[00:07:42] So Coren, it appears java doesn't like the job scheduling system http://fpaste.org/11748/36840365/ [00:07:57] though when the command is not run with the job sched. system, it runs just fine [00:08:08] JasonDC: It does, the key is "Could not reserve enough space for object heap" [00:08:39] JasonDC: By default, jsub only grants a very conservative 256M of VM [00:09:04] JasonDC: Give it 1G to start with (-mem 1g) and check its actual usage to see if you can scale it back down. [00:09:52] is it 1024mb or 1000mb for the 1g command (some systems love to round) [00:10:26] gridengine does worse. "1g" is 1 billion octets. :-) [00:10:37] fun, seems to be working though :) [00:11:07] I don't see it running? [00:11:11] Short run? [00:12:25] nope, it showed up for about a second then died, same error [00:12:52] i even tried telling the VM to run with a max of 800MB of RAM just to make sure, doesn't seem it likes that [00:13:52] same at 512 [00:18:09] JasonDC: It uses actally mapped memory. Try -mem = 2g [00:18:33] JasonDC: It's a guard against overcommit, not actual resident size usage. :-) [00:19:01] Java probably has a large map [00:19:11] -mem 2g [00:19:13] No = [00:19:29] nope still hates the command :( [00:20:33] o_O [00:20:50] Go all out, try 8g and we'll see what the hell java is smoking. :-) [00:21:20] If it's that bad, I'll make a java-only queue that has different metrics. [00:23:38] i think it wants 4g of ram [00:23:51] stays pretty solid there from the status indicator [00:24:21] idk why it's so fussy though, it has run on 1g laptops [00:24:24] oh well :P [00:25:29] JasonDC: That's the difference between the mapped size and the actual resident set. vmem includes the executable, shared libraries, etc. The program _could_ use all of that memory in certain circumstances, even though it almost certainly never will. [00:26:21] lovely, will note that [00:26:42] i guess i'll let it run as is, or would you like me to restart it with the smaller ram usage? [00:28:23] Java should be reasonably consisten it its memory use. Uf you say 4g you'll be pretty close. [00:38:34] awesome, restarted the task with the right mem usage and setup the cron job to run 2 times a month [00:38:41] thanks again Coren :) [00:39:07] JasonDC: My pleasure. [00:39:14] Oh, wait... [00:39:24] If you set up a cron job, make sure to add the -once option [00:39:37] ah, good point [00:40:20] incidentally, did you intend that bot to be continuously running (e.g. restarted if it happens to crash) [00:40:48] no, the terms of the BRFA were certain amounts per month given the sheer number of edits it makes [00:41:37] Allright, then just -once will do. [03:36:54] Ryan_Lane: notpeter tells me you are the openstack guru. :) [03:37:10] yes, but I'm not really here right now :) [03:37:14] Ryan_Lane: apparently UCB is using it (or will be soon) so I'm reading your docs. [03:37:17] Ryan_Lane: *nod* np [03:37:24] I'm here for about 4 more minutes [03:37:40] I may inquire about how to setup a kvm instance at some point; just getting the ssh keys and shell request stuff filled out now. [03:37:43] no rush. [03:37:51] cool [03:37:52] just wanting to say hello! ^_^ [03:38:00] what's UCB? [03:38:04] UC Berkeley [03:38:18] I'm interviewing for a l4 sysadmin position in their IT/EECS department on Tuesday. [03:38:27] figured it would be nice to get some familiarity with Openstack beforehand. [03:38:36] and using a working instance seems to be a good way to do so. ;0 [03:38:37] Ryan_Lane: Hey there. Odd time to pop up! :-) [03:38:38] :) [03:38:38] ah. cool [03:38:51] Coren: heh. I'm just online checking on for my flight tomorrow [03:39:03] if I stay on too long my GF will be unhappy ;) [03:39:13] Ryan_Lane: http://tools.wmflabs.org/?status <-- fun [03:39:23] hehe, understood. [03:39:31] Good night and good flight! [03:39:34] GF > irc [03:39:43] Ryan_Lane: Next week -> DB sprint. [03:39:48] Coren: [03:39:49] yep [03:40:00] I'm also upgrading openstack [03:43:05] neat. [03:43:28] I was looking to see if I can run it on FreeBSD; appears like the answer is presently no; but it runs kvm... so at least I can run it *on* Openstack... [03:43:53] better than nothing. ;) [03:44:03] heh [03:44:15] I don't see a reason it wouldn't work on it [03:44:24] assuming you can install all of the deps [03:44:46] ok. gotta run [03:44:50] apparently it has some linuxisms, like iptables, and use of brctl etc. [03:44:55] ttyl! :) [03:44:59] see ya :) [04:10:14] Coren: how long does it take for http://tools.wmflabs.org/?list to update? [04:11:09] Did you know that you can disable these tips but doing touch ~/.suppresstips <-- typo, but that is wonderful :D [04:12:43] legoktm: I believe it's taken directly from the file system. [04:13:23] hm [04:13:27] well it just updated now :) [04:13:34] legoktm: You can look at /data/project/.system/public_html/content/list.php. [04:13:45] <-- lazy :P [09:14:30] is there a step by step guide to getting started on labs for people used to toolserver? my searching doesn't turn up anything straightforward enough for my aging brain [09:21:55] hippietrail: https://wikitech.wikimedia.org/wiki/User:Magnus_Manske/Migrating_from_toolserver [09:23:38] aha thanks! [09:24:39] !log wikidata-dev wikidata-dev-9 edited crontab (after demo time) [11:29:33] mhhm, what is the cpu stat on http://tools.wmflabs.org/?status ? [11:35:05] addshore: CPU time, i guess. http://en.wikipedia.org/wiki/Cputime [11:35:40] mhhn [12:14:34] addshore: CPU time [12:48:57] Coren, can you create an "X!s-tools" project for me? [12:49:59] addshore, got your memo. [12:50:38] Not with a bang in its name, I can't. :-) /[a-z0-9-]{3,16}/ [12:51:35] Coren, how about "X!s-super-awesome-highly-depended-on-tools" [12:52:03] :-) *bzzzt* Sorry, try again. :-) [12:53:07] Cyberpower678: Incidentally, there is no reason why you can't just create the tool yourself now. :-) [12:53:13] Coren, "Coren-is-amazing" [12:53:19] oh? [12:56:47] Coren, how do you make a tool? [12:57:30] From https://wikitech.wikimedia.org/wiki/Special:NovaProject, look at 'Add service group' in the tools project actions bar (far right) [12:58:09] Coren: could you add me to the "tools" project? [12:58:28] hippietrail: What's your wikitech username? [12:58:45] hippietrail too i believe [12:59:03] "Successfully added hippietrail to tools." [12:59:03] Cool I can even add users [12:59:29] Cyberpower678: Well yeah, that's how you add maintainers. :-) [12:59:36] thanks. trying to see what labs is all about. i've had stuff on toolserver for years [12:59:42] nothing really big [13:00:51] Coren, Failed to create service group. [13:01:21] Cyberpower678: Like I said: /[a-z0-9-]{3,16}/ [13:01:34] You probably used bad characters. :-) [13:01:46] xtools [13:02:22] That's because someone already /has/ a xtools. :-) [13:03:13] Darn you Jack [13:04:01] Coren, add me to local-admin. [13:04:11] Err, no? [13:04:25] :-) [13:04:27] But whyyyyyyyyyyyy? [13:04:30] :p [13:04:36] Too much paperwork to fill. :-) [13:06:54] Hi, I want to join wikimedia Bot project [13:07:01] what should I do? [13:07:17] I'll add you. [13:07:24] thanks [13:07:26] What's your wikitech username. [13:07:33] Reza [13:07:49] also I have toolserver [13:08:03] Then you can't have an account. :p [13:08:04] account and I used screen and crontab with putty [13:08:06] Just kidding [13:08:20] reza: You want bots (experimental, unstable) or tools (stable) [13:08:39] (experimental, unstable) [13:08:58] reza: Ah, mad science time? :-) [13:09:34] how can i connect to server? [13:09:43] before I used winscp [13:10:04] tools-login.wmflabs.org [13:10:06] via SSH [13:10:30] Coren: is it ok to set up a "test" tool? is that the right way to go when you just want to explore the labs setup rather than start an actual new tool? [13:11:02] i'm following the instructions at https://wikitech.wikimedia.org/wiki/User:Magnus_Manske/Migrating_from_toolserver [13:11:27] hippietrail: Oh, yes. The experimental/not experimental distinction is at the infrastructure level. tools has a well tested infrastructure with supported packages, new experimental versions go to bots first for testing. [13:11:32] Coren, cough, I thought any member of a group can add another member to a group. [13:11:50] Just found out what projectadmins are for. [13:11:55] Cyberpower678: *service* groups. Not projects. :-) [13:12:02] :/ [13:12:18] service group is scary new jargon d-; [13:12:31] * Cyberpower678 feels like an idiot in front of Reedy  [13:12:39] Reza: You're on bots now. [13:12:39] *reza [13:12:55] @coren thanks [13:14:05] Coren, where are the bots service groups located. [13:14:28] Cyberpower678: They don't have any, yet. Petan is the one to ask. [13:15:10] So where is cluebot running from ? [13:17:37] hippietrail: service group = MMP [13:17:37] project = cluster [13:17:59] what's an MMP? [13:18:25] hippietrail: TS-speak. "Multi-Maintainer Project" which is the default (and only) option on tools. [13:18:34] Although Multi includes "exactly one" :-) [13:19:02] ah yes i know that. it's a good idea. just a bit odd when you're starting out - who will want to share my messing about? (-: [13:20:14] The point isn't where someone will at any one point in time, but that someone /can/ if it ever becomes necessary. [13:20:24] * Cyberpower678 is hacking labs. [13:20:31] >:D [13:25:47] Hi. I've requested shell access recently https://wikitech.wikimedia.org/wiki/Shell_Request/Ebrahim can I have access to tool labs? I had toolserver account in the past (~ebr) :) [13:25:57] to convert to TS jargon [13:26:34] project admins = roots [13:26:34] what else do we have? :-) [13:28:33] ebraminio: No, you have to ask for that separately, like you just did. You have access to tools now. :-) [13:29:12] ebraminio: Getting started -> https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/Help [13:31:11] is there any shared place for dumps on any policies about dumps at all? [13:31:43] hippietrail: Dumps are in /public/datasets [13:31:58] Coren: Thanks!!! :) [13:35:33] Coren: do they get put there automatically or can we put the ones there that we need? [13:35:55] hippietrail: All public wikis are dumped there automagically. [13:36:13] cool (-: [13:36:47] 857 of 'em at last count. :-D [13:37:27] very nice [13:39:08] don't type "set" at the shell prompt [13:42:04] Cyberpower678: You've got serious incoming spam from one of your tools. [13:42:40] Cyberpower678: You might want to check your mail. [13:43:37] Coren, why? [13:43:46] and no new messages [13:44:12] -rw------- 1 cyberpower678 wikidev 2380834 May 13 13:43 /var/mail/cyberpower678 [13:44:41] Coren, I have no clue what that means. [13:45:04] Just 'less /var/mail/cyberpower678' for now. :-) [13:45:13] Or use 'mailx' [13:45:35] Huh? I haven't ever received an email from any tool yet. [13:46:03] You're receiving PHP errors from your mediawiki install [13:46:18] Where? [13:46:46] It's working just fine. [13:47:36] ah, i spies a reza [13:47:44] or not [13:49:49] Cyberpower678: Most seem to be 'PHP Fatal error: Uncaught exception 'BadEntryError' with message 'Bad Entry Error: BadConfig (A non-existent configuration file was specified.)' in /data/project/cyberbot/Peachy/Init.php:239' [13:50:13] Ah, not your wiki install [13:50:30] Also 'PHP Warning: file_get_contents(/home/cyberpower678/bots/isUpdating): failed to open stream: Permission denied in /data/project/cyberbot/bots/rfx-tally.php on line 46' [13:52:32] Coren, where are you seeing this? [13:52:41] ... in your mail! [13:52:56] I don't have mail!? [13:53:16] You've got over 2 megs of it! [13:53:30] Direct me to it. [13:54:10] It certainly hasn't been emailing me. [13:54:58] Use 'alpine', 'mailx', or even less like I said above. It's mail *in the project*. We'll get forwarding to external email addresses soon once the domain name has been sorted out. [13:55:25] Coren I would prefer keep it local XD or let the users pick [13:55:37] I would never want to receive this kind of mail to real inbox [13:56:15] petan: Depending on what legal decides, it's either going to be an optional .forward or an obligatory .forward. Not my call. [13:56:27] o.O [13:56:34] who cares about legal! :D [13:56:55] * petan slaps legal with huge torrent [13:56:56] There are legal issues with being an email provider they have not yet settled. [13:57:03] oh... [13:57:09] dear... [13:57:12] god... [13:57:16] O.O [13:57:49] How is my bot even running. [13:57:50] ? [13:58:13] Cyberpower678: Ask its maintainers... oh. m :-) [14:10:09] Coren, I know what's causing the BadEntryError. [14:10:19] Missing config file. [14:45:29] petan: are you around? I got blocked on wikitech [14:45:42] yes [14:45:44] hashar: I am. Wikitech blocks are random [14:45:46] blocked how? [14:45:48] The block was made by Petrb. The reason given is Spamming links to external sites. [14:45:49] Start of block: 10:52, 9 May 2013 || Expiry of block: infinite || Intended blockee: Novaadmin [14:45:54] wtf [14:45:59] can you please remove this one? [14:46:03] https://wikitech.wikimedia.org/wiki/Special:BlockList [14:46:06] sure [14:46:13] but why it blocked you?? [14:46:27] and I get a block ID #21 [14:46:43] petan: Don't block on Wikitech right now. There's something funky going on between LDAP, CanonicalUserName, and autoblocks. [14:46:57] Ryan is aware of the issue [14:47:09] I cloned from mediawiki and keep seeing this: http://tools.wmflabs.org/fatestwiki/w/ Can I have access too apache log for resolving this error? [14:47:15] o.O [14:47:22] Error: Block ID Novaadmin not found. It may have been unblocked already. [14:47:52] and block ID #21 needs to be removed as well :-] [14:47:59] hashar what is your username [14:48:02] hashar [14:48:28] I can't remove it :/ [14:48:37] ebraminio: Does your tool own the PHP script? [14:48:37] Coren can you remove it from db? [14:48:46] I don't have so much powar [14:49:07] petan: I had no trouble removing #21 [14:49:11] hashar: All clear? [14:49:13] where is a list of all root users for bots-gs? [14:49:23] weird [14:49:27] Coren: it is mediawiki code, yes it has php [14:49:29] Coren: yup works for me now thx! [14:49:34] Pyfisch in motd [14:49:43] ebraminio: I mean, the actual PHP file. Is it owned by the tool? [14:50:14] Coren: hmm, so I must chown it to tool group? [14:50:16] Ah, I see that it is not. :-) [14:50:46] ebraminio: You can't chown /to/, but you can 'take files_and_dir' from the tool account. [14:51:02] I.e.: [14:51:07] become fatestwiki [14:51:13] take public_htmll [14:51:17] take public_html * [14:51:28] that'll "Just Work™" [14:51:38] petan: motd? in the message of the day of the irc it is not of course, what do you mean? [14:53:06] I mean message of the day on server [14:53:08] bots-gs [14:53:23] when you login to that server, it is printed out I guess [14:54:06] ebraminio: Literally, type 'become fatestwiki' [14:54:15] ebraminio: That'll switch you to the tool account. :-) [14:54:24] Coren: oh, ok [14:54:37] ebraminio: From there, you can use the take command. [14:55:25] ebrahim@tools-login:/data/project/fatestwiki$ become fatestwiki [14:55:25] sudo: sorry, a password is required to run sudo [14:55:33] o_O [14:55:47] Did you log out and back in since you created the group? [14:56:12] Your group membership is set at the beginning of your session. :-) [14:57:24] ebraminio: There you go. :-) [14:57:40] Coren: Thanks!!!! [15:00:04] Coren: is it possible to have a second group for deving? i.e. addshore-dev? [15:00:12] petan: Hey, where is the build directory for apache2.4 you were working on? [15:00:30] Coren what you mean? I installed it all from apt [15:00:37] addshore: Sure, Just create a new service group. [15:00:45] oh yes, i forgot we can create them now ;p [15:00:46] packages are in /data/project/.shared/ [15:00:55] petan: You made the .deb. Where is the build directory? [15:01:01] petan, Damianz, Ryan Lane, addshore: someone changed a file in my home directory on bots-gs which contained a login password for wikimedia. The file must be changed at maximum two days ago (more likely today), but the timestamp is from 2013-04-28. I can't imagine that one of you do such things, but you are the only ones who can have read and changed with your root rights. I don't find this funny! [15:01:26] Coren nah, I just installed the one from debian together with all dependencies, why should I recreate it? [15:02:13] petan: Because I need a couple more modules, so I was going to rebuild it that way. But if you can find the apache2-ipw for 2.4 (which is part of the same sdeb) that'll do. [15:02:33] petan: I've moved the packages you already did in our project-local apt repo. [15:02:41] you should put modules to external packages [15:03:00] petan: I know, the same sdeb also sources libapache2-ipw [15:04:58] I mean, apache2-mpm-itk [15:05:20] But you just grabbed the debian prebuilts? [15:06:59] petan: Nevermind, then. I just downloaded it. :-) [15:07:13] Coren sorry I am in office and quite busy [15:07:22] Coren yes I grabbed the prebuilds [15:07:25] from sid repo [15:07:25] petan: No worries. RL work takes priority. :-) [15:13:20] petan, Damianz, Ryan Lane, addshore: someone changed a file in my home directory on bots-gs which contained a login password for wikimedia. The file must be changed at maximum two days ago (more likely today), but the timestamp is from 2013-04-28. I can't imagine that one of you do such things, but you are the only ones who can have read and changed with your root rights. I don't find this funny [15:13:48] or is it normal that you look for such information [15:13:50] ? [15:14:02] Hi I have problem with login to server with putty [15:14:14] when i want to login with tools-login.wmflabs.org it shows [15:14:29] Pyfisch hi, why do you think that anyone did it? [15:14:52] are you sure it wasn't changed by some of your tools? [15:14:52] server connection error [15:15:11] reza can you try tools-dev.wmflabs.org [15:15:28] now i try it [15:16:09] it shows the same error [15:16:23] petan: I am 100% sure that my tools did not change it. And please explain me how they should have changed it without changing the Modify time? If it was not my tools and not you who else should have changed it? [15:16:24] Pyfisch if the file has timestamp of 2013-04-28 it likely wasn't changed since then. It's possible to manipulate the timestamps, but it's complicated and I quite fail a reason why anyone would do that [15:17:29] if your file is somewhere in your home or /data/project it could be even some gluster failure (filesystem problem) [15:17:31] Pyfisch: what changed in the file? [15:18:19] addshore: one letter was removed so that my bot can't login anymore and it was disrupted exactly at 13:30 today [15:18:34] petan: my file is in /home/pyfisch [15:19:39] Pyfisch well, that /could/ be some filesystem issue... otherwise I can't think of many other reasons for this :/ [15:21:17] After mediawiki installation I am redirected to http://tools-webserver-01/.... what can I do for it? I must install install mediawiki through a proxy? [15:21:40] I had not such problem on toolserver :( [15:21:45] ebraminio: You need to make certain, when you configure your mediawiki, that the wgServerName is "tools.wmflabs.org" [15:22:09] ebraminio: You're not allowed to install MediaWiki on the toolserver. :-) [15:22:27] ebraminio: Lemme go look at your LocalSettings.php [15:23:04] Coren: I can not even download the localsetting.php :( [15:23:23] ebraminio: It's not there, the config didn't finish. Lemme check. [15:25:00] ebraminio: Did you do the configuration of your wiki yet? [15:25:11] ebraminio: http://tools.wmflabs.org/fatestwiki/w/mw-config/index.php [15:27:20] Coren: With opening of that link, my browser is just redirecting me to invalid place [15:27:34] ebraminio: o_O? It works from here. [15:27:52] Oh, wait, I know what happened. [15:31:42] Cyberpower678: you appear to be running datefix.php on -login :/ [15:32:05] ?? [15:32:14] pid 14685 [15:32:21] ?? [15:32:34] ebraminio: When you configured, you didn't set the server name properly. Lemme see if I can fix that in partial config. [15:33:03] datefix is running from Cyberbot [15:33:12] yes, but it shouldnt be running on -login [15:33:19] it should be submitted to the grid :) [15:33:24] ?? [15:33:25] Cyberpower678: You need to submit it to the grid. [15:33:33] How? [15:34:03] I'm bad with linux remember? [15:34:27] Coren: Oh, I didn't see such thing on the installation [15:34:44] ebraminio: Give me a moment while I try to remember how to fix a partial install. :-) [15:35:04] Coren: Sure. Thanks a lot! [15:35:51] how do I look at a crontab? [15:36:01] crontab -e :) [15:36:03] Cyberpower678: I'll help you in a minute. [15:36:22] Or addshore might be able to guide you a bit in the meantime. [15:37:45] Cyberpower678: rather than running your script directly from cron , your line should look something my my example below [15:37:46] 0 9 * * * jsub -mem 512m -once -N en-usertalksubst php /data/project/addbot/bot/en-usertalksubst.php [15:38:45] and where does that go? [15:39:04] addshore what is irc feed for wikidata [15:39:12] wikidata.wikipedia [15:39:15] k [15:39:53] addshore, Cyberpower678> and where does that go? [15:41:00] On to the execution nodes of the grid :-) [15:41:06] -.- [15:41:17] Laptops just died, brb :/ [15:41:35] That was very helpful. [15:41:47] It doesn't matter where they go as long as it runs right? :p [15:42:00] Is that a question? [15:42:32] Well, since it's running right then it doesn't matter. [15:42:37] Ni think they are tools-exec-01 and 02 and 3 and 4 [15:43:14] You'll have just as much luck talking to me in Chinese. [15:43:34] Oh wait, where does that line go? In your from, or remove the from ttimings from the ffront and just run it as a command on -login [15:44:06] *cron, apparently this thing has auto correct... [15:45:02] In your from, or remove the from ttimings from the ffront and just run it as a command on -login [15:45:04] What? [15:46:34] addshore, you are being very confusing right now. [15:46:37] right, my laptop is back. How are you running your tasks currently? [15:47:00] From a crontab [15:47:17] Like on toolserver [15:47:19] okay, so what does your crontab line say? [15:47:30] what line? [15:47:35] in your crontab [15:48:31] standard stuff. All of the bot scripts. [15:48:39] hashar, is PrivateSettings.php on labs versioned somewhere? [15:48:43] so, something like the