[07:44:53] Event logging is done on a schema that is a namespace! I agree! [07:44:55] But how do we create a namespace on mediawiki local environment? (That's confusing me) [08:09:32] Hi all. we had an old mediawiki 1.7.1 with a private zone(via Windows AD group to connect). A user says he can't connect even if we put him in the right groups... how can i see from the linux server or mediawiki some info about error or logs please ? [08:10:10] apache access_log or error_log doesn't trace login connection (no trace of my login name for example) [08:40:48] ok found the problem [08:41:38] We use AD groups to guive access and seems that matching a group doesn't works when user is in a group that is itself member of the wiki_ok_goup [08:41:54] like if the test works only for direct group [11:16:13] another question, is it complicated to migrate v1.7 to a recent one ? (i'll use a debian server, if possible a package version) [11:16:31] i remember that some versions are critical [13:11:48] dcpc007, afaik any stable should be ok [13:12:54] I was long time not active in development (actually since GCI ended, because School life), but I remember that newer versions have a migration script [13:13:21] i hope it will not be too complicated, i see that there is a part of wiki protected with login. the apache access/error_log is full of hundred of errors/warning mainly php or missing files [13:13:26] Reedy, hi btw! :) [13:13:37] and maybe a custom theme [13:13:54] dcpc007, which PHP Version? [13:14:39] i look, it's a linux SUSE 11.1 [13:15:32] mediawiki 1.7.1 / PHP 5.2 / Mysql 5.0.26 [13:15:54] some extensions and a hook "userCan: namespacePermissionsCheckNamespace" [13:17:05] hey *smile* and no doc or history on what they choose or how they create it :) [13:17:31] was installed by a user in 2006 [13:18:04] url for connection is : http://server/~user1/wiki/index.php [13:18:09] yeah !!!! roxx [13:18:36] this should be apache user_mod if i remember [13:19:01] something like transforming all folders of user's home as web site :) [13:21:12] Ugh, definitely outdated server software :/ [13:21:37] I recently moved to a new server using Nginx with PHP 7 now [13:22:25] MediaWiki 1.7.1? [13:22:45] That's quite old. [13:23:17] héhé, for mediawiki i'll have via debian 1.27.3-1 (deb 9 or 8) or 1.19.20 (deb 7) [13:23:20] yes :) [13:23:46] that's why i warn to urge migrate it, but hope it will not be too complicated [13:24:06] and config is really not clean [13:24:26] enven while editing the tool bar is incomplete (some icons in error) [13:24:50] See which newer version supports old PHP versions, since I remember doing cleanups in January [13:25:20] So the newest will definitely run only on 5.6 or higher I think [13:25:30] *PHP [13:28:12] i could have php5.4.45 with current "new" template debian 7 [13:28:42] latest stable debian is 7.0+49 [13:30:14] PHP 5.4 is too old :( [13:31:04] https://m.mediawiki.org/wiki/Special:MyLanguage/Manual:Installation_requirements [13:31:16] it for mediawiki 1.19 [13:31:24] debian 7 old now too .... [13:31:43] My old server was running 5.6 on Debian 7 [13:31:44] but it's the standard "new" base i install since 4 years to replace all the older servers [13:31:53] ha ? in stable ? [13:31:56] You need to use an apt repo [13:32:33] https://packages.debian.org/search?suite=default§ion=all&arch=any&searchon=names&keywords=php5 [13:32:39] https://serverfault.com/a/694866 [13:32:41] php 5.6 is jessie, [13:32:47] debian 8 [13:32:53] Do you know how to use apt sources? [13:33:16] yes but i don"t use external non supported repo unless necessary [13:33:22] You just follow the instructions for Wheezy and it's ok [13:33:24] it's not a personnal project :) [13:33:41] but as you see it's necessary, both for security and stability [13:34:07] Or just throw Debian away, their repos are so badly maintained [13:34:13] and external repo break standard of debian, break a lot upgrade process and sometimes are not updated for security patches :-( [13:34:32] Ubuntu on the other hands has far more better maintained repos and also supports PPAs [13:34:38] they are well maintained.. just learn their role ... "stable" is for production env [13:34:47] lol please ... not :) [13:35:00] dppc007, trust me, I was using it on my old server for years, because I was too lazy to reinstall [13:35:04] they install and deform what they want and change thing like the wind [13:35:25] yes, and me i speak of mainting 100+ servers [13:35:34] maintaining [13:35:35] Ubuntu Server? No, it's not as bad as their desktop os [13:36:05] near never had probllem with config and follow on debian stable (or unstable for some desktop) [13:36:41] he only have to use 1-2 years old versions ... but that's why it's called stable ... one or 2 years of tests and validation [13:37:19] and ubuntu-server never look, but i've seen what are doing ubuntu guys [13:37:43] I am using LTS versions only lol [13:37:50] even going until introduce invasive promotional selling link inside your stations [13:38:14] But as I said, no other choices, install Debian 8 or use external repo if you wanna use latest MediaWiki [13:38:24] remove good project only because a commercial one paid ubuntun to remove it [13:38:51] yes, i need to upgrade all edbians but alone ... and have very more older server to do