[00:10:06] is anyone else getting $ ssh wmflabs ssh_exchange_identification: Connection closed by remote host? [00:10:29] oh nvm, i see the topic, labs down for maintenance [00:18:11] Coren: Wow, this is taking a long time. [00:19:40] anomie|away: Yeah. :-( [00:20:14] At least we finally end on new NFS hardware. [00:20:59] Is the new hardware "hopefully no more problems", or "definitely no more problems"? [00:23:33] Krinkle: you were looking for me yesterday? [00:27:27] YuviPanda: Yeah, I was looking for someone who can tell me where the php.ini is for tool labs web servers. [00:27:46] I couldn't figure out where it was in puppet. [00:29:37] anomie|away: "hopefully" [00:29:56] anomie|away: we'll know soonish I'd imagine [00:30:09] because we currently have a very silly configuration (as php 5.4 does by default) that makes it impossible (or at least hard) to output gzip pages (e.g. in mediawiki or some other php script) [00:30:15] Krinkle: ah [00:30:33] Krinkle: in toollabs? I thought we had 5.3? [00:30:41] yeah, 5.3 too [00:31:03] is tool labs down? [00:31:08] http://tools.wmflabs.org/ [00:31:09] Krinkle: nfs [00:31:11] replacement [00:31:14] see /topic [00:31:15] ERR_CONNECTION_REFUSED [00:31:21] ah [00:31:23] k [00:31:34] YuviPanda: - https://bugzilla.wikimedia.org/show_bug.cgi?id=55547 [00:32:03] Krinkle: i'm not sure if it is puppetized at all, I bet it's just the default [00:33:03] k, we'll need to either replace the default or add something in conf.d [00:33:20] Krinkle: yeah, should be able to do that easily. [00:33:25] can play around when toollabs is back up [00:33:42] I have no idea where it should go. I know where it could go but not where it should. [00:33:54] Coren: ^ [00:34:11] Krinkle: have you played around with SPDY at all? [00:34:36] YuviPanda: ? [00:34:54] Coren: Krinkle is trying to figure out where in puppet to put php.ini confs for toollabs [00:36:25] It's not puppetized yet, AFAIK. [00:36:31] yeah, what i thought [00:36:56] Coren: I suppose we can make it a conf.d and put it in modules/toollabs/files and add the appropriate rule in modules/toollabs/manifests/webserver.pp [00:36:59] Krinkle: ^ [00:37:16] Sounds about right to me. [00:37:33] for a second i read "trolllabs" [01:08:51] Coren: things up yet? [01:09:17] No. I despair, but I shan't give up. Then I'm going to beat up people who have 20G logs. [01:10:45] Coren: my whole / is less than 1g [01:10:59] That wasn't directed at you. :-) [01:11:08] (Nor at anyone specific, really) [01:11:26] Coren: whomever has 20g log files has something broken [01:11:35] Clearly. [01:11:57] log files shouldnt often exceed 500mb [01:13:51] we should build a central logging system (with logstash) [01:14:04] but then everyone'll bitch and moan about having to modify their code, so why bother? :) [01:14:23] Coren: or they should be stored in /temp and subject to random deletions :P [01:22:07] Hi. [01:22:21] you're here [01:22:23] LABS DOWN FOR PLANNED MAINTENANCE [01:22:25] Yes, I'm here. [01:22:27] Hello. [01:22:28] I'm here. [01:22:33] I checked the mailing list. [01:22:41] Coren: Is there an outage list or something? [01:22:45] You can't make me subscribe to labs-l. [01:22:52] But I might subscribe to actually important shit. [01:23:05] Elsie: hire a secretary [01:23:08] I checked the archives earlier and it said NFS for like 30 minutes to an hour I think. [01:23:11] YuviPanda: I have Coren. [01:23:12] (o; [01:23:30] I was on the server when it apparently went away. [01:23:33] Not a clean break. [01:23:34] Elsie: The advantage of being subscribed is that you'd also have gotten the regular updates. :-) [01:23:35] Coren said there was some unexpected data transfers that are slowing the patching down. [01:23:53] Coren: I get about a 100 wiki e-mails a day as it is. :-( [01:24:03] I'm apparently now on design, in addition to wikimedia-l