below [15:48:57] MaxSem: I don't think it is. it is in .gitignore [15:48:59] 0 0 * * * php /data/project/cyberbot/script.php [15:49:04] ok, thanks [15:49:24] As your crontab is on -login, this means that your script is execusting on the login server [15:49:40] You need to change that line to something like the below [15:50:05] 0 0 * * * jsub -mem 512m -once -N taskname php /data/project/cyberbot/script.php [15:50:06] But I assigned it as cyberbot [15:50:13] !log deployment-prep Created a dummy $wmgRedisPassword = 'not set'; # MaxSem 13.05.1 [15:54:40] addshore, I was just doing what X! was doing. [15:54:47] xcombelle: [15:55:08] Before he retired. [15:55:08] sorry :/ [15:55:15] ebraminio: It's a very good place for it. Unlike the TS, we don't have a rule against running mediawiki. :-) [15:55:46] Coren: great! thanks! [15:55:52] Cyberpower678: X! was using a method that even on the TS was obsolete. [15:55:55] http://pastebin.com/1PZZ18cr [15:56:08] addshore, ^ [15:56:09] give me a few secs [15:56:12] !ping [15:56:12] pong [15:56:13] freaking wm-bot [15:56:13] petan, !pong [15:56:13] don't do that again! :D [15:56:13] you fucking scared me [15:56:33] petan, lol how? [15:56:33] !pong [15:56:33] :< [15:56:34] don't play with me [15:56:34] I thought that was a bot [15:56:39] and that it is fucked more than I thought [15:56:40] @suppresson [15:56:44] How did that go again? [15:56:45] Where's the doc? [15:56:47] whoami [15:56:52] You are trusted identified by name .*@wikipedia/.* [15:56:56] :/ [15:56:57] @trusts [15:56:57] @trusted [15:56:57] I trust: petan!.*@wikimedia/Petrb (2admin), .*@wikimedia/.* (2trusted), .*@mediawiki/.* (2trusted), .*@wikimedia/Ryan-lane (2admin), .*@wikipedia/.* (2trusted), .*@nightshade.toolserver.org (2trusted), .*@wikimedia/Krinkle (2admin), .*@[Ww]ikimedia/.* (2trusted), [15:57:08] @seen Ryan_Lane [15:57:08] petan: Last time I saw Ryan_Lane they were talking in the channel, but they are not in the channel now and I don't know why, in #wikimedia-labs at 1/3/2013 5:33:36 AM (130.10:23:31.8064980 ago) [15:57:14] 130 days ago :D [15:57:19] here we go [15:57:32] petan, I thought I was an admin? [15:57:45] Cyberpower678 are you subscribed to wikitech-l? [15:58:01] I don't think so. [15:58:08] I saw you promote me though. [15:59:20] http://lists.wikimedia.org/pipermail/wikitech-l/2013-May/069368.html [15:59:21] read it [15:59:40] the poor thing was removed from filesystem by accident [16:00:41] so Cyberpower678 your first 4 cron lines would be this http://pastebin.com/QUCa2YGb [16:01:10] I need to be an admin on #wikipedia-en-accounts-unreg [16:01:22] It trusts no one. [16:01:46] petan, ^ [16:01:55] ok [16:02:22] @trusted [16:02:22] I trust: petan!.*@wikimedia/Petrb (2admin), .*@wikimedia/.* (2trusted), .*@mediawiki/.* (2trusted), .*@wikimedia/Ryan-lane (2admin), .*@wikipedia/.* (2trusted), .*@nightshade.toolserver.org (2trusted), .*@wikimedia/Krinkle (2admin), .*@[Ww]ikimedia/.* (2trusted), [16:02:33] and I'm supposedly an admin of it here. [16:02:52] addshore: You might want to add -cwd to those jsubs if they expect to be run from the directory [16:03:04] Coren: yes! [16:03:09] jsub/qsub chdir to home by default [16:03:40] addshore, you deleted my output directory. [16:03:45] Cyberpower678: here is an ammended version > http://pastebin.com/JJx5Cda0 [16:04:07] Cyberpower678: they will output to /data/project/cyberbot/taskname.err and taskname.out [16:04:16] the name is the parameter that follow -N [16:04:24] sec [16:04:40] addshore, I want my out files to go in my CyberbotI and CyberbotII folders [16:04:47] and my err files [16:05:18] Successfuly added .*@wikipedia/Cyberpower678 [16:05:18] @trustadd .*@wikipedia/Cyberpower678 admin [16:05:24] Cyberpower678: take a look here >> https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/Help#Submitting_Jobs [16:06:05] expand the section that says "jsub parameters", and take a look at -o and -e [16:07:43] You deleted the rest of my crontab. [16:08:32] addshore, ^ [16:09:26] petan, where is the bot doc again? [16:09:39] !log bots petrb: deleted bots-1 instance [16:09:53] petan, pyfisch' password issue might just be his own misconfiguration - while a bot is logged in, it wil generally use cookies instead of logging in every time [16:12:19] addshore, ooh. It can store new logs everytime? [16:12:36] naaah, generally I dont need to look at the logs ;p [16:12:47] its a perfect script *cough* [16:13:35] Cyberpower678: you have to convert the rest ;p [16:14:16] And to think I almost got someone to do the dirty work for me. :p [16:14:39] I'm going to get something to eat first. [16:16:44] hehe Cyberpower678 , I would have done it all until I realised how big it was! [16:21:25] addshore, I've got a lot of scripts. [16:21:29] addshore: Regular expressions exist :-). [16:22:41] scfc_de: i love regex ;p [16:27:48] /(I|we) lo{1,4}v{1,2}e(s|z)? reg(ex|ular( |-)expressions)\.?/i [16:29:02] Matches I love regex [16:29:13] We love regex [16:29:26] I loves regex [16:29:33] We loves regex [16:34:12] I lovez regex [16:34:13] we lovez regex [16:34:14] meh [16:34:14] too many to list. [16:34:14] Back to getting food. [16:34:15] hmm Coren does a service groups webspace take a while to get into the apache config? [16:34:17] It make take a minute or two. [16:34:17] What service group? [16:34:18] addshore-dev :/ [16:34:18] getting 403 [16:34:22] oh, just started working ;p http://tools.wmflabs.org/addshore-dev/test [16:39:06] Seems mosh-server fails to start on tool-login.wmflabs.org "mosh-server: invalid option -- 'l'" [17:55:43] fucking netsplits [17:59:13] addshore, is a crontab merely a timer to feed the terminal a command line that you could feed yourself exactly as is? [17:59:32] You are admin identified by name .*@wikipedia/Cyberpower678 [17:59:37] @trusted [17:59:38] I trust: petan!.*@wikimedia/Petrb (2admin), .*@wikimedia/.* (2trusted), .*@mediawiki/.* (2trusted), .*@wikimedia/Ryan-lane (2admin), .*@wikipedia/.* (2trusted), .*@nightshade.toolserver.org (2trusted), .*@wikimedia/Krinkle (2admin), .*@[Ww]ikimedia/.* (2trusted), .*@wikipedia/Cyberpower678 (2admin), [17:59:45] Cyberpower678: no [18:00:00] no what? [18:00:02] it is a separate daemon that launch the commands in clean environment [18:00:20] Ah. [18:01:00] Are the commands identical to what could be used on the command line? [18:01:53] Cyberpower678: In most cases, yes. [18:02:23] I just came to that realization as I was rewriting the crontab. [18:02:34] Coren: Has the issue with cron, jsub and PATH been addressed? (Default PATH not including /usr/local/bin.) [18:02:56] scfc_de: Not that I can recall. What is the bz number? [18:03:19] No BZ yet. [18:03:19] blergh [18:03:23] There's your problem. :-) [18:03:30] andrewbogott, scfc_de: thanks thanks! :) [18:03:43] With the schedule I have until the Hackaton, if it's not in BZ it doesn't exist. :-P [18:03:59] Coren: whatcha think? ;p [18:04:01] https://github.com/addshore/addwiki/blob/84feca2c149fcee0d0f52e4c5450fa3fecf673ff/bot/wikidata/deploy.php [18:04:41] Cyberpower678: So, in a crontab you either have to use "/usr/local/bin/jsub" instead of "jsub" on the command line, or add a line "PATH=/usr/local/bin:/usr/bin:/bin" as the first line to your crontab. [18:04:43] buh: Meeting now. Will look at it later. [18:05:00] naaaw :P [18:06:15] buh: What exactly do you want to achieve? [18:07:32] buh, you dodn't mention that to me. [18:07:43] scfc_de: long story [18:07:50] Cyberpower678: mention what? [18:08:06] scfc_de> Cyberpower678: So, in a crontab you either have to use "/usr/local/bin/jsub" instead of "jsub" on the command line, or add a line "PATH=/usr/local/bin:/usr/bin:/bin" as the first line to your crontab. [18:08:20] O_o [18:08:25] you do...? [18:09:37] addshore: Yes, unfortunately. Coren might change the default PATH in the future, but at the moment you have to specify the full path (or re-define PATH). [18:09:51] I guess I should start using that then ;p [18:10:48] addshore: Re deploy.php, why not just loop over all languages, start (-once) a job for it, and let the grid sort it out? I mean, that's the whole advantage of the grid, that you don't have to do this scheduling on your own and with limited information about system load. [18:11:16] some of the langs have 2 articles, some have 900,000 [18:11:36] I want to keep checking each languages remaining articles (the plan is daily) [18:11:47] Then one job will finish early, and the other later :-). [18:12:03] but then I have to keep looking through 250 languages to see what isnt running ;p [18:12:16] also I would have to spawn 250 grid jobs [18:12:31] It's a machine, it doesn't mind :-). [18:12:36] this keeps the number of jobs down and slowly cycles through everything ;p [18:12:49] last time I tried there were not 250 job spaces open ;p [18:13:22] Now there are: "0/0/1000" :-). [18:13:40] meh [18:13:51] I think I'm gonna like it this way ;p [18:14:36] Well, it's your bot :-). I just like to keep things as simple as possible, so my brain can still understand what the script is supposed to be doing, two weeks from now :-). [18:14:37] addshore, isn't it a bit early to use buh as your nick? [18:14:48] O_O? [18:15:15] scfc_de: I did have it simply spawning 250 processes a day on bots, but its just so ugly xD [18:15:18] That should be reserved for christmas so everyone can see how much you hate it. [18:15:28] addshore, ^ [18:16:40] addshore: I'm a fan of minimalism :-). But the more important question: Does deploy.php work?! :-) [18:16:58] scfc_de: we will find out in a second ;p [18:17:07] scfc_de, that attitude will gay you nowhere. [18:18:06] Cyberpower678: The minimalism or the "does it work" bit? :-) [18:18:23] Yep. [18:19:00] You want to make sure it's more than "it works". You want to make sure it stays working in the long run. [18:19:03] :p [18:19:15] That should be reserved for christmas so everyone can see how much you hate it. [18:19:15] addshore, ^ [18:19:59] buh buh buh [18:20:12] Cyberpower678: That's why I try to keep the number of moving parts minimal, so there's less chances for things to fuck up :-). [18:20:24] addshore, btw, it's buh-humbug [18:20:37] Freenode's collapsing again. [18:22:08] addshore: Is is actually running? I don't see it neither as an SGE job nor on tools-dev. [18:25:55] hehe, got a bug or 2 ;p [18:26:34] it couldnt find job and not even if I specify the exact location [18:28:01] addshore: Sure? [18:31:00] it works on login.. [18:31:58] Is the (current) deploy.php accessible somewhere? [18:33:49] If you add "> ~/deploy.log 2&>1" to the end of your crontab, do you errors in that file? [18:34:26] Make that "2>&1". [18:42:05] Coren, addshore: New crontab successfully installed. [18:42:41] Yeay [18:43:13] Coren, everything's broken now. [18:43:21] But completely stalled. [18:43:24] *Bot [18:43:47] Output files aren't going where they are supposed to. [18:45:10] Coren, HELP [18:45:26] Check your -o and -e options [18:45:34] (In a meeting right now) [18:45:48] Restoring old CronTab [18:46:18] Done [18:46:39] scfc_de: I was watching deploy run, ;p on dev its missing commands, on login it ran [18:46:39] addshore, explain why the new crontab broke everything. :p [18:47:03] ill try and have a look in a sec [18:47:08] scfc_de: may be able to help you ;p [18:47:49] hehe scfc_de you can see my little spurt of activity on http://tools.wmflabs.org/addshore/addbot/status/ [18:48:55] and scfc_de it is https://github.com/addshore/addwiki/blob/29799b3306b62ecdb1bd50dc1b0aa877f6ce290c/bot/wikidata/deploy.php [18:49:43] Pyfisch: your password issue might be a misconfiguration on your end - a bot will typically not need a password while logged in, so the misconfiguration will only be noticed in maybe 30 days or so [18:49:59] so you might have changed the file by accident at the end of april (hence the modification date) [18:51:29] valhallasw: my bot needs password if it puts a page to wikidata (no item changes, a page). It needs the password pretty often. [18:52:56] Pyfisch: if it's pywikipedia, it should just cache the cookies [18:53:15] it is pwb [18:54:28] mhhhm scfc_de can not run screen on -login from a service group account :< [18:54:40] this just keeps getting better and better [18:55:00] addshore: what's the error you get? [18:55:06] addshore, have you had a look yet? [18:55:22] cannot open your terminal bla bla /pts/1 - please check [18:55:37] Cyberpower678: not yet :/ [18:55:44] paste one of the cron lines in here [18:55:47] Not that I really care where it runs as long it runs correctly. [18:56:17] Cyberpower678: the problem is if you run on -login you are slowing down the only point of entry to the cluster [18:56:20] affecting everyone [18:56:49] This one runs the reporter [18:56:51] */15 * * * * cd /data/project/cyberbot/bots && jsub -mem 512m -once -N rfx-report php rfx-report.php -o $HOME/CyberbotI/rfx-report.out -e $HOME/CyberbotI/rfx-report.err [18:57:01] addshore: right, that is a general problem with running screen as a different user [18:57:20] Cyberpower678: did you not see scfc_de 's fix above? [18:57:54] This is affixed to the top of the crontab [18:57:54] PATH=/usr/local/bin:/usr/bin:/bin [18:58:02] addshore: try first running script /dev/null [18:58:06] ahh, I would just change all jsub 's to the location [18:58:24] valhallasw: Script started, file is /dev/null [18:58:29] addshore: now try screen again [18:58:33] I will not restore the crontab until the output is directed to the correct folder. [18:58:41] val, perfect [18:58:52] addshore: you'll need to run that also before re-attaching [18:58:58] and unfortunately it adds another logout layer [18:59:02] Cyberpower678: you rick just having all of your processes killed the way you are running them.. [18:59:11] but it's the best method I could find [18:59:20] valhallasw: interesting :P I just hope I remember :) [18:59:36] addshore, Can you tell my why my output is going where it's supposed to? [18:59:44] *isn't [19:00:15] Cyberpower678: not really, i dont use -o or -e, I just let it all collect in my project folder [19:00:42] I don't want my project folder being cluttered like that. [19:01:16] I never look in my rpoject folders ;p [19:01:25] heck, i barely have to log into the users now xD [19:01:35] petan, can you tell me why the output is being directed to the location I want it to? [19:02:00] Cyberpower678: you're passing your arguments to php, not to jsub [19:02:07] move them to the left of 'php rfx-report.php' [19:03:00] I'll fix that later. I have a doctor's appointment now. [19:03:16] scfc_de: my deploy works :) [19:09:42] Cyberpower678: valhallasw is right, BTW. [19:11:01] addshore: Nice. BTW, the graphs are really cool. When Ryan is back (holiday or something?), we should discuss with him how he feels about using Ganglia for user-level stuff as well. [19:23:41] scfc_de: that would be awesome. [19:23:53] I'd move gerrit reviewer bot to labs pretty