before :(' [13:39:23] i have some deb 8 for specific cases, and starting to create my templates .. but now start again as deb 9 released :) [13:39:46] and debian is not the real problem ... compared to tomcat 5 migration... [13:40:15] and house soft development of 2003-2007 without any doc or sometimes source code ... [13:41:06] (and they don't want to dev from scratch .... too expensive they told ... => 5 years to migrate all servers ... and not finished) [13:41:32] and for choice i could at least take the debian 7 versions, the mediawiki package will be compatible with their php version, no doubt [13:41:48] would be far better with 1.27 than 1.7 :D [13:41:58] and easier to migrate after i think too [13:42:41] ok seems ther is migration instruction for each versions ... i'll have to read all between 1.7 and 1.27 or + [13:42:52] and some hours with a VM to test ) [13:55:41] at least no need to search for work, i have many like this pending [13:56:14] thx for info, i'll try to look the docs, and will see when start install test + migration data/conf if any problem or questions [13:57:15] like if private zone via user login (connected to AD) is still the good way to do [14:03:20] hey I'm after 2 (hopefully easy) plugins if possible... [14:03:25] is it possible to inherit a block from one template to another? [14:03:51] a) a plugin that'll show (on any article) at the top of the page... the categories it is a member of (rather than at the bottom)... maybe just a little one liner - nothing huge [14:04:06] b) a plugin that'll check any article attempting to be saved - and stop you if no category is set [14:04:10] any ideas? [14:05:19] b would be possible with AbuseFilter I'd guess. [14:05:57] ooh thanks will look at that [14:07:03] a) could also be show the categorie(s) that the page is a member of - in the sidebar [15:43:27] Hi, How can I browse all the pictures implemented on my wiki [15:47:00] Special:ListFiles [15:51:18] thanks bawolff [16:06:36] I'm beginning to actively hate mysql databases. They're such an infrequent pain in the rear that you forget how you fixed a thing that came up awhile ago because it was awhile ago. [16:07:58] When you have a new slave in an array and get the session hijacking warning whilst trying to log in, what does that mean [16:08:35] Ulfr: I assume you are storing sessions in db [16:08:52] (e.g. $wgSessionCacheType is CACHE_DB or MainCacheType is CACHE_NONE ) [16:09:00] That's true, yeah. [16:09:24] Either your slave is corrupted, or there is slave lag issues, but MW should be able to handle the slave lag properly [16:09:27] Had an old caching server setup that was causing ghost entries to appear so I said cache me outside [16:09:32] and then didn't go outside. [16:09:38] oh, ok. [16:10:09] Well let's say I had an intermittent issue where edits would be refused because the database had to catch up with itself as well [16:10:13] Is that related? [16:10:58] If slave lag is really high, potentially the sessions (stored in objectcache table) haven't made it to the slave yet [16:12:19] Well it's playing catchup with a backup last night. Hum. [16:17:00] So I'm not super familar with the whole dealing with slave lag code, so I may say something stupid here, but [16:17:21] If its only one slave that's lagged, then mediawiki i believe would use the other slaves [16:17:33] but if you only have 1 slave, then MW will always use it, even if lagged [16:17:57] So you might want to set the weight for that slave server to 0 until the lag on it falls below 30 seconds [16:18:33] pardon my noobishness, but how can I measure what the slave lag is? [16:18:46] because right now nobody can log in and everything is timing out on the master if I run it with no slaves [16:20:50] you can do the query [16:20:52] SHOW SLAVE STATUS; [16:20:55] on the slave [16:22:11] oh, well it's at 0 [16:22:51] yeah, it's pretty much on point atm [16:23:15] Ok, so if you're having issues with sessions its probably some totally other cause [16:25:01] I'd agree with you, but this only comes up when I add the slave [16:25:28] Hm. evidently setting it to CACHE_ANYTHING might have fixed it. The documentation on what that means kinda sucks [16:25:38] https://www.mediawiki.org/wiki/Manual:$wgMainCacheType [16:28:09] Oh wait, looking at SqlObjectCache, if i'm reading this right, CACHE_DB always seems to use the master db, unless i'm missing something, so slave lag shouldn't be an issue [16:28:43] CACHE_ANYTHING means basically use the main cache type [16:28:55] Which is usually apc or memcached [16:33:02] well I haven't set anything up related to caching, but people were able to log in with CACHE_ANYTHING so I have no idea what's going on [16:33:24] turns out someone added a restore timestamp from two years ago to my slave inputs [16:33:28] so I've been restoring very old data [16:34:19] might be part of my problem. [16:34:23] darned interns [16:35:02] Well if the restore deleted/overrided things in the objectcache table, it could cause the problem you are seeing [16:35:39] storing sessions in memcached/apc/etc is generally better than db anyways for performance reasons, assuming your cache is relatively reliable [16:35:54] I should probably set one of those up. [16:36:20] If CACHE_ANYTHING changed anything, then you probably already do [16:36:24] but