[00:30:55] Ryan_Lane, Coren, I'm thinking about 58997 a bit, what it means to change behavior by region. [00:31:09] I presume the distinction is whether OSM is running on virt0 vs virt1000... [00:31:21] bugzilla? [00:31:29] but in theory OSM can manage instances on multiple regions... [00:31:35] um, yeah, https://bugzilla.wikimedia.org/show_bug.cgi?id=58997 [00:31:38] ah [00:32:02] well, the service groups aren't distinct by region [00:32:11] right. [00:32:12] they should be exactly the same across region [00:32:44] Coren was thinking that we would make a gradual transition, basically have eqiad use the new system and pmtpa the old system, until we finish migrating, then purge the code and remaining ldap stuff for the old system. [00:33:00] how would that work? [00:33:07] this is based on the instances and puppet [00:33:08] That's kind of my question :) [00:33:19] that's not a good idea [00:33:31] Wait, what does puppet have to do with anything? [00:34:52] I think that specifically Coren wanted to implement storage in eqiad using the new model without breaking everything in the meantime. [00:35:52] Hm, hard to discuss this without him here to voice his desires [00:40:03] hm. I guess it doesn't really affect puppet [00:40:42] I think it's fine to have the two kinds of groups coexist, it's just hard to think through when to use which… [00:41:04] Rather than trying to do something clever with datacenters it might be better to ensure that for any given service group both models are implemented. [00:41:13] Then once the new model is fully supported we can just turn off and purge the old model [00:41:17] hm... [00:49:49] I think that's going to cause more problems than it solves [00:50:06] why don't we just switch? [00:50:09] also, on nova-precise2 I have a refactor of public/private dns going [00:50:20] with an abstract class and public and private extending that [00:50:29] it's slightly broken right now [00:50:43] I'll finish it up tomorrow and push it in [00:51:00] anyway, it should simplify things some [00:51:12] gotta run [01:30:35] * Coren is baaa-ack! [01:31:11] andrewbogott: Yeah, you know, that would actually work. Just have both, and ldap.conf picks the right one. [01:31:39] They are actually mutually compatible since the UIDs are already unique. [01:32:14] It's just that this way each group is in two places with two names. [01:32:30] And instances pick the place they like. [01:33:38] This is - by far - the easiest and simplest solution. [01:42:00] This pro tip brought to you by 'Wow, telnet is much better that having to dial in to a terminal server!' [01:53:36] Coren, Ryan was squarely in the "why don't we just switch?" camp. Which I haven't thought about very much… seems like that requires lots of simultaneous patching and debugging. [02:00:47] andrewbogott: Switching will require about five dozen small patches to as many scripts and bits and pieces. None of it is complicated, but the potential to create subtle bugs is high. [02:00:59] yeah [02:02:08] But you are perfectly correct that, after thinking about it for a minute, there's absolutely no reason to pick one or the other. We can simply have both until pmtpa dies with nothing breaking; and then we can patch the things that care as we migrate them. [02:05:52] Yeah, I'm pretty sure that's possible. I'll work on it a bit and see if I come to my senses :) [03:45:53] hi [03:47:38] I have a problem logged [03:48:00] This error comes to me [03:48:06] Disconnected: No supported authentication methods available (server sent: publickey) [03:49:25] Is there one you please help me [03:49:26] .؟ [10:01:10] hi [10:01:20] I work with windows [10:01:49] and i can't run my bot [10:01:56] already [10:02:10] i have an eror message [10:02:23] I do not know why [10:03:02] But it can not read the password of the bot account [10:03:02] Dekel: it really helps if you could tell us the /exact/ error you go [10:03:02] got* [10:03:14] ok [10:03:19] one minute please [10:04:43] WARNING: Running on Windows and transliteration_target is not set. Please see http://www.mediawiki.org/wiki/Manual:Pywikipediabot/Windows Password for user DekelBot on wikivoyage:he: No handlers could be found for logger "pywiki" Logging in to wikivoyage:he as DekelBot via API. Login failed. Wrong password or CAPTCHA answer? API login failed, retrying using standard webpage. Password for user DekelBot on wikivoyage:he: Logging [10:05:35] login failed* [10:07:22] my user-config.py: [10:07:49] # -*- coding: utf-8 -*- family = 'wikivoyage' mylang = 'he' usernames['wikivoyage']['he'] = u'DekelBot' [10:12:29] Jasper, While i'm trying to do simple edits (like add text), I'm getting this message again [10:16:11] what should i do? [10:37:43] ? [10:48:46] ? [11:03:19] I'm not sure why you expect help about windows issues at a labs channel whose lab is running a linux cluster [11:10:17] because thre is no eindows channel [11:10:43] windows* [11:11:35] there is #pywikipediabot [11:12:19] although admittedly people who run the bot could know here, the platform doesn't