[04:18:37] anyone alive? how do I stop puppets from ruining LocalSettings? [04:19:17] i had a perfectly configured instance, now every now and then some rogue puppet runs and resets everything :( [04:38:57] yurik, for a labs instance only modify orig/LocalSettings.php. For a vagrant instance, see https://www.mediawiki.org/wiki/MediaWiki-Vagrant/Customizing [04:39:33] spagewmf, you mean it modifies /srv/mediawiki/LocalSettings.php, not the one in orig? [04:39:46] i think i might have been modifiing orig/ for a long time [04:39:55] i mean - i was modifying non-orig [04:40:08] i just saw a note at the top of that file [04:40:15] so modified the one in orig/ [04:40:19] will see if it works :) [04:40:32] hi yurik [04:40:38] hi Amir1 [07:48:33] Oren_Bochman get a better client, like pidgeon [07:48:35] :P [07:48:37] you won't be tired [08:09:11] pidgin is a horrible client [12:00:11] How to get SVN commit access? [12:00:39] o.O [12:00:47] where? [12:01:06] Pywikipedia [12:01:27] you usually ask a maintainer of that repository? [12:01:42] I don't think pywikipedia has anything common with labs [12:06:11] But how come http://svn.wikimedia.org/users.php has my labs user name in it, and http://www.mediawiki.org/wiki/Developer_access/Subversion#Updating.2Freplacing_your_SSH_key seems to say that the users with labs account can update their SSH key in labs? [12:10:32] I've posted on http://www.mediawiki.org/wiki/Thread:User_talk:Siebrand/How_to_be_a_coder%3F, but the response is so/too slow. [12:19:57] ? [12:33:59] Leaving, I'll see the log. [12:34:18] addshore, ping [12:47:17] Cyberpower678, you'll always succeed in pining me as long as my router PPPoE is connected. [12:47:43] zhuyifei1999, ?? [12:51:02] I mean, as long as my router is on and my IP isn't changed, you can succeed in pining me. (But I can disable it too.) [12:51:21] zhuyifei1999, totally random, but ok. [12:55:16] addshore, ping ping [12:55:23] pong pong pong [12:55:33] addshore, https://github.com/cyberpower678/Peachy [12:56:01] :D [12:56:55] addshore, and you don't need to worry about accidental uploads of personal config files. [12:57:03] They won't commit [12:57:03] ? :O :D [12:57:06] :> [12:58:15] addshore, all have to do is fork the repo, make your changes and submit a pull request. [12:59:06] * addshore goes to have a look [12:59:35] Cyberpower678: does it all sort of work? :> [13:00:16] With the exception of a few minor bugs, yes. [13:00:30] You are looking at alpha-3 [13:04:54] i will have a play with it soon :D [13:08:05] Oink, oink. [13:08:23] Today on the schedule: federated commons test. :-) [13:08:30] Coren, S7? [13:09:05] Cyberpower678: Dunno. I'll ask Asher as soon as he comes in (he's in PST, so in a couple hours) [13:09:19] Coren, legal? [13:09:45] Cyberpower678: Last update status I had was that they were (a) swamped and (b) still working on the previous question (re email) so not yet. [13:09:52] Coren can I haz IP [13:09:58] toolsbeta [13:10:02] I never thought I'd say this, but "we need more lawyers" :-) [13:10:14] no please :P [13:10:21] Coren, I am getting hammered with questions on why the deleted edit count isn't working. :p [13:10:21] no more lawyers... [13:10:53] Cyberpower678 reply with "forward your question to Marc-Andre pls" :D [13:10:59] :D [13:11:02] you could even make a template from it [13:11:50] Coren this needs merge https://gerrit.wikimedia.org/r/#/c/69133/ [13:38:06] Sorry for Cyberpower678, I thought you meant pinging my IP. And Petan, can you answer my question now? [13:42:42] zhuyifei1999 related to svn? no [13:42:51] I have no idea who maintain pywikipedia svn [13:43:38] Cyberpower678: going to go and try to amke some stuff with peachy now :) [13:43:47] zhuyifei1999: Siebrand answered the question before your last a few hours ago; you need to have a little patience. Or type "@notify siebrand" so that a bot will notify you when he joins IRC. [13:45:18] zhuyifei1999: we should be on gerrit in two months [13:45:20] Scfc_de, where in