595]<cmlenz>looks pretty constant so far
[09:20:21.610375]<cmlenz>and who wrote that stupid template engine again :P
[09:20:28.647458]<cboos>aliens?
[09:21:09.389927]<cboos>as it's way beyond human understanding ;-)
[09:21:14.321725]<rsmarples>is there a lot of dislike over genshie because of it's cpu hungry nature?
[09:21:18.274571]<cmlenz> PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
[09:21:18.413911]<cmlenz> 9375 www-data 20 0 2051m 1.1g 4964 S 100 37.8 142:54.16 trac.fcgi
[09:21:45.062061]<cmlenz>so gdb?
[09:21:54.753293]<cboos>yes, that seems to be worth it
[09:22:15.683820]<cboos>just get the path to the python interpreter used, then
[09:22:24.250056]<cboos>gdb /usr/.../bin/python
[09:22:29.865129]<cboos>then:
[09:22:44.082539]<cboos>(gdb) attach 9375
[09:24:21.866591]<cboos>then gdb should interrupt it,
[09:24:22.546590]<cmlenz>heh
[09:24:29.906662]<cboos>simply do "bt" at that point
[09:24:29.937080]<cmlenz>gotta apt-get gdb first
[09:24:34.325966]<cboos>oh no
[09:24:58.199779]<cmlenz>does that not work?
[09:25:04.781244]<cboos>Linux is getting corporate... doesn't even come with gdb installed by default :-/
[09:25:28.165158]<doki_pen>cboos: depends on the linux, I'd guess
[09:25:32.705628]<cboos>well yes, should work, just install it then
[09:25:41.658153]<rsmarples>cboos: i disagree. gdb is not a tool for end users
[09:26:08.658048]<cboos>since when end users are supposed to administrate Linux boxes? I must have missed an episode ;-)
[09:26:11.086323]<cmlenz>well, linux is not a os for end users ;)
[09:26:15.246065]*cmlenz ducks and runs
[09:26:24.046145]<doki_pen>I bet slackware comes with it by default
[09:26:30.193657]<rsmarples>tell that to android users :P
[09:26:47.146755]<rsmarples>netbsd comes with it by default
[09:27:53.330577]<cboos>cmlenz: if that process is still there when you're back installing gdb, then I'd say I'm **really** interested knowing what that process is doing...
[09:30:40.318770]<cmlenz>yeah
[09:30:45.325304]<cmlenz>so far it's still running
[09:31:42.194618]<cmlenz>as is the gdb installation
[09:34:11.934988]<cboos>Well, in any case this seems to be a very serious bug. You didn't notice such a process yesterday (or was that Monday?)
[09:36:39.210938]<cboos>reminds me #7785, also fcgi and CPU caught in a busy loop
[09:37:20.405754]<cmlenz>http://scratchpad.cmlenz.net/9a70413e37fd5b4d21ac2ed9dcd0ae98/
[09:37:37.593647]<cboos>what's the uname -a there btw?
[09:38:07.537362]<cboos>Debian Lenny?
[09:38:08.610291]<cmlenz>Linux lynx 2.6.27-11-server #1 SMP Thu Jan 29 20:13:12 UTC 2009 x86_64 GNU/Linux
[09:38:32.634942]<cmlenz>debian sure, lenny dunno
[09:38:52.517168]<doki_pen>cmlenz: there is a debian file in /etc, me thinks
[09:38:52.753495]<cboos>ok, so now on the gdb prompt, just type: bt
[09:39:48.608872]<doki_pen>also, it seems dumb to use 64 bit os in this instance
[09:39:55.170426]<cboos>cmlenz: and after 'bt', please also paste the backtrace
[09:40:19.322183]<cmlenz>http://scratchpad.cmlenz.net/1d58fe22165ca5bc0a81ed4712db5247/
[09:40:27.731653]<cboos>although seeing "__lll_lock_wait" there /really/ makes me think about another incarnation of #7785
[09:40:34.221163]<cboos>thanks
[09:41:09.886721]<cboos>argh
[09:41:52.997631]<doki_pen>is that what a deadlock looks like?
[09:41:53.526837]<cmlenz>?
[09:44:19.262785]<cboos>we're in Py_Finalize, so we're cleaning up everything, including the pool, but probably it's too late and it was already cleaned-up by the atexit handler of apr
[09:44:42.858472]<cboos>so yes, it's really #7785, I'll reopen it.
[09:44:49.181093]<cmlenz>can I kill that proc now?