and wikitech-l. [01:24:07] Have some empathy! [01:24:17] * YuviPanda unsubscribes Elsie from wikimedia-l [01:24:20] that should help [01:24:23] And the two Toolserver lists, of course. [01:24:32] Plus nagging e-mails from the Toolserver. [01:24:51] "Your code sucks. Sincerely, ts-nag-bot" [01:24:55] Every day. [01:26:24] Coren: What about a status site? :-) [01:26:32] http://status.wmflabs.org/ [01:27:10] http://lists.wikimedia.org/pipermail/labs-l/2013-October/001750.html [01:27:21] I wonder if I have gigs of logs. [01:27:58] http://lists.wikimedia.org/pipermail/labs-l/2013-October/date.html#1750 [01:28:02] Oh, I already pasted a link. [01:30:13] See, I planned ahead. I already did a rsync yesterday, thinking that copying "just the touched files" since last night would be swift. [01:30:55] * Coren stares malevolently at nearly 300G of logs, all told, not counting git pulls touching hundreds of thousands of files. [01:31:01] And dumps. [01:31:26] * Elsie takes a dump on the channel floor. [01:31:33] Elsie: You're not helping. [01:31:36] Coren, copy copy copy faster [01:31:38] :P [01:32:37] And the copy is already annoyingly slow because it has to be over the network to avoid triggering the controller stalls. [01:33:06] percentage? [01:33:37] MRX: Not knowable. rsync copies in arbitrary order; but I think I've seen most tool names scroll by by now. [01:33:51] It's currently stuck on a particularily... beefy one. [01:34:00] heh [01:34:52] Not sure what Cyberbot's datefix /does/, but it produced 16G of logs. [01:35:26] rfx-report is "more reasonable" at only 8G of logs. [01:35:44] * Coren makes a note to have a chat with Cyberbot. [01:36:18] haha [01:36:52] The sad thing, of course, is that much of the stuff that takes so long could very well have been nuked. [01:37:45] But like I said earlier, the version of rsync packaged for Precise doesn't support the --ignore-crap option. :-) [01:38:29] Good thing too, I might have an account after it's done. [01:38:33] might not* [01:48:48] Coren: there was a reason I had a quote per user :) [01:49:11] quite a few users were writing out 10G log files because they never truncated them [01:49:15] in the bots project [01:49:18] Ryan_Lane: I have a LART. [01:49:23] LART? [01:49:30] Look it up. :-) [01:49:49] ah, I never use the term luser [01:50:19] It doesn't need to go with luser specifically; it's a BOFH thing. :-) [01:50:51] * Ryan_Lane nods [01:51:49] At one of my past jobs, I had two LARTS on my desk in my office: LART-1, a long stretch of packing foam (think Nerf); and LART-2, a crowbar. Whenever someone was at my desk asking me something silly or because they goofed, I'd make a show of hesitating over which to grab. (I always wacked with LART-1) [01:53:44] The only time I "used" LART-2 was when one of the dev managed to bring down the build box with a fork bomb. I was already hearing them tease the poor dude who had goofed, so I stamped into the dev room with a grimace brandishing LART-2. [01:54:04] It took them about 2 hours to recover from the severe laughter. :-) [01:56:02] (Oh, yeah, I burst in the dev room bellowing "Who was it?" in my best impression of an insane lumberjack on a rampage) [01:56:23] I read that as 'insurance salesman on a rampage' [01:56:28] i've no idea why [01:56:41] * YuviPanda imagines Coren as an Insurance salesman [01:56:46] Are they known for attacking devs with crowbars? :-) [01:57:13] printers maybe [01:57:44] * YuviPanda imagines Coren doing an 'Office' style hardware bashing, but with labstore3 than a printer :) [01:58:03] heh. [01:59:25] * Coren misses LART-2 now. [01:59:27] -rw-rw---- 1 50404 local-50404 282G Oct 10 17:59 /mnt/srv/tools/project/cyberbot/CyberbotII/PCbot.out [01:59:33] Ima kill 'im! [01:59:53] 282G [01:59:56] :| [01:59:58] right [02:03:12] Coren, please don't delete PCbot.out, I've been reading