http://bots.wmflabs.org/~wm-bot/logs/%23wikimedia-labs/20130618.txt [13:45:41] apart from that, you can submit patches on sourceforge.net [13:45:57] https://sourceforge.net/p/pywikipediabot/patches/ [13:47:13] zhuyifei1999: http://www.mediawiki.org/wiki/Thread:User_talk:Siebrand/How_to_be_a_coder%3F [13:48:11] addshore, tell me how it goes. If you need help, let me know, but the framework makes everything so simple, you shouldn't need much help. :) [13:50:09] valhallasw, that seem to be inactive, the last one seem to be 7 days ago. scfc_de, he hasn't replied to me yet. [13:51:17] Or I can wait for gerrit. But bough [13:53:16] I mean both git.wikimedia.org and gerrit.wikimedia.org are sometimes buggy, and even offline for a while. [13:57:56] zhuyifei1999: I know that he hasn't replied to you yet, that's the reason why I suggested a little patience. If you experience bugs or downtimes in Gerrit, please report them. [13:58:53] gerrit is also replicated to github so you can contribute there [13:58:59] github also has readonly svn if you want [14:01:31] YuviPanda: readwrite, actually [14:01:38] oh?! [14:01:40] I didn't know that [14:01:42] YuviPanda: if you use your own fork, at least [14:01:53] https://github.com/blog/1178-collaborating-on-github-with-subversion [14:07:41] Cyberpower678: bug tracker? [14:10:36] Cyberpower678: ? :< [14:10:53] addshore, what's wrong? [14:11:06] You can open an issue on GitHub [14:11:08] well, is there somewhere i can report things? :P [14:11:12] kk :) [14:12:03] addshore, you haven't forked the repo yet? [14:12:09] nope [14:13:24] *tries to work out how to do inline comments* [14:18:38] addshore, inline comment? About what? [14:19:01] brb [14:32:51] hey Merlissimo - just so you know, there is now a graphvserv instance on a labs project. [14:33:54] :D [14:34:01] ok lol [14:34:49] all those ping timeouts -- looks almost as bad as the wikimedia de office wifi ;) [14:52:52] can somebody with python experience please look at http://pastie.org/8055628 and help me fixing it? [14:53:44] Pyfisch: can you show me the code? [14:53:48] probably a missed .encode somewhere [14:54:01] it is on tools [14:54:07] YuviPanda toolsbeta-mc is ready for your experiments [14:54:23] Pyfisch: tools where? [14:54:32] here: /data/project/wdrdr/cgi-bin/index.php [14:55:15] Pyfisch: err, index.php? [14:55:23] ah [14:55:23] cgi [14:55:57] YuviPanda: you can also see the error at http://tools.wmflabs.org/wdrdr/cgi-bin/index.cgi?item=q1 [14:56:08] addshore, I'm back [14:56:25] You found old stuff, waiting to be updated. [14:56:27] if you set your browser lang to russian or some other not ascii script language [14:56:32] howdy ;p [14:57:27] Pyfisch: i just urlencoded that, check /tmp/index.cgi see if it works? [14:57:33] (I don't have write access to your file) [14:58:10] petan: sweet [14:59:03] (there are other issues with the python file, I'd reccomend using flask :)) [14:59:10] (it'll take care of things like this) [15:00:11] Pyfisch: let me know how that goes :) [15:00:47] YuviPanda: just checking :-) [15:00:52] :) [15:03:55] YuviPanda: is not working: http://tools.wmflabs.org/wdrdr/cgi-bin/indexX.cgi?item=q1 :-( [15:04:16] where do you see the traceback? [15:04:37] * YuviPanda waits for Coren to setup WSGI sometime, so I can start telling people to just move already [15:04:58] YuviPanda I can set it up but I thought Coren should approve it [15:05:14] petan: indeed, he commented on the bug a while ago about 'found a nice way to do it' so am waiting on him :) [15:05:20] + I would prefer to test it first on toolsbeta [15:05:42] yeah yeah, I can wait [15:05:56] YuviPanda: nowhere because it does not create traceback for webbroser like normally [15:06:08] Pyfisch: are you just starting this tool? [15:06:12] or is it being ported frome lsewhere? [15:06:22] if you are just starting this, I highly reccomend using something like Flask [15:06:28] it will kill all these stupid problems [15:06:40] writing CGI directly is out of date by about 10-15 years [15:06:51] Pyfisch: I can help