[00:19:58] * Betacommand grumbles about server differences [00:47:20] server admins need to stop killing server kittens, its not nice [00:50:17] Betacommand: What specifically are you having difficulty with? [00:50:36] Hasteur: several things [00:50:49] different configurations between servers [00:50:55] is the root of several issues [00:51:10] Consistency against the compute cloud machines? [00:51:25] Betacommand: what is different? [00:51:49] YuviPanda: one was an issue with whois [00:52:08] webservers have /bin and the login servers have /usr/bin/whois [00:52:39] IE on login servers I can just call whois whatever [00:53:02] however on webservers I must define the full path due to apache screwing things up apparently [00:53:18] ah, right. path not being inherited from env [00:54:05] after defining the path correctly I still get errors on 2/3 of the requests (keep in mind Im just refreshing the samepage) [00:54:28] it works once then doesnt the next refresh [00:55:04] hmm, let me check if whois is in puppet [00:55:08] if it isn't, that'll explain all the problems [00:55:25] YuviPanda: its part of the core linux [00:56:10] also config and env should be 100% the same across all servers [00:56:37] if its going to not work it should not work on all servers [00:57:00] Coren|Away: ^ [00:57:23] right now it works 100% on login, and about 1/3 on the web servers [00:57:48] want to talk about wanting to gouge my eyes out?? [00:58:08] you really cant make it much more of a pain in the ass to debug stuff [00:58:30] what else did you run into? [00:59:04] Apache being about as obnoixous as a [00:59:31] Apache being about as obnoixous as a 4 year old on crack, caffeine and about 10kg of sugar [00:59:47] are your tools in php? [00:59:51] No [00:59:51] or something else? [00:59:54] python [01:00:12] hopefully in a month you won't have to deal with apache then. I should be able to get uwsgi up by then. [01:00:35] uwsgi ?? [01:00:50] http://projects.unbit.it/uwsgi/ [01:01:01] more importantly, that means you'll get proper error / access logs for python too [01:01:53] YuviPanda: will I need to change my code, or will it just make things work? [01:02:07] you won't need to change your code. [01:02:14] like how you start your jobs with jsub / jstart now [01:02:17] there'll be uwsgistart [01:02:24] so you just do uwsgistart [01:02:28] and that'll be it [01:03:07] right now apache refuses to display anything unless I send headers with a content type\n\nHTML [01:03:39] if there is a space on the second line god forgive because apache sure as hell wont run your code [01:03:44] wait, I made a big assumption. that you are using not using CGI. [01:03:56] if you want to use CGI, you'll be stuck with Apache. [01:04:08] YuviPanda: then Im still fucked [01:04:17] well, if you're using CGI in 2013, then yes, you are fucked [01:04:38] that's the entire reason I've been doing the uwsgi work. because fuck CGI. [01:05:15] YuviPanda: the code isnt that complex and these scripts are invoked process the information and die, which works great on the toolserver [01:05:39] coming to labs has been one nightmare after another [01:06:17] as I said, if you're using CGI in 2013, you're fucked at most places. [01:06:25] YuviPanda: not really [01:06:27] * YuviPanda needs to go afk in a couple of minutes [01:06:32] okay, let me rephrase [01:06:43] in my opinion, if you're using CGI in 2013, you're fucked at most places :) [01:07:03] anyway, going afk now. [01:07:06] good luck! [01:07:09] if things where uniform across servers and apache wasnt OCD things would be decent [01:07:24] sure, the things not being uniform sounds like a bug [01:07:39] IIRC you emailed labs-l about it, so Coren should be able to get to it when he wakes up [01:08:19] ciao! [05:55:22] !ping [05:55:22] !pong [05:55:30] it's just quiet around here, then :) [13:56:53] Hey there! I would like to have a clone of the Wikidata database on my tools instance. How can I load a dump there? Should I download it from here? http://dumps.wikimedia.org/wikidatawiki/ And is it correct if I create the new database as described here? https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/Help#Creating_new_databases [14:29:46] Granjow: A dump? There is a replica of the Wikidata db. [14:30:08] https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/Help#Database_access You should be able to access it. [14:31:50] Silke_WMDE: I also want to insert data. Downloaded the wikidata dump now and will try to load it with mwdumper ... [14:34:37] Shouldn't my project tool account have a replica.my.cnf