[09:45:07.082582]<cboos>sure ;-)
[09:45:12.298533]<cmlenz>k
[09:45:19.362006]<cboos>just "kill" from within gdeb
[09:45:21.494714]<cboos>just "kill" from within gdb
[09:46:30.409898]<cmlenz>so, progress on a "taking over the infrastructure" proposal would be nice
[09:46:39.534171]<cboos>but still - the other things we discussed so far (charts in Bitten, possibility to go beyond > max_connections), those are also valid reasons for hangs
[09:47:11.017291]<cboos>about the infrastructure move, you could say a word on the trac-dev thread, as I asked for some clarifications from your part
[09:47:15.860169]<cmlenz>if I don't hear back from jonas in the next couple of days, I'll give you and osimons access to the server
[09:47:30.322282]<cboos>ah, good!
[09:47:38.006985]<asmodai>boo :P
[09:47:41.218663]<cmlenz>oh I didn't see anything requiring clarification, will look again
[09:47:51.571647]<osimons>cmlenz: i don't want it - i don't have the know-how to make sense of any of this - i'm just following silently here and learning... :-)
[09:47:53.038643]<cboos>(about the clarifications) namely if you were just talking about t.e.o or all of egdewall.org
[09:48:01.202718]<cmlenz>osimons: ok
[09:48:09.449571]<cmlenz>edgewall.org
[09:48:41.286049]<cboos>so taking the whole stuff (all 5 projects) including the domain name itself?
[09:48:49.509098]<cmlenz>yeah
[09:49:05.561205]<doki_pen>I think moving to 32bit os would drastically reduce your memory consumption
[09:49:09.665348]<cmlenz>I don't see how anything inbetween makes sense
[09:49:15.457600]*asmodai eyes doki_pen
[09:49:17.035549]<cboos>Shane Caraveo proposed hosting by ActiveState
[09:49:27.407026]<cboos>what about coderesort? ;-)
[09:49:29.041957]<asmodai>doki_pen: What on earth makes you think that?
[09:49:43.305424]<doki_pen>asmodai: all ints are 64 bit == more memory usage
[09:49:47.862801]<doki_pen>asmodai: do some googling
[09:49:55.541942]<cmlenz>AFAIK the hosting is completely paid for by the adwords stuff, so the financials isn't much of a problem, it's more organizational
[09:50:04.675249]<doki_pen>I switched from slicehost to linode for just that reason
[09:50:07.389647]<asmodai>If you're actually using an ILP64 box yes
[09:50:10.589902]<asmodai>most are LP64
[09:50:12.218242]<doki_pen>slicehost only offers 64
[09:50:27.654502]<cboos>hosting *.edgewall.org without the ads would be nice ;-)
[09:50:42.114419]<cmlenz>true
[09:50:44.130962]<doki_pen>asmodai: I don't know what that means. I only know that in my experiance, 64bit os uses much more memory
[09:50:48.669716]<cmlenz>so then you'd go with a sponsor model
[09:50:58.557947]<doki_pen>for the same exact deployment
[09:51:05.121794]<cboos>why not, indeed
[09:51:05.993852]<asmodai>ILP = integer, long, pointer, and the number refers to the bit size.
[09:51:29.298223]<asmodai>ILP64 -> int, long, ptr are all 64 bits, LP64 -> 32 bits int, 64 bits long and ptr
[09:51:36.637663]<cmlenz>s/you'd/we'd ... I'm not out yet :P
[09:51:36.650745]<evil_twin>cmlenz meant: so then we'd ... I'm not out yet :P go with a sponsor model
[09:51:52.954595]<cmlenz>well not quite evil_twin
[09:51:58.861023]<cboos>cmlenz: we still need you for Genshi 0.6 :P)
[09:51:59.162140]<doki_pen>and do you think slicehost is using ILP64?
[09:52:03.157550]<cboos>after that ... ;-)
[09:52:27.077718]<doki_pen>doesn't a language like python use a significant amount of pointers?
[09:53:05.719580]<doki_pen>and isn't memory addressable in 64 bit blocks anyway?
[09:53:19.649818]<doki_pen>so how could you get 32 bit ints?
[09:53:59.334937]<asmodai>http://en.wikipedia.org/wiki/ILP64#Specific_data_models
[09:54:12.166406]<doki_pen>I'll have to admit that I'm only guessing why. But if you just try it you'll see.