it over the last few weeks and am on the 254th gig [02:03:18] rm $(find /srv/tools/project/cyberbot -name '*.out') [02:03:48] spagewmf: I have no choice; this file alone will take hours to copy. [02:04:43] * Coren feels like a sucker now. [02:04:43] oh nooes, it started to get really gripping around the 247th gig. Did Alice and Bob get back together? [02:04:50] * spagewmf trolling [02:05:13] spagewmf: Non, Eve screwed the relationship up anyways by stealing Alice's love letters. [02:05:32] In the end, it turned out that Alice was actually the US government [02:05:34] and Bob was Elsie [02:05:35] so [02:06:14] spagewmf: hey! you use a standing desk, right? [02:06:25] spagewmf: any resources on how to properly position oneself when using it, etc? [02:07:45] YuviPanda: http://www.amazon.com/Ergotron-WorkFit-C-Sit-Stand-Workstation-24-213-085/dp/B005K96QJU/ Now everyone's using the http://www.focaluprightfurniture.com/ [02:08:09] spagewmf: ah [02:08:18] spagewmf: mine is a stack of books on a dining table :| [02:10:36] YuviPanda: briefly, everyone is different. I like the top of monitor _above_eye level, high above the keyboard, so I like the Ergotron [02:13:29] spagewmf: hmm, right [02:18:58] I sit down. [02:35:40] Sooo tired. Be done already!!! [05:23:55] [bz] (8NEW - created by: 2MZMcBride, priority: 4Unprioritized - 6normal) [Bug 55603] Impose per-user home directory quota in Tool Labs - https://bugzilla.wikimedia.org/show_bug.cgi?id=55603 [09:58:45] [bz] (8NEW - created by: 2Tim Landscheidt, priority: 4Unprioritized - 6trivial) [Bug 48625] Provide namespace IDs and names in the databases similar to toolserver.namespace - https://bugzilla.wikimedia.org/show_bug.cgi?id=48625 [10:38:05] [bz] (8NEW - created by: 2Matthew Flaschen, priority: 4Unprioritized - 6normal) [Bug 55612] mediawiki_singlenode can not run installer due to missing /usr/bin/php - https://bugzilla.wikimedia.org/show_bug.cgi?id=55612 [10:48:16] !log deployment-prep beta is back up :-] [10:48:17] Logged the message, Master [11:01:11] valhallasw: reviewing your change [11:01:14] valhallasw: and amending it :-] [11:01:21] ok :-) [11:01:31] it was almost correct [11:01:39] you just forgot to apply the new template to the project [11:02:27] and had to change the Git command to make sure it is not fetching submodules [11:03:22] submodules should be exempted anyway, as externals is not listed in hte pyflakes command, right? [11:03:26] !jenkins pywikibot-core-pyflakes [11:03:26] https://integration.wikimedia.org/ci/job/pywikibot-core-pyflakes [11:03:28] nothing wrong with not cloning submodules, though [11:03:54] valhallasw: example build https://integration.wikimedia.org/ci/job/pywikibot-core-pyflakes/652/console [11:04:01] 11:03:41 + pyflakes pywikibot scripts tests [11:04:01] 11:03:42 pywikibot/__init__.py:15: 'logging' imported but unused [11:04:02] 11:03:42 pywikibot/__init__.py:18: 'sys' imported but unused [11:04:03] .. [11:04:08] looks good to me [11:04:14] :-) [11:05:06] what was the issue with pfyalkes . already ? [11:05:37] https://integration.wikimedia.org/ci/job/pywikibot-core-pyflakes/651/console [11:05:40] that one has pyflakes . [11:09:43] copy pasted question on https://gerrit.wikimedia.org/r/#/c/86729/ [11:13:21] hashar: it takes ez_setup into account, but it indeed already ignores externals [11:16:37] okkk [11:17:08] valhallasw: job is already updated, I am merging the change [11:23:45] petan, [11:29:33] labs corrupted my bot. :-( [11:35:30] Coren|Sleep, for when you wake up. PLEASE PLEASE shut down the spambot task first. It relies on so many files, that the maintenance almost sent it on a rampage to falsely tag 30000 articles on WIkipedia. It also corrupted the bot. I took 45 minutes to fix itself since the files where providing conflicting data, essentially the bot hung up duiring the recovery