I pretty much just carte blanche disabled any caching because I was sick of having stuff deleted hours ago coming back and getting me yelled at [16:36:48] as otherwise CACHE_ANYTHING defaults to CACHE_DB [16:36:56] well, that's news to me considering I set up the server templates and whatnot, but if it's something simple to setup it might have snuck in from a cookbook or something [16:37:28] deleted stuff coming back is not good (definitely should not happen) [16:38:09] sometimes if varnish/file cache/squid is misconfigured, it can cause stale data to appear relatively easily [16:38:19] but stale data for memcache/apc is generally rare [16:38:32] It was wonky stuff with the sidebar links, annoying as all getout to troubleshoot. [16:38:40] Oh, yeah I definitely do have varnish setup [16:38:47] * Ulfr has the smarts. [16:39:04] oh hmm, i think there was a sidebar bug from a long time ago [16:39:14] well, do I? Hm. [16:39:42] well, my boss ultimately solved the problem by removing the bloody sidebar [16:39:52] Apc/memcached will probably have more of a performance improvement impact than adding a slave server [16:39:52] navigation isn't so great, but we don't have things we deleted coming back [16:39:53] lol [16:40:27] the problem isn't with the software itself responding slowly, just the amount of traffic we get [16:40:49] If you were still interested in troubleshooting sidebar issues, in all probability, its probably related to localisation cache [16:41:11] Not even remotely. like I said, I was ordered to profane the vector skin and so there's no sidebar. [16:41:30] I'm just hoping that the session cache thing stops crawling up in my business [16:41:31] The take a hammer to it approach :P [16:41:44] the frightening thing is he's a heart surgeon for his day job [16:42:02] and that's his preferred management style. irritating problem? just amputate. [16:42:43] help [16:42:48] Are you using just a single MediaWiki server or multiple? [16:42:55] ilanosortap: how can we help? [16:42:56] one [16:43:04] The server question was for Ulfr [16:43:09] I've got a single english language wiki with a pair of frontends and a master/slave setup [16:43:16] Hi! [16:43:22] i want to work on huggle [16:44:19] we used to have separate installs for different languages and a small amount of content, but there wasn't a lot of administration for them so it's mostly just advertisements and spam. Kinda just said buh-bye to it all [16:44:23] Ulfr: Because, if you just have a single php server (since apc is only a good caching solution if you have one php server, otherwise you need memcached), you probably already have apc installed, so all you'd have to do is set $wgMainCacheType = CACHE_ACCEL; [16:44:44] No, I've got two. One gets flattened in a DDoS too easily :( [16:45:15] ilanosortap: This is not really the right channel. Huggle has their own irc channel #huggle [16:45:36] ilanosortap: If you're not familar wiht irc, you can join the huggle channel by typing [16:45:40] /j #huggle [16:46:43] yeah, no varnish install on these new php7 frontends [16:46:48] it's fast enough without it I think [16:47:18] images are shared with s3fs [16:47:20] How much varnish can help depends a lot on the breakdown between logged in and logged out users [16:47:48] well it was a lot slower with 32-bit unbuntu 10.04 and php5.5 [16:48:02] so I added varnish and it was fine [16:48:12] yeah, php7 gives a big speed increase [16:48:40] It's glorious. I'm hoping I can just clear the stupid old timestamp and get the actual latest database snapshot and let people log in that way [16:48:59] parsoid is enough of a pain to remember how to set up on top of everything else, no need to have caching I don't need to boot [16:49:43] yeah, I've never even tried to set up parsoid [16:50:14] it's not that bad lately, but figuring it out was a pain. really like visual editor though so it's worth it [16:50:31] I'd still reccomend trying out memcached though. It will reduce the load on your db servers quite a bit (probably) [16:50:38] the irony here is this whole situation was caused because my management server plum forgot about one of the pair of database servers [16:51:01] apparently replication the way it's set up won't work unless both servers are managed [16:51:21] MW didn't care that master and slave weren't talking though [16:51:59] MW should care. Its supposed to go into lagged slave mode (show a bunch of warnings all over the place) if the lag exceeds 30 seconds [16:53:09] Oh trust me it does [16:53:57] that comes up fairly frequently at peak usage [16:55:55] but the crappy event handling kinda turned everything from fixable to hot mess very very quickly [16:56:08] * Ulfr learned his lesson about letting scripts change server roles [21:12:25] Example If I want to translate the page https://www.mediawiki.org/wiki/Manual:Hooks/SkinBuildCodeBar but I need to create the same page with the "URL prefix" /ru at the end? [21:56:32] How come i cant edit T176055 ? [21:56:33] T176055: Update of QueryPages failing on commons with "MASTER_POS_WAIT() or MASTER_GTID_WAIT() failed: MySQL server has gone away" - https://phabricator.wikimedia.org/T176055 [22:14:15] hello friends [22:15:57] Hi [22:17:14] how are you [22:17:23] Im fine [22:17:43] Salaheddine 20 years from morroco [22:19:01] you