quickly then ;-) [19:29:50] Anyone know if there is a command to check to see previous job's that were run on tools? (some sort of job run history) [19:32:10] mhhhm JasonDC|BNC not that I know of [19:32:52] * Coren beats apache 2.4 into compliance. Run, dammit! [19:33:06] SGE can log jobs, but afais only to a central file [19:33:33] alright, reason i ask is because i have a cron job that should have already run a few hours ago and I recieved no output to my log file, so i'm debating if it didn't run (because my cron is poorly scripted) or something else [19:33:59] JasonDC|BNC: check with qstat [19:34:00] JasonDC|BNC: You can get recently finished jobs with 'qstat -s z' [19:35:06] wow, wiki is rather lagged atm [19:35:28] I just wondered why all my tasks were not editing at its because maxlag is shooting up [19:36:38] oh no, or my maxlag detection has somehow broken in the move... [19:41:44] clever me :D [19:46:22] Coren: got my wikidata bot running around in a lovely 10job loop across the whole db :) [19:47:41] addshore: Success! [19:53:17] lovely :D [19:54:25] but Coren im having to run the deploy on -login as -dev doesnt have the command 'job' currently ;p [19:55:13] Will be fixed soon. [19:55:23] [= [20:15:35] Coren: where can I see the source behind http://tools.wmflabs.org/?status ? [20:15:54] addshore: /data/project/.system/public_html [20:19:33] would be awesome to be able to pass a parametr to it and see only the jobs of a specific tool, e.g. if you click on that tools name from the origional list [20:21:29] since a shutdown 8 days ago because of "restart required" messages, my "instances" are not correctly working, and I see missing libs in the error log files [20:21:40] is someone here having the same problems ? [20:21:51] andrewbogott *ping* [20:22:01] Tools project seems unafected. [20:22:16] Coren: I already contacted Ryan [20:22:38] but my OpenID testwikis are still not wroking correctly [20:23:18] "/usr/bin/php: error while loading shared libraries: libheimntlm.so.0: failed to map segment from shared object: Cannot allocate memory" or [20:23:26] "fatal: Out of memory? mmap failed: Cannot allocate memory" [20:23:41] and the other instance shows [20:23:54] andrewbogott: [20:24:04] do you see my messages above ? [20:24:21] Ryan told me, you updated puppet.... [20:24:27] Wikinaut: You're seeing an openid bug that was revealed (but not caused) by some changes I made in puppet. [20:24:47] I diagnosed the problem but didn't revert my change since it looked like the openid fix was en route. [20:24:50] Lemme see if I can find the bug number [20:25:03] you mean the protocol // [20:25:05] issue ? [20:25:15] It's this one: https://bugzilla.wikimedia.org/show_bug.cgi?id=44293 [20:25:19] no, see the error log messages [20:25:52] they indicate more and other problems ! [20:25:55] Those 'failed to map' messages are unrelated to the problem you are seeing; I don't know why they are happening but they are harmless as best I can tell. [20:26:11] andrewbogott: Speaking of, there is an odd issue that you might have an idea if it's on your side or not: sometimes (about 1:5) when creating a service group the home gets set wrong. No obvious pattern, deleting the service group and recreating it sets it right again on the new one. [20:26:22] "usr/bin/php: error while loading shared libraries: libheimntlm.so.0: failed to map segment from shared object: Cannot allocate memory" [20:26:36] this was NOT happening before your chnage [20:26:45] Wikinaut: Yep, I see those too… I don't know what's causing them but they're not related to the problem. [20:26:46] until ~ 06. May everything worked perfectly [20:26:58] *everything* [20:27:02] Coren: What does the home get set to? [20:27:29] the other instance runs the same, but has other error messages, one moment: [20:28:22] "/usr/bin/php: error while loading shared libraries: libasn1.so.8: failed to map segment from shared object: Cannot allocate memory" [20:28:29] this is on the second instance [20:28:34] a different message ! [20:28:38] andrewbogott: ^ [20:29:17] andrewbogott: do you know the two instances, and can you perhaps look what is going wrong ? [20:29:34] this would be very kind and helpful [20:29:35] Wikinaut: As I said, I see those 'Cannot allocate' errors as well, but I have also determined that they are not the cause of any noticeable symptoms. I don't know why they are appearing or what they mean. [20:29:38] andrewbogott: /data/project/local-group rather than just /data/project/group (i.e.: it keeps the prefix even though it's not in the pattern) [20:30:08] andrewbogott: the error messages were _not_ present before your change/s last week [20:30:41] Wikinaut: I believe you. Nevertheless, in cases when I have fixed the openid failure, those messages remained. [20:30:41] the OpenID issue: I don't know. I never had this problem with my own wikis running on https:// servers [20:30:57] How have you fixed this, I need to know this [20:30:59] please [20:31:05] The problem isn't with https, it's with relative paths. [20:31:25] Wikinaut, I take it you're not able to contact any of the openid devs? I thought you were one yourself. [20:31:26] can you fix it on my servers ? [20:31:35] I am the OpenID dev [20:31:52] let me know how you fixed it [20:32:38] Well, the bug is with relative server paths. If you change $wgServer to an absolute rather than a relative path than the symptom goes away. [20:32:54] but i did not touch that ! [20:33:11] Core, is it possible that in those cases the user actually requested a service group with of 'local-'? And then my code filtered out the 'local' part in some places but not in others? [20:33:42] Wikinaut, this conversation is frustrating to me because I feel like you are panicking and not reading what I'm typing. [20:33:56] no I am not panicking [20:34:00] andrewbogott: In this case, could we add a check that service groups with that prefix can't be requested? [20:34:07] you changed something in "my" systems [20:34:09] "andrewbogott: Wikinaut: You're seeing an openid bug that was revealed (but not caused) by some changes I made in puppet. [20:34:09] [3:24pm] andrewbogott: I diagnosed the problem but didn't revert my change since it looked like the openid fix was en route. [20:34:10] [3:24pm]" [20:34:36] can someone approve https://gerrit.wikimedia.org/r/63569 please? it's pretty uncontroversial:) [20:34:50] scfc_de: If that is in fact what's happening then that's a good idea. That's why I asked 'is it possible...' [20:34:51] ## The protocol and server name to use in fully-qualified URLs [20:34:52] if ( $_SERVER['SERVER_NAME'] == "openid-wiki.pmtpa.wmflabs" ) { [20:34:54] $wgServer = "http://openid-wiki.pmtpa.wmflabs"; // Ryan Lane 20130128 [20:34:55] } else { [20:34:57] $wgServer = "http://openid-wiki.instance-proxy.wmflabs.org"; // Wikinaut 20130128 [20:34:59] } [20:35:42] andrewbogott: why should this be causing trouble now? [20:35:58] Wikinaut: Is this LocalSettings.php or orig/LocalSettings.php? (Assuming you're using mediawiki::singlenode...) [20:36:14] orig/LS [20:36:25] Look at LocalSettings.php. [20:36:31] (moment) [20:37:02] Wikinaut: So, at the moment we have two options, a band-aid fix and a proper fix. I can apply the bandaid, but I would encourage you to pursue the proper fix instead. [20:37:02] Is there a specific reason why you prefer not to do that? [20:37:02] Wikinaut, because I changed wgServer to use a relative path in puppet. [20:37:05] ok [20:37:06] MaxSem: pushed. [20:37:19] Sorry if that wasn't clear before, that's what I meant when I said I 'revealed' the bug. [20:37:22] LS is overwriting orig/LS because it comes later [20:37:23] It amazes me how the load for -login is higher than any other instance on tools >.