process. [11:37:52] Cyberpower678: jstop? [11:38:15] or qdel [11:38:25] I was asleep when this happened. Someone on Wikipedia disabled the runpage thankfully. [11:40:30] Spambot which has probably become very well known by content editors relies on several files during it's run. If it goes missing. The bot may go haywire. [11:54:21] Cyberpower678 [11:54:44] Will Coren|Sleep get that message I wrote above? [11:54:56] which one [11:55:12] The one right above valhallasw [11:55:21] A few lines up [11:55:23] this is all I received in notification window [11:55:25] (11:23:45) petan, [11:55:46] PLEASE PLEASE shut down the spambot task first. It relies on so many files, that the maintenance almost sent it on a rampage to falsely tag 30000 articles on WIkipedia. It also corrupted the bot. I took 45 minutes to fix itself since the files where providing conflicting data, essentially the bot hung up duiring the recovery process. [11:56:06] ok so you need to shutdown some task? [11:56:14] No. [11:56:34] Before ever doing maintenance again. Shut down that task first. [11:56:40] Should be easy to find. [11:56:50] why it happened? [11:56:55] you could also fix your bot [11:56:58] your task should be optimized for such events [11:57:02] giftpflanze +1 [11:57:45] The maintenance caused some files to get damaged. They were outdated which fed the bot wrong information. [11:57:53] There's no defense against that. [11:57:56] but I agree with Cyberpower678 that all tasks should be probably stopped before maintenance [11:58:06] I thought Coren|Sleep did tha [11:58:07] t [11:58:33] I found out because someone filed a bug report on my talk page. [11:58:58] This bot is optimized for crash recover, but is powerless in detecting faulty data. [11:59:25] It does have a runpage though. [11:59:41] So the user disabled my bot while I was asleep. [11:59:47] yes, this is fucked up maintenance I guess [11:59:57] he should have stop all tasks before messing up the fs [12:00:14] maybe just send him a mail [12:00:25] that is more reliable than "Coren: ping" [12:00:26] My bot's the winnner of large log files. :p [12:00:45] howcome [12:00:53] isn't that 990GB logfile guy a winner [12:01:24] Coren|Sleep, said that Cyberbot is generally the winner through it's high volume of large files. [12:02:13] 05:04 < wm-bot3> False [12:02:28] wonders why the bot said that without an obvious reason :) [12:02:50] !false is False [12:02:50] Key was added [12:02:54] !falser [12:02:58] !false [12:02:58] False [12:03:10] petan, wm-bot> False [12:03:30] what [12:03:36] ?? [12:03:41] oh wait mutante False [12:03:49] wtf is going on now [12:15:16] petan: Cyberpower678 tried to suggest a reason for mutante's observation of wm-bot saying 'false' [12:15:41] valhallasw: does it do that? :P [12:15:43] @rss-off [12:15:43] Rss feed has been disabled on channel [12:15:57] ^^^ [12:16:17] That is what causes "False" [12:18:58] petan: it's the rss "False" bug I mentionefd a while ago. [12:19:21] is it reported? [12:20:31] You can reset the rss-style to something usable, but when it false to retrieve, it sets the text to false instead of turning that feed off. [12:20:52] And I'm not sure if I put in a ticket or not. [12:20:54] oh, so it told me "couldnt read RSS feed" [12:20:55] gotcha [12:21:12] that makes sense, i use it on another channel to read various feeds [12:22:08] If you do @info you can see which feed and fix it or disable it correctly. [12:22:48] It'll be the one with false in the text column. [12:33:56] thanks [12:38:49] Cyberpower if a tools needs a clean shutdown before a mainetance operation that is advertized well in advance, you should at /least/ have told me (and ideally stopped it yourself) :-) [12:40:46] Coren|Sleep: there is a file that can't be changed owner of [12:41:10] petrb@tools-dev:/data/project/gabrielchihonglee-bot/public_html/w$ sudo chown