look spectacularly related [11:13:41] it looks like the warning is related to silly cmd.exe while wrong password is an actual problem [11:15:21] Dekel: sry, I didn't get pinged [11:15:40] I'm a Windows fan and personally I'd avoid Windows and use Linux for this purpose [11:16:18] gry: looks like an incomplete installation to me [11:16:40] possibly [17:20:33] hi [17:25:07] I need help with tools-login.wmflabs.org [17:26:22] I need help with tools-login.wmflabs.org [18:28:31] Hello All! [18:28:47] I have a question regarding DB setup. [18:29:35] In concrete user database. [18:30:07] Anybody? [18:31:39] giftpflanze: Hello! :) Do you (by accident) use user database? Some experience? [18:55:56] DrTrigon: don't ask to ask... ask :p [19:09:50] valhallasw: Is it right that I do have to create user database myself? On every host dewiki, enwiki, etc.? [19:10:01] Coren: Many thanks for intl package :) [19:10:24] DrTrigon: yes [19:10:28] DrTrigon: Yes, provided it fits the correct name (user__%) [19:11:08] ...somehow anoying isn't it? [19:11:20] ... what? [19:11:35] That's by design. [19:11:39] That I have to create the databases myself of EVERY host... [19:12:13] Well, if that's what you application needs, then yes. But then again, what are you doing that needs a DB on every host? [19:12:15] On TS they were "just there" which was quite useful... ;) [19:12:15] Yes, and the tables, and the data. [19:12:27] On the TS, they also were not 'just there' as far as I know. [19:12:34] DrTrigon: No they weren't. [19:12:46] DrTrigon: There is no way to do that with mysql, period. [19:13:20] "By default, once a user is created an empty u_username database is created as well" https://wiki.toolserver.org/view/Database_access#User_databases (but maybe it is outdated...) [19:13:32] It's also not useful, as the databases are not syncronised [19:14:08] I just need it for joining etc. not to store data [19:14:12] DrTrigon: Creating a database by default was just silly. Almost no tools use databases on the replicas at all, and most that do need one database on one cluster. [19:14:38] DrTrigon: Wait what? You don't need a database to join, unless you have data in it to join /with/. [19:15:41] hedonil: That's what they pay me the big^H^H^Hmedium bucks for. :-) [19:15:50] If you would like your database to interact with the replica databases (i.e., if you need to do actual SQL joins with the replicas, which can only be done on the same cluster) https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/Help#Steps_to_create_a_user_database_on_the_replica_servers [19:16:08] Coren: ^^^ [19:16:29] Well yes, but that's it. Joins between your data and the project's. If you don't have data to join against, why would you need to join with your DB? [19:16:58] The keywords in that quote is "your database to interact with the replica". That presumes you have data. [19:17:27] Coren: Thanks anyway, makes me happy ;) btw. is the channel topic still applicable [19:18:20] No. Hasn't been for weeks. Nobody noticed. :-) [19:20:12] Coren: Can you have a look at /data/project/drtrigonbot/pywikibot_drtrigonbot/public_html/cgi-bin/sum_cat_disc.py? Can I doo all the SQL "magic" in there without needing an user database? [19:21:10] (what about the temporary table?) [19:21:44] (sorry should have been "pywikibot-drtrigonbot" NOT "pywikibot_drtrigonbot") [19:21:56] Well yeah, if you use a temporary table then you /are/ storing data in your database! :-) [19:22:19] You're just not keeping it around for a long time. [19:23:49] Coren: So I have to write my script resp. SQL in such a way that it creates needed databases if not there...? How to do this in a smart way? Ideas? ;) [19:24:39] There isn't a 'create or replace' equivalent for databases, but if you 'connect your_db' and catch errors, you can respond to the error by creating it. That's the most robust way I can think of. [19:25:43] CREATE DATABASE IF NOT EXISTS? [19:26:03] https://dev.mysql.com/doc/refman/5.0/en/create-database.html [19:26:10] valhallasw: Does that even exist? Huh. Learn something new every day. :-) [19:26:15] * Coren is more of a postgres dude himself. [19:26:16] Do I have to use the 'CREDENTIALUSER__DBNAME' scheme? [19:26:21] DrTrigon: Yep. [19:26:30] Coren: I'm surprised mysql even throws an error if the database already exists ;-) [19:31:46] Coren and valhallasw: Thank you very much! I will go over this and come back if I still have problems... ;) Greetings!!! [19:48:59] Ping Coren [19:49:24] !b 59084 [19:49:24] https://bugzilla.wikimedia.org/59084 [21:05:38] Coren and valhallasw: Thank you , the DB part works now on tool-labs also! The open issues are related to locale which I will solve another day... (valhallasw - no not unicode this time truly locale I want to convert time-date strings) - Good night! [21:05:55] ;))) [21:06:21] T13|sleeps: Need moar details on what you're trying to test. [21:09:15] I full protected MediaWiki:Common.js and want to see if a TE with editinterface can still edit it. [21:09:58] I'm guessing no, but have been told by others