< [20:37:38] addshore: Because I'm building PHP. :-) [20:37:43] The relative path makes other things work better here and there. We can switch to an absolute path on your instances if necessary. [20:37:54] <^demon> Coren: Building on the login instance? ;-) [20:37:54] andrewbogott: I do understand now [20:38:08] I think I can comment that [20:38:43] ^demon: Yeah, I know. I didn't realize where I was until I installed all the effing build deps. :-) [20:38:53] The change will get overwritten by puppet, unfortunately. I'll have to put in some special-case handling to disable the puppet setting in certain situations... [20:40:07] ^demon: I should be done soon. [20:40:14] <^demon> Oh, not bothering me. [20:40:20] <^demon> I'm just heckling :p [20:40:37] andrewbogott: OpenID works again. Google OpenID has changed (must look, why this happened) [20:41:15] I do know, that Google OpenID is depending on the requesting consumer [20:41:57] but right now, it broke the present OpenID in my account... in a real environment a disaster [20:43:43] Hey, can someone add me (User:Sitic on wikitech) to the 'bots' project? And yes I know the difference between 'bots' and 'tools' project, I'll ask here for 'tools' permissions when it's ready [20:48:47] sitic: {{done}} [20:48:56] Actually, I lied. [20:49:18] Coren: thanks :-) [20:49:26] oh. [20:49:35] sitic: Wait a sec. It didn't work for some reason. Tracking down why. [20:49:40] Coren: what do you think of this arrow flow? http://grab.by/mvcc [20:49:43] sure [20:50:56] sitic: It failed because you were already there. :-) [20:51:33] oh ok.... I didn't asked for it yet... [20:51:53] Someone beat you to it apparently; or I clicked twice without noticing. [20:52:06] You're there either way. :-) [20:52:17] thanks :-) [20:55:05] Coren: Core, is it possible that in those cases the user actually requested a service group with of 'local-'? And then my code filtered out the 'local' part in some places but not in others? [20:56:54] andrewbogott: Not likely. Everywhere /else/ was using the right name -- just the home. [20:57:14] addshore: Actually, tools-dev is accesible from the outside, but otherwise it's good. [20:57:24] it is O_o [20:57:58] it doesnt have an ip listed on https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools ;p [20:59:07] addshore: Odd. That's out of date. [21:05:38] You are unknown to me :) [21:05:44] addshore: can you fix that [21:05:59] addshore: @trustadd petan.* trusted :)) [21:06:04] without :)) [21:06:06] Successfuly added petan.* [21:06:06] @trustadd petan.* trusted [21:06:19] !coren is The toolmeister: http://www.mediawiki.org/wiki/User:MPelletier_(WMF) [21:06:19] Key was added [21:06:53] btw here is old db http://bots.wmflabs.org/~wm-bot/db/backup/%23wikimedia-labs.htm if you find some useful key, plis add it back [21:07:34] !tooldocs is http://www.mediawiki.org/wiki/Wikimedia_Labs/Tool_Labs/Help [21:07:35] Key was added [21:07:45] !petan is OMG Petan deleted me and now he is going to have to spend 2 days trying to put me back together..... [21:07:45] This key already exist - remove it, if you want to change it [21:07:49] !petan [21:07:50] Petr Bena - http://enwp.org/User:Petrb [21:07:53] :/ [21:08:02] !tunnel is ssh -f user@bastion.wmflabs.org -L :server: -N Example for sftp "ssh chewbacca@bastion.wmflabs.org -L 6000:bots-1:22 -N" will open bots-1:22 as localhost:6000 [21:08:02] Key was added [21:08:08] Type @commands for list of commands. This bot is running http://meta.wikimedia.org/wiki/WM-Bot version wikimedia bot v. 1.10.8.0 source code licensed under GPL and located at https://github.com/benapetr/wikimedia-bot [21:08:18] !tooldocs is https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/Help [21:08:18] This key already exist - remove it, if you want to change it [21:08:27] Coren: I just added it :P [21:08:34] You added the old one. :-) [21:08:39] ah [21:08:49] !tooldocs del [21:08:49] Successfully removed tooldocs [21:08:57] There's a redirect, but better point at the right one. :-) [21:08:59] !tooldocs is https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/Help [21:08:59] Key was added [21:09:16] * addshore was just looking up how to delete a key to do that ;p [21:09:46] petan|ic: Does that mean that I can now use "/nick petannotiam" and am trusted by wm-bot?! :-) [21:09:47] !wmflabs is this is authentic foto of labs http://i215.photobucket.com/albums/cc152/estaquente/funny_pictures/e744f91c29ec99f0e662c9177946c627.jpg [21:09:47] Key was added [21:10:05] !petan... is OMG Petan deleted me and now he is going to have to spend 2 days trying to put me back together..... [21:10:05] Key was added [21:10:25] hehe [21:10:47] scfc_de: yes :> [21:10:48] !shellrequests is http://bit.ly/10eZZoa [21:10:48] Key was added [21:11:10] nobody would ever pretend to be me :0 who would dare! [21:11:13] !wm-bot is wm-bot [21:11:13] This key already exist - remove it, if you want to change it [21:11:17] fail [21:11:23] !addshore [21:11:32] !addshore is fail [21:11:33] Key was added [21:11:37] !addshore [21:11:37] fail [21:11:50] You are trusted identified by name .*@wikimedia/.* [21:11:55] oh.. [21:11:56] lol [21:12:17] @trustdel petan.* [21:12:18] You can't delete yourself from db [21:12:21] :o [21:12:38] dat bot is smarter than me [21:12:41] :O [21:12:43] !sh [21:12:43] http://bit.ly/10eZZoa [21:12:43] @trustdel petan.* [21:12:44] User was deleted from access list [21:12:47] ;p [21:12:59] * petan|ic feels deleted [21:12:59] andrewbogott: Is LocalSettings.php overwritten every xx minutes ? [21:13:11] Yep, that file is managed by puppet. [21:13:14] arrrgh [21:13:26] How can this be changed ? [21:13:38] "andrewbogott: The change will get overwritten by puppet, unfortunately. I'll have to put in some special-case handling to disable the puppet setting in certain situations..." [21:13:58] OMG [21:14:36] are you Tyler Romeo, or does he have some obscure IRC name? I never see anyone around by that name. [21:14:37] petan|ic: would this do? petan!uid11516@.*irccloud.com.* [21:14:46] not sure how wmbot has its hosts [21:14:47] andrewbogott: tyler romeo doesn't do irc [21:14:52] andrewbogott: remember that lines of code in orig/LS: [21:14:52] maybe I could just identify :) [21:14:56] ## The protocol and server name to use in fully-qualified URLs [21:14:58] if ( $_SERVER['SERVER_NAME'] == "openid-wiki.pmtpa.wmflabs" ) { [21:15:00] $wgServer = "http://openid-wiki.pmtpa.wmflabs"; // Ryan Lane 20130128 [21:15:02] } else { [21:15:03] petan|ic: thats the better idea ;p [21:15:04] $wgServer = "http://openid-wiki.instance-proxy.wmflabs.org"; // Wikinaut 20130128 [21:15:05] } [21:15:35] andrewbogott: take a look at mediawiki/vagrant -- specifically the readme file in the 'settings/' subfolder, and the last few lines of 'LocalSettings.php' [21:16:07] Wikinaut: Wouldn't it be easier to just apply Tyler's fix? [21:18:43] scfc_de: I will check this. [21:19:01] I checked it some time ago, and his fix broke it fully [21:19:08] (at that time) [21:19:18] but I will check now [21:19:48] At the other hand, Ryan and I we changing the server, as you can see, depending on how it is accessed [21:20:05] Coren, where is the code