local-gabrielchihonglee-bot fuckup [12:41:11] chown: changing ownership of `fuckup': Invalid argument [12:41:33] petan: Is that a new tool? [12:41:45] not sure, it's not mine [12:41:49] zhuyifei1999? [12:46:34] petan: It's just owned by him; he can use take. [12:46:59] Coren: he told me he couldn't? nor sudo chown worked [12:47:26] sudo chown is deprecated anyways [12:48:03] petan: Have you seen https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/Help/NewWeb ? [12:48:17] not yet [12:50:28] Coren: When I tried touching ~/lighttpd.RESTART just now, it dumped some errors about being unable to unlink the pid file. And I'm not sure if it actually restarted. [12:51:07] It should have. The bit about unlinking the pid file is a known issue that has no effect. [12:51:26] Coren: I'm just fixing Gabrielchihonglee's bot test wiki [12:51:26] if the file is gone, and hasn't been replaced by a .STOPPED then it's restarted. [12:52:03] he gave me the right to deal with them [12:52:33] zhuyifei1999: You can use take to chown the file; have you tried it? [12:52:47] anomie: What tool? [12:52:52] already [12:52:54] Coren: File is not gone. 'anomiebot' tool. [12:53:33] anomie: The file's existence is checked every 120s or so. Lemme check. [12:53:39] Coren: already done long ago [12:53:58] zhuyifei1999: Can you try it and tell me the output, please? [12:54:13] Invalid argument [12:54:48] just like petan's [12:54:55] Just "Invalid argument"? [12:55:27] more, but I forgot [12:55:46] I'd appreciate it if you could try it now and paste the result. [12:55:49] &very hard connect today [12:56:31] anomie: Odd; it's really not starting for you. Lemme try and figure out why. [12:57:40] zhuyifei1999@tools-dev:~$ cd /data/project/yifeibot [12:57:50] zhuyifei1999@tools-dev:/data/project/yifeibot$ touch testing [12:58:01] zhuyifei1999@tools-dev:/data/project/yifeibot$ become yifeibot [12:58:14] local-yifeibot@tools-dev:~$ take testing [12:58:22] testing: Invalid argument [12:58:44] Coren: ^^ [12:59:11] Odd. I'll look into it [13:03:19] anomie: Oh, I see why. There's a bug in the RESTART logic. [13:04:52] fwiw, wmbot read RSS feeds just fine yesterday, it's just today that it switched to all "False" [13:07:23] anomie: Fix't [13:07:34] Pro tip: qstat -u root -j httpd-anomiebot [13:08:42] Coren: auto restart failed again with about 4 of my bots [13:09:27] Betacommand: That's probably my bad; I was so tired when I finally could restart things that some exec nodes were rebooted early with only partial filesystem access. [13:09:40] * Betacommand grumbles [13:09:40] Betacommand: I thought I had caught all, but possibly not. [13:09:57] those bots have yet to survive a reboot [13:09:59] Coren: BTW, any interest in including that rule in ~local-anomiebot/.lighttpd.conf in the default config? If not, I'll put it on the webpage under "recipes for common configuration tasks". [13:11:25] anomie: It'd make sense (to make it default). After I caught up on sleep. The only reason I'm here at all is for some emergency support. :-) [13:36:21] (03CR) 10Yuvipanda: "Ow, okay :D Can we make it to be Draft1, to be consistent with PS1, PS2, etc?" [labs/tools/grrrit] - 10https://gerrit.wikimedia.org/r/88037 (owner: 10Hashar) [14:09:20] YuviPanda: yes [14:09:54] (03CR) 10Hashar: "Yeah go for Draft1 :-] Can you amend it yourself? I am too busy right now sorry. Else will do next week." [labs/tools/grrrit] - 10https://gerrit.wikimedia.org/r/88037 (owner: 10Hashar) [14:11:07] I am off [14:18:06] (03CR) 10Yuvipanda: "Sure! Just didn't want to step on toes :P I'll amend it later tonight" [labs/tools/grrrit] - 10https://gerrit.wikimedia.org/r/88037 (owner: 10Hashar) [14:20:50] YuviPanda: if awake, can you restart/check on the unicorn? I'm getting 500s now. [14:42:48] andrewbogott: ah, okay [14:42:50] moment [14:43:14] thx [14:43:51] andrewbogott: I