that editinterface might override that protection [21:11:38] So, figured if it was on testwiki I could know/demonstrate for sure. [21:11:52] Wouldn't it be easier to just test this on enwiki? You don't need to make the actual edit -- if the editor is even offered 'edit' rather than just 'view source' it'll tell you what you need to know. [21:12:49] On enwp TE doesn't have editinterface. [21:12:53] But FWIW, I'm pretty sure you are correct in the first place; editinterface protection and edit restrictions are orthogonal to each other. [21:13:02] Oh, right. [21:13:35] The intent is to unbundle editinterface into termplateeditor? [21:13:41] Which is why I need to be able to prove my case it can be done for the RfC to request it. [21:13:43] Because that would be a really, really bad idea. [21:15:12] I still need to get actual statistics, but I believe most changes to MediaWiki: are per TE edit requests anyways and my position is TEs are more technically adept at making such changes. [21:16:54] So, if it's technically possible to allow TEs to edit most MediaWiki: pages and protect the few exceptions like Common.* [21:17:27] I don't see why it would be bad. [21:17:40] I honestly don't know what process we normally go through to change testwiki settings, and adding userrights may have interesting consequences (well, more precisely, removing them afterwards would). I'll have to check. [21:18:40] I did put in a request on bz, so feel free to comment there if I'm not here when you know. :) [21:19:19] T13|sleeps: The general consensus is that editinterface is already too widely distributed by having it bundled with +sysop to begin with; giving it to more editors doesn't sound like a great idea to me. But I'll ask around and see what the usual process for okaying changes to testwiki is. [21:20:54] I'd honestly think it would be better for TEs to have instead of adminstrators.. They are more likely to know what they are doing and research the interface message before applying a change... [21:21:22] But I'm sure I'd be considered biased. [21:21:32] "Instead" would make sense. Good luck getting /that/ to fly. :-) [21:22:27] I did spend quite a bit of time on Mediawiki wiki researching the interface (which was and is still severely lacking documentation). [22:39:09] andrewbogott_afk: https://gerrit.wikimedia.org/r/104320 [23:23:28] Hi [23:23:49] Is it possible to use jsub and pywikibot? [23:23:51] hi [23:28:14] hi Cyberpower678 [23:28:26] pirsquared, hi [23:29:45] Cyberpower678: ever used pywikibot? [23:30:04] When I use jsub, it gives "pywikibot: module not found" or a similar error [23:30:14] but using python2 directly works [23:31:14] it's in the pythonpath... even adding sys.path.append('/shared/pywikipedia/rewrite'); sys.path.append('/shared/pywikipedia/rewrite/externals/httplib2'); sys.path.append('/shared/pywikipedia/rewrite/scripts'); does not work [23:33:56] Betacommand: any idea? [23:33:56] pirsquared, used? yes. Liked it, no. [23:34:21] pirsquared, I'm not your person for that. [23:34:41] It's not for me, it's for someone who refuses to come to this channel [23:35:18] they followed these instructions https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/Help#Using_the_shared_Pywikipedia_files_.28recommended_setup.29 [23:36:54] "used? yes. Liked it, no." :D [23:37:37] giftpflanze: well, do you know about python import paths and stuff? [23:37:57] no, sorry [23:42:08] giftpflanze, another user who refuses to step to that dark side. :D [23:42:28] me? [23:42:30] Ys. [23:42:36] Python. [23:42:41] I [23:42:42] HATE [23:42:46] PYTHON [23:42:48] PERIOD [23:42:55] rofl :) [23:43:01] Why? Just the whitespace? [23:43:39] pirsquared, Hmm? [23:43:45] i would use whitespace anyway [23:44:29] pirsquared, dump the script into the Pywikibot folder and import the names of the needed files. [23:45:14] giftpflanze, what's your language? [23:45:32] tcl *cough* [23:45:56] I use unlambda to write my bots. [23:46:24] i have never heard of that [23:46:36] umm... [23:46:38] very esoteric [23:47:04] extremely. [23:47:09] :P [23:48:31] umm? [23:48:47] looks the the code is not written correctly if you pastebin the imports and text above ill take a look [23:49:06] i have pywiki and jsub working [23:49:33] pirsquared, Betacommand is your obvious choice for Python bot related questions. [23:49:39] Betacommand: the code is not mine, but it is the same as in http://toolserver.org/~mfw/Dellie (replace import wikipedia with import pywikibot) [23:50:31] * Cyberpower678 still admires Betacommand's ability to melt down a server. [23:50:32] yeah, ive been coding python for the 7-8 years [23:53:12] can i get the error traceback? [23:53:39] pirsquared, I'm more of your goto for PHP related stuff. [23:54:49] Betacommand: ImportError: No module named pywikibot [23:55:07] even though it is set up as described on wikitech [23:55:34] thats useless i need the full traceback [23:56:03] pastenin is your friend [23:59:37] Betacommand: http://pastebin.com/Twin6tqT [23:59:59] when I do jstart -blablabla Dellie.py