[09:54:31.797327]<asmodai>As it notes most Unix/Linux systems + compilers use LP64
[09:55:27.004839]<asmodai>Well, memory addressing is often done with malloc() and family, which are pointers and use size_t's for specifying size.
[09:56:47.994973]<doki_pen>I could be completely wrong, but isn't a short something the compiler enforces? You must still allocate the int, right?
[09:57:05.933338]<asmodai>Sure, if you allocate enough variables you will start using memory.
[09:57:54.755294]<doki_pen>what I do know, is that I am using about 30% less memory on 32bit, and I have read many reports from others who have similar experiences
[09:58:22.237978]<doki_pen>I have also heard of ppl using 32bit libraries on a 64bit kernel
[09:58:23.955871]<osimons>cboos: my many small-ish projects on sqlite is not quite scaled for single projects of trac size :-) that said, i'd have no problem with helping out & sponsoring something that is community owned and run
[09:58:26.330757]<asmodai>That's scary enough.
[09:59:55.442187]<rsmarples>doki_pen: freebsd uses a 64-bit kernel and a 32-bit userland on sparc64 due to memory reasons
[10:00:14.637763]<rsmarples>on the other hand, netbsd is pure 64 on the same box and doesn't seem to have any memory issues
[10:00:40.692173]<rsmarples>I have sparc64 i use for testing that my code is endian clean
[10:01:21.853994]<asmodai>rsmarples: heh, also a good testbed for bus errors
[10:01:43.486389]<rsmarples>yes, that also :)
[10:02:06.003640]<rsmarples>first time I saw my code SIGBUS I had to look up what that signal meant
[10:02:17.423083]<asmodai>I know that feeling.
[10:07:32.350847]<cmlenz>bbl
[10:19:49.198109]<evil_twin>t.e.o: Ticket #7785 (Trac slow / unreachable objects / database locked) reopened - <http://trac.edgewall.org/ticket/7785#comment:18> - cboos
[10:19:50.789674]<doki_pen>what does that mean? wrong endian?
[10:26:23.438437]<asmodai>Well, if you have a datatype, like an int
[10:26:35.690619]<asmodai>is the most significant bit (MSB) on the left or right side?
[10:33:14.242165]<doki_pen>asmodai: I meant the sigbus
[10:34:47.074680]<evil_twin>t.e.o: Ticket #8629 (Trac should call mimetypes.init() as it is not thread-safe to not do so ...) created - <http://trac.edgewall.org/ticket/8629> - Leonardo Santagada <santagada@…>
[10:37:47.675002]<asmodai>SIGBUS is related to alignment errors, RISC vs CISC
[10:44:07.894181]<doki_pen>asmodai: another reason is that 64bil libraries are bigger, and are loaded into memory
[10:44:15.299394]<doki_pen>that is probably the biggest reason
[10:58:22.882814]<doki_pen>instructions are also bigger on 64bit
[11:01:42.445280]<thm>hi! I'd like to do sth like req.redirect('mailto:foo@bar.tld'), but that doesn't work. any hints?
[11:02:14.995693]<raster>Hi all
[11:02:39.077708]<raster>Is it possible to have more than one bazaar repository in the same tracd?
[11:03:05.866353]<raster>I don't know how to set the second repository with trac-admin
[11:04:39.621228]<doki_pen>raster: that is not supported in any official trac release
[11:05:00.053597]<raster>Ok, thanks
[11:05:02.873251]<doki_pen>raster: there is a multirepo branch that is experimental, but I don't think it supports bzr.
[11:05:09.729628]<doki_pen>raster: I could be wrong though
[11:05:25.263290]<doki_pen>raster: most ppl have one trac project per repo
[11:05:41.555710]<raster>Thanks doki_pen. I'll launch two tracd with one tcp port each one, then.
[11:06:19.357658]<doki_pen>raster: you can have mutli projects with on tracd instanc
[11:06:55.766159]<doki_pen>raster: see --env-parent-dir
[11:07:06.166844]<doki_pen>they must be in the same parent directory though
[11:07:14.685422]<raster>Oh, ok
[11:19:47.901856]<evil_twin>t.e.o: Ticket #7785 (Trac slow / unreachable objects / database locked) updated - <http://trac.edgewall.org/ticket/7785#comment:19> - cboos
[11:24:25.736812]<raster>thanks, but --env-parent-dir is not needed. Just putting both directories worked fine.