guess it needs a restart? [14:43:56] Killed by signal 1. [14:45:02] (03PS3) 10Yuvipanda: draft exposed as 'draft-1' instead of 'PD1' [labs/tools/grrrit] - 10https://gerrit.wikimedia.org/r/88037 (owner: 10Hashar) [14:45:43] (03PS4) 10Yuvipanda: draft exposed as 'Draft1' instead of 'PD1' [labs/tools/grrrit] - 10https://gerrit.wikimedia.org/r/88037 (owner: 10Hashar) [14:46:42] (03CR) 10Yuvipanda: [C: 032 V: 032] draft exposed as 'Draft1' instead of 'PD1' [labs/tools/grrrit] - 10https://gerrit.wikimedia.org/r/88037 (owner: 10Hashar) [14:47:49] andrewbogott: rebooting [14:49:23] (03CR) 10Yuvipanda: "Merged and Deployed. Would be awesome if you could test it at some point." [labs/tools/grrrit] - 10https://gerrit.wikimedia.org/r/88037 (owner: 10Hashar) [14:50:10] andrewbogott: still can't ssh in. can you take a look? [14:50:17] yep [14:51:10] andrewbogott: machine seems dead [14:51:10] andrewbogott: https://metrics.wmflabs.org/ doesn't work either [14:51:10] responds to ping tho [14:51:10] * YuviPanda awards Coren|AFK a Barnstar of Staying Up Watching Things Copy [14:51:26] YuviPanda: I can log into proxy-dammit both as root and as Andrew [14:51:48] andrewbogott: ah, i can now too [14:51:55] So just a delay in reboot [14:51:58] probably [14:52:05] oh, no screen [14:52:07] right [14:52:36] andrewbogott: should be running on 5000 now [14:53:16] Yep, looks better. Thanks! [14:53:38] yw [14:54:11] YuviPanda: did you get anywhere with checking code into gerrit? [14:54:19] andrewbogott: nope. needs push right [14:58:11] YuviPanda: Can you push now? I added group lolrrit since that's basically just you anyway [14:58:18] trying [14:59:18] andrewbogott: woo, that worked! :D https://github.com/wikimedia/labs-invisible-unicorn [14:59:51] Cool. [15:00:14] I'm going to be afk off and on today, but I'll think about making a deb out of this... [15:00:58] YuviPanda: Now that things are imported do you want to keep push rights? [15:01:07] ah, no. take it off! :) [15:01:12] 'k [15:01:22] andrewbogott: it might need a newer version of python-redis and python-flask :( [15:01:24] not sure about flask [15:01:28] but probably redis [15:01:40] Yeah, and that special nginx package [15:01:43] Will be a mess [15:01:55] andrewbogott: i can rewrite it to use the system packages [15:02:07] andrewbogott: for flask and redis at least [15:02:13] If that's not a huge amount of work, that'd be really good. [15:02:16] it's not [15:02:23] let me know when you stop testing so I can take a look [15:02:44] Redis might be *slightly* slow on system package, but since this is just the API shouldn't be a problem [15:03:27] I'm done testing for the moment. [15:03:53] andrewbogott: grr, it uses flask SQLalchemy, which has no package [15:03:54] :( [15:05:54] Everything you're doing is similar to things that Openstack services do (using wsgi and sqlalchemy) so if you look at OS source maybe it'll be obvious how to adapt? [15:06:02] Since presumably we already have all the packaging we need for Openstack APIs [15:06:25] - back soon - [15:06:48] andrewbogott: hmm, okay [15:06:51] will do! [15:34:06] YuviPanda: Um… not that I want to necessarily encourage you to rewrite everything, but the OS services use a shared core called 'Oslo' that might magically do all the things you want. [15:34:33] I haven't looked at it in a while, though, it might have grown to be excessively complex. [16:00:38] andrewbogott: right. I think Ryan_Lane linked me to it at some point. [16:41:23] Coren|AFK: err, why are all the projects home directories in / ? [16:49:18] Coren|AFK: and / and /data/project/ are not the same... [16:50:52] Coren|AFK, petan, and I can't write to /data/project/gerrit-patch-uploader/ as it's owned by root... [17:11:44] a bunch of my jobs are in error state [17:11:45] error reason 1: 10/11/2013 06:54:12 [40004:29226]: can't