you with setting up flask on toolserver too, if you want [15:07:14] also consider using a git repository. indexX, indexX2, etc values get messy very quickly [15:07:42] YuviPanda: thanks for your help, but I don't really want also to start learning flask a little bit after django and cgi [15:07:50] :( [15:07:51] okay [15:08:06] I can't do much about CGI though, sorry [15:08:10] legoktm might be able to help [15:09:17] this should be a VERY little tool so i thought that i can do it with cgi and i didn't think that I will need git [15:09:43] I personally think using CGI directly is a very bad idea for anything, ever. [15:10:29] YuviPanda: a question for a different project: how easy is flask? [15:10:35] very [15:10:37] it is a microframework [15:10:49] so it takes about 20 minutes to fully understand it [15:10:49] much easier than django? [15:10:52] yes [15:10:55] it is the anti-django almost [15:11:15] http://flask.pocoo.org/ [15:11:26] look at the sample code, etc [15:11:27] super simple [15:13:49] Pyfisch: I can set that up + virtualenv for you on your tool account pretty easily [15:17:27] YuviPanda: how does the web server call into it? is WSGI working yet, or is it fcgi? [15:17:36] there are working CGI wrappers [15:17:47] it's 4-5 lines of code, copypasteable [15:35:15] * Coren wonders what commonswiki tables are *actually* joined with the other project databases in practice. [15:38:48] Coren: I asked multichill, he probably has an idea [15:41:41] valhallasw: Thanks. [15:42:00] * Coren has user and revision working for both wikidata and commons from enwiki now. [15:45:08] Oh yeah, also page. [15:53:34] Coren hey I have a question [15:53:46] Coren imagine you have a server which you want to shut down [15:53:57] I have lots of answers. Maybe one of them even relates to your question. :-) [15:54:08] Coren it is exec node, and there are running jobs, you want to disable it so that running jobs will finish but new jobs don't get submitted [15:54:24] petan: Simple: just remove it from the host group. [15:54:38] petan: http://i.qkme.me/3uwcag.jpg ;-) [15:54:50] Coren will it preserve the currently running jobs? [15:54:58] * Coren nods. [15:55:04] The queues are only checked when scheduling. [15:55:23] ok [15:55:48] petan: and you should always be able to force a re-schedule of jobs in the continuous queue [15:56:30] wait lol there is "halt" command I didn't even know it [15:56:36] I always did shutdown -h "now" [15:59:30] !log bots petrb: scheduled bnr3 for removal [15:59:33] Logged the message, Master [16:02:01] there are /still/ people running bots on bots-4 by hand o.O [16:02:19] they didn't manage to even convert them to grid, how could I want them to move them to another project... [16:51:55] petan: Set deadlines :-). [17:01:23] hi [17:04:52] are there any limitation in framework usage? (more specifically, Laravel) [17:19:22] fale: Not specifically, although the more obscure one may find fewer people to help with. [17:19:26] ones* [17:20:54] Coren: I see. Laravel should not be obscure (the intent of using an MVC framework is to make code not obscure) :). So there are no milits for "performaces" reasons (on ToolServer was against the rule to install sw like MediaWiki andsimilar) [17:22:24] fale: We have a fair bit more resources than the TS had, so it is unlikely to be an issue. If it's really out of the sky, we may need to make special arrangement but -- as a rule -- there is no a priori limitations. [17:23:51] Coren: I see :) (MVC framework should, in any case, not make the project heavier, maybe bigger (the 100mb that the TS gived me were just enough for 3~4 project) :) [17:24:24] Coren: I'm tring to change my mind-set from a TS pov to a Labs pov :). thanks a lot for your help :) [17:24:56] fale: Shouldn't be too much of an issue: labstore1.pmtpa.wmnet:/keys 36T 10T 26T 28% /public/keys [17:24:56] If you manage to eat a significant chunk of that, /then/ we will have words. :-) [17:25:26] Coren: :D I'll take no more than 200mb for now :D [17:47:16] Coren: Any plans to work on https://bugzilla.wikimedia.org/show_bug.cgi?id=48851 ? [17:49:24] multichill: I am aware of the issue; this will "solve itself" once the new maintenance scripts are in place (user databases will be organized slightly differently on the DB which should alleviate the issue) [17:50:20] And of course you have a bug open for this change? [17:54:47] coren? [17:59:35] multichill: well, it seems the bug you linked would act as that bug :) [18:02:18] multichill: Ryal_Lane, as is oft does, speaketh sooth. [18:04:39] bully [18:04:53] wait. am I using the wrong time period/place? :) [18:13:43] Coren: are there any query limits? (or, is there any page that describes all limits doft and hard) [18:17:38] fale: None yet. I'll put some into place iff they become required, and infom the labs-l if that ever becomes the case. [18:18:18] Coren: so I can run a 8-minutes query without worring about long query and things like that? [18:18:57] Coren: any eta on commons joins? [18:19:12] At this point, right. Albeit if your query takes this long, it might be worthwhile to review it to see if it can be made more efficient through tweaking or application logic (they usually are) [18:19:54] Betacommand: I've got a prototype in place on enwiki <-> commons and wikidata, with only a couple of tables. Once I know which tables are really needed, I'll have a better eta. (atm is user, revision and page) [18:20:14] Coren: image [18:20:17] Coren: Thanks :). I'll into it, even if I'm pretty n00b at query optimization, and eventually I'll try to ask some help here :) [18:20:30] Betacommand: That... should have been obvious to me for commons. :-) [18:20:45] * Coren adds image now. [18:20:46] Coren: KISS [18:20:57] Its always the small stuff you forget [18:22:47] Coren: might I suggest logging and filearchive [18:22:50] Betacommand: commonswiki_f_p now has image [18:23:19] Coren: whats up with the bizarre name? [18:23:35] Where can I find the exception logs for a load.php failure on http://en.wikipedia.beta.wmflabs.org? [18:23:38] commonswiki_(federated)_(public) [18:23:46] ah [18:24:17] you cant just call the federated commonswiki_p ? it would make things far easier for tools [18:25:02] Betacommand: I _can_ but I'd rather not. There are gotchas about how to "properly" use federated tables that I want tool maintainers to keep in mind; forcing the table rename is the best way to flag that. [18:25:42] * Betacommand grumbles about the mass re-write needed [18:26:15] Betacommand: Assuming your code is already careful about using indices, it's just a single sed. :-) [18:26:28] s/commonswiki_p/commonswiki_f_p/g [18:26:29] :-) [18:27:32] Betacommand: No filearchive. That one isn't replicated at all to begin with. logging is no problem though. [18:27:35] * Coren goes to add that. [18:27:56] Coren: adding file archive would be useful for the file hashes [18:28:55] you could search for previously deleted files via hash [18:30:12] Coren: why would filearchive be an issue? [18:31:41] Betacommand: That one has been axed by legal and isn't likely to come back; I might be able to weasle them into creating a view with just hashes if I ask nice enough. :-) [18:32:31] Coren: my thought would be hash, filename, uploader, date,time [18:32:58] having the bare hash withoout context is almost useless [18:33:17] Betacommand: Noted. I'll ask. [18:33:53] Coren: what part of the file archive would be deemed too sensitive? [18:34:13] Other than the hash the rest can be found in the logging table [18:34:43] except for the upload summary [18:34:58] hashar, do you know where the exception logs are for a load.php to http://en.wikipedia.beta.wmflabs.org/ are? [18:35:04] I have the exact URL if that helps. [18:35:04] Betacommand: It may well be the hash itself (being proof that file X was on enwp during a range of dates). [18:35:06] * Betacommand points to log/upload [18:35:24] Betacommand: I'll have to ask. [18:35:34] Coren: why would that be a legal issue? [18:35:51] Betacommand: IANAL. :-) [18:35:59] superm401: /data/projects/logs iirc [18:36:09] superm401: at least that is where the udp2log are written too [18:36:25] All I can find is the Apache log, which doesn't help. [18:36:29] superm401: there are some apache error logs there as well [18:36:30] Betacommand: At the very least, I expect, it doesn't fit the "only avaliable to non-admins" baseline. [18:36:54] superm401: /data/project/logs/exception.log should have what you need [18:36:55] hashar, It's hard without the error code (2ddf5, etc.) you sometimes get in prod. [18:36:59] Betacommand: Which, mind you, is a baseline and not a hard requirement. Hence, asking for it back may be okayed. [18:37:06] Alright, I'll take another look in there. [18:37:15] superm401: the fatal are not logged though. They are hardcoded to be sent to production, I got some patches pending in ops/puppet.git though [18:37:35] "They are hardcoded to be sent to production" [18:37:38] What does that mean? [18:37:44] I think it probably is a fatal. [18:37:45] It's a 500 [18:37:50] definitely a fatal :D [18:37:58] * Betacommand goes to get a link [18:38:04] so we catch fatal errors using a PHP extension 'wmerrors' [18:38:42] its configuration is deployed by puppet which has something like: wmerrors.log_file = udp://someproductionIP:666 [18:39:10] superm401: https://gerrit.wikimedia.org/r/#/c/68830/1/modules/applicationserver/templates/php/wmerrors.ini.erb,unified :) [18:39:30] Coren: for non-deleted info http://dumps.wikimedia.org/commonswiki/20130613/commonswiki-20130613-image.sql.gz lists all of that info [18:39:59] its trivial to cross reference several dumps and get a list of what has been deleted [18:40:11] hashar, oh, I see, the fatals are logged to the prod log. [18:40:19] But it doesn't even log there since the networks are isolated, right? [18:40:25] I hope so :) [18:40:32] Betacommand: I am aware. [18:41:30] Betacommand: logging (and logging_userindex) now in federation. [18:42:21] Coren: thanks [18:42:22] Betacommand: You can test with enwiki now [18:42:57] Coren: it wont be till this weekend when I have enough time to migrate my db code to labs [18:43:16] Betacommand: FYI, the "gotcha" about federation is fairly simple to cope with: if you are querying a federated table, keep your result set as small as possible through where clause(s) on index columns. [18:44:11] Coren: Any specific reason why you not federate simply all tables? [18:44:11] some queries are still going to be a bitch [18:44:13] scfc_de: It's a pain to maintain. [18:45:07] Betacommand: Honestly, I think that most cases that won't fit this fairly generous model can be rejiggered advantageously by talking to commons_p directly instead and using application logic. [18:45:30] Betacommand: I'd love to be shown a query that won't work right that's in actual use though, to see if I can work around it. [18:46:00] Coren: Show Puppet config, and maybe I show pity :-). Isn't it simply "delete all federated tables commons_f_p.*; for TABLE in commons_p.*; do something with $TABLE; done"? [18:48:50] Betacommand: I should say, perhaps that is worded to stringently. The actual gotcha is "keep the [part of] the result set [that comes from rows on the federated table] as small as possible". [18:49:18] There is no particular limit on the examined rows on the /local/ db. [18:51:20] A 1:1 join on an index column should, for the most part, be almost exactly as fast as if they were both local databases. [19:06:36] Coren: Get a list of missing files and what articles that are use them on enwiki [19:07:24] You means files that are in articles but neither on enwiki nor commons? [19:08:21] yeah\ [19:08:49] IE "missing files" [19:09:21] * Betacommand goes to find a the code that he tried to use for that using app logic [19:09:35] it couldnt load enough data into memory [19:27:32] Coren / petan: could I have qmake on tools-dev? :-) [19:29:44] valhallasw: Caveat: qmake is also a variant of qsub where individual tasks are run with SGE. But you mean Qt's make, I