[11:24:30.630866]<raster>Thanks anyway! :)
[11:34:48.346324]<evil_twin>t.e.o: Ticket #7785 (Trac slow / unreachable objects / database locked) updated - <http://trac.edgewall.org/ticket/7785#comment:20> - cboos
[11:51:42.726916]<thatch>morning all
[12:09:48.806540]<evil_twin>t.e.o: Ticket #454 ([patch]Edit ticket comments) updated - <http://trac.edgewall.org/ticket/454#comment:171> - Ryan Ollos <ryano@…>
[12:19:51.087991]<evil_twin>t.e.o: Ticket #7785 (Trac slow / unreachable objects / database locked) updated - <http://trac.edgewall.org/ticket/7785#comment:21> - Felix Schwarz <felix.schwarz@…>
[12:42:21.336518]<ronny>hi
[12:42:44.166783]<ronny>who is handling the vcs abstraction stuff of trac?
[13:12:02.062326]<thm>ronny: mainly cboos, I think
[13:12:43.217622]<thm>ronny: am I wrong, or do I know you from another channel ;)
[13:26:49.882770]<ronny>thm: probably true
[13:27:15.735481]<ronny>thm: should be #pylib :P
[13:27:33.425316]<thm>I thought of #mtn
[13:27:43.751194]<ronny>oh, that was ages ago
[13:27:53.705783]<ronny>im on hg since years
[13:28:00.675159]<thm>ah, ok.
[13:28:22.939174]<thm>the mercurial plugin was also written by cboos
[13:28:25.794968]<thm>iirc
[13:28:33.221213]<ronny>thm: when is cboos around here?
[13:28:58.801361]<thm>seldomly. today he was in the (european) afternoon.
[13:29:09.726851]<ronny>im writing a vcs abstracton lib, i'd like to see what he things at some point
[13:30:19.606144]<thm>will this be part of pylib?
[13:30:37.961796]<ronny>no, separate
[13:31:53.122954]<ronny>but i'll most likely implement the same path api's for history inspection and commit building
[13:35:33.537848]<thm>better write him a mail, I'd say.
[13:37:04.830817]<ronny>ok, thanks
[14:34:50.566211]<evil_twin>t.e.o: Ticket #3564 (Alphabetized Bullets) updated - <http://trac.edgewall.org/ticket/3564#comment:4> - ronaldholden
[14:54:53.954783]<evil_twin>t.e.o: Changeset [8562]: 0.12dev: When editing a ticket comment, save the old comment in virtual ... - <http://trac.edgewall.org/changeset/8562> - rblank
[15:11:53.119246]<collink>Can someone point me to how I install TracForge?
[15:19:52.006871]<evil_twin>t.e.o: Ticket #454 ([patch]Edit ticket comments) updated - <http://trac.edgewall.org/ticket/454#comment:172> - rblank
[16:39:55.711969]<evil_twin>t.e.o: Ticket #8630 (Wrong "Adding plugin ..." reported in log file) created - <http://trac.edgewall.org/ticket/8630> - Augusto Roccasalva <augustoroccasalva@…>
[16:41:30.718957]<peeps[work]>is there something i can put into my svn commit messages to have trac link it with a ticket?
[16:44:53.854205]<evil_twin>t.e.o: loader.patch attached to Ticket #8630 - <http://trac.edgewall.org/attachment/ticket/8630/loader.patch> - Augusto Roccasalva <augustoroccasalva@…>
[17:09:54.005351]<evil_twin>t.e.o: Ticket #8631 (ValueError: list.remove(x): x not in list) created - <http://trac.edgewall.org/ticket/8631> - lyma
[17:24:56.164201]<evil_twin>t.e.o: Ticket #8631 (ValueError: list.remove(x): x not in list) closed - <http://trac.edgewall.org/ticket/8631#comment:1> - rblank
[17:32:38.763080]<skyfive>Hi, all
[17:34:54.526436]<evil_twin>t.e.o: Ticket #454 ([patch]Edit ticket comments) updated - <http://trac.edgewall.org/ticket/454#comment:173> - rblank
[18:48:59.743122]<skyfive>I have a question.
[18:49:52.537705]<skyfive>I want to manage multiple git repositories on one trac instance, so how do I edit my trac.ini file?
[18:50:08.602092]<skyfive>I already installed trac-git-plugin
[21:33:19.046781]<pacopablo>skyfive: you need to be running the multirepos branch
[21:33:34.889723]<pacopablo>skyfive: and then you go into the web interface, in the Admin section, and add repositories