stat() "/data/project/legobot//GA_bot.out" as stdout_path: [17:12:43] i killed all of them except one [17:13:18] my logs are just under 5M! [17:13:19] nice [17:13:42] legoktm: there's something wrong with the project mount, I think [17:13:52] :/ [17:15:46] legoktm: /data/project *and* / contain home directories for projects, but they are not the same [17:15:51] hm, tried activating lighttpd by creating .lighttpd.conf (https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/Help/NewWeb), but I'm not sure how to configure it to work with python.. Anyone know? [17:16:09] danmichaelo: http://flask.pocoo.org/docs/deploying/fastcgi/ [17:16:18] I haven't tried it myself yet, though. [17:17:02] https://panel.djangoeurope.com/support/manual-setup/ under Set up your Django site in lighttpd might also help [17:17:13] I was planning to work on it tonight, but my projects' home folder is broken [17:18:47] danmichaelo: you could probably also just use cgi scripts, though. [17:19:19] yeah, that's what I did with apache [17:19:40] I'm working on the remaining kinks. Something went broky with chown when root on the new NFS server I'm trying to debug (hence inability to change ownership) [17:21:32] Find it a bit weird that the python script itself is specified in "bin-path". Seems like you have to add a "fastcgi.server" block for each script then. Not that I have that many scripts, just seems a bit redundant [17:23:31] danmichaelo: the fastcgi.server maps an http address ('/yourapplication.fcgi') to an application (/var/www/yourapplication/yourapplication.fcgi) [17:23:42] danmichaelo: the application doesn't have to be (and shouldn't be, really) in public_html [17:24:22] Sorry about the (brief) filesystem stall, changing a setting. [17:28:37] danmichaelo & valhallasw : looks like you can either map every single application, or create one socket and map processes to that, e.g: http://home.badc.rl.ac.uk/lawrence/blog/2006/10/26/exploring_web_server_backends_-_installing_fastcgi_and_lighttpd ? [17:28:54] sorry, tell your application to connect to the socket that's available [17:30:04] hmm [17:47:59] The first time I touched lighttpd.RESTART the server restarted and lighttpd.RESTART was deleted, but I also got the lighttpd.STOPPED file that I cannot delete (owned by root, rw-r-r) and the second time I touched lighttpd.RESTART nothing happened (it's eight minutes ago). I wonder if the presence of lighttpd.STOPPED is blocking the restart [17:48:32] also, nothing is written to error.log... [17:50:12] danmichaelo: It is. [17:50:27] danmichaelo: But you should be able to remove the lighttpd.STOPPED file [17:50:47] Coren|AFK: liar! :P (AFK) [17:52:01] Coren: any idea why /data/project/gerrit-patch-uploader is owned by root? [17:52:26] and, more importantly, could you chown it to local-gerrit-patch-uploader? :-) [17:52:34] valhallasw: Yes, because there is a problem with permissions atm prevent root from chowning stuff. [17:53:04] Hey all, quick question regarding the javascript API [17:53:12] what would be the time when mw.user is ready within the mw.loader? [17:53:28] I don't understand why, but yes, I could remove it 8) [17:53:44] Ryan_Lane: thinking of any timeline for migration from public IPs to proxy? [17:53:58] Ryan_Lane: asking to make sure I can be available around that time [17:55:30] Coren: ahhhh. OK [17:57:04] YuviPanda: whenever we have the interface merged and the api packaged [18:03:12] valhallasw: I've chowned it manually from the server. [18:07:15] Coren: cool, thanks. [18:08:29] Coren, now that the hardware upgrade is complete, how about expanding my instance? :-) [18:08:58] Cyberpower678: so you can make more 1T log files? :) [18:09:26] None of my files are no where near that big. [18:09:51] Read the email again. :p [18:10:02] it said 2G by mistake [18:10:03] he meant 2T [18:13:04] Can you verify that? [18:13:42]