suppose? [19:29:59] *grin* yes :-) [19:31:58] Betacommand: Hm. With ~50m imagelinks this would be a painful thing to run even on local databases indeed. [19:35:18] * Coren tries something fancy. [19:35:22] Pyfisch: hey, do you still need help? [19:35:41] legoktm: yes [19:36:58] Pyfisch: somewhere near the top, add in "import cgitb; cgitb.enable()" [19:37:07] that will spit out a traceback if possible [19:37:18] legoktm: this is already in. [19:37:25] hm [19:37:44] does it work if you run the script from commandline? to catch syntax errors, etc [19:39:14] legoktm: it is running and working fine, except for non ascii letters. [19:40:05] o.O [19:40:26] ascii internally? or as a url parameter? or.... [19:41:25] legoktm: I can't print non ascii letters from a wikipedia article url. [19:42:04] hm [19:42:10] is it giving a traceback? [19:42:35] legoktm: yes. [19:43:18] legoktm: what is your default browser language? [19:43:23] http://tools.wmflabs.org/wdrdr/cgi-bin/index.cgi?item=q1 [19:43:25] en-us [19:43:33] ooh, neat! [19:43:59] if you can set it to russian or chinese (or also german in some cases) you will see the error message. [19:45:12] http://tools.wmflabs.org/wdrdr/cgi-bin/index.cgi?item=Q712570 also works for en-us [19:45:51] Pyfisch: you need to print an urlencoded representation of the unicode string [19:47:06] urllib.quote(u'booö'.encode('utf-8')) [19:49:22] what he said ^ :D [19:54:11] valhallasw, legoktm: it works! thanks! [19:54:31] Pyfisch: cool! :-) [20:18:18] hi [20:24:06] meh [20:24:21] Coren the puppet class for webproxy doesnt work [20:24:40] petan: It's not so much "not working" as in "not written". :-) [20:24:54] Coren it say some weird error so I cant even switch to nfs on it [20:25:01] o_O? [20:25:03] have a look at toolsbeta-webproxy [20:25:19] I'm in the middle of something, can you pastebin the error? [20:27:02] yes [20:29:58] Coren: so, yeah, that NFS issue scares me [20:30:05] for switching everything to NFS [20:30:55] Ryan_Lane: Understandably so; albeit the recovery is swift. Further googling convinced me that the combination of rsize and scheduling is the cause. [20:31:15] Specifically, that scheduling is the cause and the bigger rsize made it worse. [20:32:07] heh [20:32:10] so, deadline? :) [20:32:37] I'm going to continue rsyncing data across [20:32:44] and hopefully next week we'll switch homedirs [20:32:50] * Coren nods. [20:38:11] something is not really clear to me. To do an sql query, I should: login to bastion.wmflabs.org -> log to another server -> run the query? [20:38:29] fale: you can do a direct ssh to tools-login.wmflabs.org [20:39:19] legoktm: thanks :), I'll try this way :) [20:39:37] Coren: did you see my thought about bastion and access to all shares? [20:39:47] Ryan_Lane: No, where? [20:39:58] mentioned it in irc yesterday [20:40:34] Coren: so, my thought it that since bastion doesn't allow root and everyone is a member that we might be able to share all project data shares to it [20:40:47] and mount it under directories with locked down permissions per project [20:41:03] legoktm: I'm not allowed to login, I think :( [20:41:06] I see no reason why not, but I see no reason why either. [20:41:18] Coren: so that people can scp/sftp directly to it [20:41:21] fale: er, has your account been approved to be in the tools project? [20:41:37] and so that they could push to gerrit without forwarding their keys [20:41:47] Ryan_Lane: Oh. Not useful for tools since it has its own bastion, but I see your point. [20:41:57] legoktm: not yet, I opened the request today. I'm only in the shell group for now [20:42:20] fale: ah, well Coren can approve you for that :) [20:42:39] Coren: it may be possible to eliminate the tools bastion with this kind of change [20:44:08] Ryan_Lane: I'd want to keep it for simplicity's sake (most tools users don't/can't work with proxycommand and a big fraction have windows where this is a pain at best) [20:44:21] Coren: there wouldn't be a need for proxycommand ;) [20:44:28] they would work