that actually creates those homedirs? I can't remember if I wrote it or not :) [21:20:36] andrewbogott: is it necessary to have this sequence order : [21:20:40] # Include the complete settings file generated by mw install [21:20:41] require_once( 'orig/LocalSettings.php' ); [21:20:43] $wgServer = "//" . $_SERVER["SERVER_NAME"]; [21:20:54] or can the $wgServer set before calling orig/LS [21:20:55] ? [21:21:09] this would at least ad-hoc fix my problem ! [21:21:33] andrewbogott: I suggest this sequence: [21:21:41] $wgServer = "//" . $_SERVER["SERVER_NAME"]; [21:21:43] # Include the complete settings file generated by mw install [21:21:45] require_once( 'orig/LocalSettings.php' ); [21:21:55] Rearranging won't work because the orig file always includes a $wgServer which typically does not work with https. [21:22:15] So the proper relative settings will in all cases be clobbered by orig/LocalSettings. [21:22:26] I understand [21:22:29] now [21:22:31] <^demon> OpenID standard should support protocol-relative ;-) [21:22:35] yep [21:22:38] will do soon [21:22:40] very soon [21:23:38] Wikinaut, go ahead and submit some kind of if/then change to the puppetized LocalSettings and I'll have a look. It's not immediately obvious to me how it should look. [21:24:15] andrewbogott: I will first check the OpenID protocol-fix [21:24:29] because of the many requests... [21:24:50] if it solves the issue. you don't have to apply a change [21:30:09] petan: / petan|ic are there any backups of home or project on bots? [21:34:30] Coren, can you do an apt-get install adminbot on the tools boxes? I need an upgrade. [21:35:19] andrewbogott: I don't have that package through apt that I can see. Where's the repo? [21:35:44] It's on brewster… version 1.7.2 is already installed, I need 1.7.3 [21:36:39] At least… I think it got onto tools-login using apt [21:37:35] Not seeing it at all. I have a project-local repo though I can stuff it in. [21:37:49] Where on brewster? [21:37:54] Couldn't find any package whose name or description matched "adminbot" [21:38:33] Um… if I do it on tools-login it tells me 'adminbot is already the newest version' [21:38:41] which is not what I was going for, but is different from what you see? [21:38:45] * Coren is a moron. [21:38:52] I was doing it on my home box. :-) [21:39:01] :) [21:39:09] That doesn't explain why it can't see 1.7.3 though [21:39:24] did you just install it? [21:39:33] (on brewster, I mean) [21:39:34] I just loaded it on brewster, yeah. [21:39:43] Needs aptitude update then. Doing it now. [21:39:46] Oh, I didn't do an update of course [21:40:05] Lemme do this project-wide [21:40:11] thanks [21:51:05] Coren, petan, addshore: Outputs are going to where they are susposed to but the bot is still broken. [21:51:13] Could not open input file: rfx-tally.php [21:53:12] andrewbogott: Version: 1.7.3 [21:53:26] Coren: Great, thank you! [21:56:08] Coren, petan, addshore: help, please? [21:56:37] Cyberpower678: Well, what error messages do you get in your .err files? [21:56:54] Could not open input file: rfx-tally.php [21:58:21] labs-morebots, doing ok? [21:58:22] I am a logbot running on tools-exec-04. [21:58:22] Messages are logged to wikitech.wikimedia.org/wiki/Server_Admin_Log. [21:58:22] To log a message, type !log . [21:59:23] andrewbogott: Is this the log bot that also triggers on "!log "? [21:59:34] andrewbogott: please can you temporarily switch off this updating of LocalSettings.php on "openid-wiki" ( https://wikitech.wikimedia.org/wiki/Nova_Resource:I-00000589 ) [21:59:43] Coren, any ideas? [21:59:53] scfc_de, yes. [22:00:07] Wikinaut: I can't, really, we'll need to patch the puppet code. [22:01:17] I cannot test my changes, and the former patch of Tyler is not working (i already merged that manually, so the https://gerrit.wikimedia.org/r/#/c/45474/ is outdated and can be abandoned) [22:01:44] you can disable puppet temporarily with sudo puppetd --disable, iirc [22:02:05] that will prevent it from clobbering your file, but also prevent it from pushing any other kind of updates [22:02:21] ori-l: "temporarily" . until when (end of the session ?) [22:02:34] until you run puppetd --enable [22:02:36] ok [22:02:51] you should probably clear it with andrewbogott or Coren first, though [22:02:51] that#s what I thought... [22:14:00] can I only submit jobs from the 'project'-login server and not from the grid execution nodes? So a grid job can't submit another grid job? [22:17:39] sitic: Yes, that's not possible. [22:17:57] ok :-/ [22:36:49] andrewbogott: I have a getFullurl question: [22:38:25] I do need this [22:38:27] OpenID: http://openid-wiki.instance-proxy.wmflabs.org/w/index.php?title=Special:OpenIDLogin/Finish [22:38:28] not this one: [22:38:30] $nt->getFullURL( '', false, PROTO_CURRENT ); gives [22:38:31] OpenID: http://openid-wiki.instance-proxy.wmflabs.org/wiki/Special:OpenIDLogin/Finish [22:39:16] Wikinaut: I don't know much; probably best asked in #mediawiki. [22:39:28] what is to be changed here $nt->getFullURL( ”, false, PROTO_CURRENT ); in order to get the article path [22:40:04] Can I get any help with my bot here? [22:41:11] Cyberpower678: What's the problem? [22:41:33] Coren, petan, addshore: Outputs are going to where they are susposed to but the bot is still broken. [22:41:33] Could not open input file: rfx-tally.php [22:42:15] scfc_de, I just updated the crontab to what they told me to update it to and now everything is broken. [22:42:34] Can you post the crontab to pastebin or somewhere? [22:43:33] scfc_de, http://pastebin.com/NiUuHZ3n [22:45:12] Spaces in your filenames [22:45:30] Cyberpower678: If you have spaces in directory names (bad, bad idea) you have to quote all your filenames "like this" [22:45:53] Cyberpower678: Much better idea: don't have spaces in any of your directory or file names. [22:46:12] Coren, where is there a space? [22:46:24] Cyberpower678: I don't see spaces, but you also have to move the "-e logfile.err" to the left of the script. [22:46:43] Cyberpower678: Ah, there is no space in your /actual/ directories. $HOME/Cyberbot II/status.err <-- has a space [22:47:16] Ah, there! [22:47:19] Cyberpower678: Lemme go fix your crontab [22:47:35] Coren, that would be addshore's doing. [22:47:54] Also in afd-delete, you have "*\2". That needs to be "*/2". [22:48:33] scfc_de, and that still worked somehow. [22:48:54] cron may be generous there. :-) [22:49:18] Coren, are you fixing it in PasteBin? [22:49:44] Cyberpower678: No, I'm fixing your actual crontab [22:50:11] The file or the actual settings? [22:50:15] Which, in fact, should be properly fixed now. [22:50:19] I use a file to store it/ [22:50:29] $HOME/crontab [22:50:49] Fixing that too, then. [22:51:46] There you go. [22:52:29] * Cyberpower678 goes to check [22:55:17] hi. I have a problem with $nt->getFullURL( '', false, PROTO_CURRENT ); [22:55:18] I need this output http://openid-wiki.instance-proxy.wmflabs.org/w/index.php?title=Special:OpenIDLogin/Finish [22:55:20] but always get hhttp://openid-wiki.instance-proxy.wmflabs.org/wiki/Special:OpenIDLogin/Finish [22:55:21] How can I _force_ the generation of the proto-relative canonical form ? [22:55:58] Wikinaut: andrewbogott asked you to ask on #mediawiki. That is the better forum for this kind of questions. [22:56:16] oops. sent to to wrng channel [22:56:17] sorry [22:56:23] was already in the other... [23:00:09] Coren, thanks. Bot is operational. Now to fix the error messages.