as well? [14:36:18] there shoudl be one in the tool's home dir [14:40:59] Silke_WMDE: There isn't for lifeweb. [14:44:46] Granjow: not in your own project, this apply for tools only [14:45:25] if you aren't working in tools project, you may not have access to replica, but you can of course load a dump, mwdumper is likely most fast, but it will take several days anyway for full history [14:48:10] petan: So I would take the cnf from my user's home directory? [14:48:36] For the dump; I do not need the history, only the most recent data. Does this take that long as well? [14:48:57] depends on size of wiki, simplewiki took around 1 day [14:49:34] Okay ... that's a lot. [14:55:11] Considering that wikidata is around 15 times larger, no, I won't load the dump. [14:57:29] petan: Have you ever thought about making separate IRC channels for Labs vs. Tool Labs? Or doesn't that make sense? [14:58:04] I think Ryan wanted to keep as less channels for labs as possible, but I don't have anything against that.. [14:58:10] Hm, can I install APC on my instance? [14:58:17] this channel is silent most of time anyway [14:58:31] Granjow: not sure what the shortcut stands for [14:58:41] Caching for PHP [14:58:57] aha [14:59:06] first what comes to my mind is APC UPS :P [14:59:13] :) [14:59:14] of course you can [14:59:20] :D [14:59:41] cool. How can I become root? [14:59:46] sudo su [14:59:58] are you still talking about your project right? [15:00:06] yes [15:00:13] then sudo should do that [15:00:15] what is my password? [15:00:27] if it ask for password you probably don't have root [15:00:34] :( [15:00:37] are you sysadmin of that project? [15:00:46] if it's /your/ project you should be [15:00:59] @labs-user Granjow [15:00:59] That user is not a member of any project [15:01:06] what's your labs username [15:01:35] no idea, but I created it [15:01:49] is it labs project or tools project? [15:01:55] you don't know your username? [15:03:11] Well, I log in to tools-dev atm (or tools-login), as granjow. [15:03:34] LivingShadow [15:03:40] @labs-user LivingShadow [15:03:40] LivingShadow is member of 2 projects: Bastion, Tools, [15:03:55] ok it's a tools project because you aren't member of any other labs project :-) [15:04:25] in that case, you can't install APC yourself, but you can request that, using bugzilla [15:04:35] !broken [15:04:35] oh no, so fix it! dam you [15:04:39] jesus [15:04:42] @search bugzilla [15:04:42] Results (Found 6): bz, osm-bug, report, bug, rb, tools-bug, [15:04:45] !report [15:04:46] to report new bug open: https://bugzilla.wikimedia.org/enter_bug.cgi [15:04:48] here we go [15:04:53] What is the difference between labs and tool labs? [15:05:01] Can I get root access? [15:05:07] labs is a place where tool labs are hosted [15:05:18] I think we had some terminology FAQ [15:05:24] @search terminology [15:05:24] Results (Found 1): terminology, [15:05:27] !terminology [15:05:27] https://labsconsole.wikimedia.org/wiki/Terminology [15:05:45] this is not what I meant [15:06:33] well, the labs is shortcut for wikimedia labs, which is basically a virtual host provider (you can create projects there which consist of virtual machines) tool labs is a project of wikimedia labs [15:07:27] So I have a tool labs tool. [15:07:36] Or a tool labs lab tool? [15:07:55] you have a tool which is part of tool labs (which are part of wikimedia labs) [15:08:38] tool labs are meant to host bots or tools that are simple enough that they don't need to run on dedicated servers (to save resources) [15:09:23] Oookay. I have no idea how much resources my tool needs, but I eventually have to scan the whole DB, if that works at least. [15:09:51] if your tool require so much specific configuration (configure OS parameters for your tool to work, install some software that can't be on tools project etc) then you can request own project on labs, where you would need to install everything and setup OS yourself [15:10:25] but for most tools a tools project is enough :) [15:11:08] Would be easier. [15:11:09] TBH I think scanning of whole DB using dumps can be done on tools just as on any other project, so that's no reason to require own project [15:12:27] ok. Will ty. [15:12:27] try. [15:14:03] Hm. Can I set up a wiki on my tools tool that uses the replica of wikidata? [15:15:06] ah, I can [15:15:41] not [15:15:54] because it is read-only. [15:16:27] So I guess I will just read parts of the dump ... [15:19:50] Or not. Because I cannot create a new DB. [15:20:29] moment [15:20:31] !toolsdocs [15:20:31] https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/Help [15:21:02] I followed those steps, and when I try to [15:21:04] mysql --defaults-file="${HOME}"/replica.my.cnf -h tools-db [15:21:12] then it says access denied. [15:21:33] !toolsvslabs is https://wikitech.wikimedia.org/wiki/Help:Labs_vs_Tool_labs [15:21:34] Key was added [15:22:16] Granjow: you can create new db's [15:22:32] you can always use "sql" instead [15:22:38] just type sql [15:22:40] for example: [15:22:43] sql enwiki [15:22:48] petan: How? [15:22:57] or type sql --help [15:22:59] in terminal [15:24:56] And then I create table sqluser__dbname and Maria says denied. [15:26:00] sqluser?? [15:26:09] you should replace that with your sql name [15:26:15] I mean your sql username [15:26:24] which you can read from replica.my.cnf [15:26:27] ~/replica.my.cnf [15:29:50] yes, I used create table u3409__lifeweb [15:33:36] Is this because it is only a tools tool and not a tool? [15:34:30] not a project, I mean [15:35:07] you should be able to create db's, though I am unsure if using your user account as well [15:35:07] did you actually try to do that as a tool? [15:35:07] you need to type "become toolname" [15:39:22] yes. [15:39:57] Need to leave now; will maybe try again tomorrow. [17:34:33] Is there a password-less phpmyadmin on labs yet? [17:52:30] * Hasteur makes a comment about command line and real coders [18:33:06] Hasteur, Real Coders don't use the command line http://xkcd.com/378/ [18:36:38] [bz] (8NEW - created by: 2Jesse PW (Pathoschild), priority: 4Unprioritized - 6normal) [Bug 53668] Some replicated databases are missing tables - https://bugzilla.wikimedia.org/show_bug.cgi?id=53668 [18:36:39] [bz] (8NEW - created by: 2Peter Bena, priority: 4Normal - 6normal) [Bug 48930] (Tracking) Database replication services - https://bugzilla.wikimedia.org/show_bug.cgi?id=48930 [18:37:31] [bz] (8RESOLVED - created by: 2Pietrodn, priority: 4Unprioritized - 6normal) [Bug 52370] Several replicated DB are missing tables and content - https://bugzilla.wikimedia.org/show_bug.cgi?id=52370 [18:47:08] Dispenser: phpmyadmin is not allowed in labs [18:50:15] a number of instances have been owned because of it [18:50:17] Question: Is the text of the current version of a page accessable from the labs replicated databases? [18:50:47] Hasteur: no. it's recommended to pull the text from the production API [18:51:07] Phooey... The search API is a turd. [18:51:13] it'll never be accessible because it would likely be less efficient [18:51:14] [bz] (8REOPENED - created by: 2tb, priority: 4Unprioritized - 6enhancement) [Bug 48875] Unable to explain queries on replicated databases - https://bugzilla.wikimedia.org/show_bug.cgi?id=48875 [18:51:15] search api? [18:51:19] who said search api? :) [18:51:41] you just need to do a get request against the API [18:51:45] it should be relatively simple [18:52:25] I'm looking for a specific string in multiple pages and then potentially editing the page if some criterion are met. [18:53:05] Ryan_Lane: What about something to let arbitrary SELECTs? [18:53:07] that kind of query would likely cripple the external store anyway ;) [18:53:11] Dispenser: use the CLI? [18:53:24] Ryan_Lane: https://en.wikipedia.org/wiki/Wikipedia:Bots/Requests_for_approval/HasteurBot_3 [18:53:26] Dispenser: or tunnel an application running on your local system [18:54:00] Hasteur: you really want to talk with Coren|Away about this :) [18:54:06] Ryan_Lane: For non-Labs users so they can update https://commons.wikimedia.org/wiki/User:Dispenser/Wrong_Extension on their own [18:54:18] he'll have better recommendations on how to go about this [18:54:37] The activated category I can use the category API search on, but disabled categories require using the api-search to sneak in the back end and get no more than 50 results. [20:16:40] Ryan_Lane: hi, people are talking and making very constructive comments [20:16:42] http://fa.wikipedia.org/wiki/%D9%88%DB%8C%DA%A9%DB%8C%E2%80%8C%D9%BE%D8%AF%DB%8C%D8%A7:%D9%86%D8%B8%D8%B1%D8%AE%D9%88%D8%A7%D9%87%DB%8C_%D8%A8%D8%B1%D8%A7%DB%8C_%D8%A7%D8%B3%D8%AA%D9%81%D8%A7%D8%AF%D9%87_%D8%A7%D8%B2_%D8%A7%D8%B3%E2%80%8C%D8%A7%D8%B3%E2%80%8C%D8%A7%D9%84_%D8%A8%D8%B1%D8%A7%DB%8C_%D9%88%DB%8C%DA%A9%DB%8C%E2%80%8C%D9%BE%D8%AF%DB%8C%D8%A7%DB%8C_%D9%81%D8%A7%D8%B1%D8%B3%DB%8C [20:16:59] I'll send you a report very soon [20:17:07] Amir1: great. thanks [20:18:46] :) [20:50:55] [bz] (8NEW - created by: 2Liangent, priority: 4Unprioritized - 6normal) [Bug 53689] HTTPS access is not detectable on Tool-Lab webserver - https://bugzilla.wikimedia.org/show_bug.cgi?id=53689