directly from the bastion [20:44:39] o_O [20:44:57] any reason that couldn't work? [20:44:58] Most stuff does require to be connected to tools-login interactively. [20:45:10] Start jobs, edit crontabs, etc. [20:45:11] basically the bastion would become tools-login [20:45:42] I'm not going to allow a host outside the project to become a submit host for the grid. :-) [20:45:43] * fale is pretty confused by the permission system in labs... probably I have to study more the wiki (even if topics are not easy to find) [20:45:55] fale: Give me a minute and I'll add you. [20:45:58] Coren: isn't it based on group membership anyway? [20:46:22] fale: basically, try to ignore everything that says 'labs' inless it's in the combination 'Tool Labs' ;-) [20:46:33] Ryan_Lane: It's also not going to work if there is more than one grid to submit /to/. gridengine doesn't do multitenancy. [20:46:51] Coren: http://monkeyswithbuttons.files.wordpress.com/2010/09/proxy.png?w=614 for the windows users [20:46:56] * Ryan_Lane nods [20:47:21] Coren: basically what I was looking to do was simplify workflow and docs [20:47:27] Coren: thanks :) [20:47:39] valhallasw: lol, thanks for the suggestion :) [20:47:40] no more "log into foo" and "log into bar", but "log into the bastion" [20:47:51] Ryan_Lane: Honestly, I think that's putting all the eggs in one very complicate basket for very little gain IMO. For a lot of "true" projects, scp directly to /data/project is a win. Not sure it'd be so for tools. [20:48:24] well, plus being able to push to gerrit [20:48:32] which is… difficult for most projects right now [20:49:54] electricity outage in part or Prague I live :/ [20:50:01] my laptop will stand for a bit though [20:50:07] Coren: part of the difficulty right now is people being confused by "labs" and "tool labs". we should probably start referring to "tool labs" as just "tools" [20:50:11] Coren I will send that puppet error [20:50:15] it makes explaining things way easier [20:50:57] Ryan_Lane: I didn't get to pick the name. :-) [20:51:12] we can make the names whatever we want :D [20:51:17] oh. gotta run [20:51:19] back in a while [20:51:33] e: /Stage[main]/Toollabs::Infrastructure/File[/etc/update-motd.d/40-infrastructure-banner]: Could not evaluate: Could not retrieve information from environment production source(s) puppet:///modules/toollabs/40-toolsbeta-infrastructure-banner at /etc/puppet/modules/toollabs/manifests/infrastructure.pp:23 [20:51:37] error [20:51:40] :o [20:51:41] Coren [20:51:50] Coren: there's no reason not to use a different name for external communication than for labs-internal communication ;-) [20:52:20] I'd have called it the Forge. :-) [20:52:45] "The Forge is the Toolserver replacement run by the WMF" [20:52:46] that works. [20:52:52] !ping [20:52:52] pong [20:52:59] am I here? [20:53:05] petan: yes [20:53:08] petan: no [20:53:10] petan: Yeah, but you're not all there. :-) [20:53:41] petan: That puppet is broken. It looks like its configuration is so broken it can't read its configuration anymore. :-( [20:54:04] Coren: Calling it The Forge would be good in distinguishing it both from TS and the rest of labs, actually [20:54:23] fale: You've been added. [20:54:34] Coren: thaks :) [20:56:25] !ping [20:56:25] pong [20:56:36] eh [20:56:41] did you see what I write before? [20:56:47] there is electricity failure :/ [20:57:23] I am on batteries now [21:01:06] mysql query are fine to be done from cli or is better to execute them from gridscheduler? [21:13:50] * fale is confused: ERROR 1142 (42000): SELECT command denied to user 'u2896'@'10.4.0.220' for table 'page' [21:14:10] Betacommand: How long does that missing file query usually take for you? [21:16:01] fale: did you use the replica.my.cnf file instead of the .my.cnf file? [21:16:09] (you should use the replica one) [21:16:18] legoktm: yep [21:16:31] legoktm: mysql --defaults-file=~/replica.my.cnf -h itwiki.labsdb [21:16:34] weird…. [21:17:57] fale: You're doing this to the itwiki_p table right? [21:18:08]