[00:00:41] petan: I hope I'm getting these right... there wasn't any info on the PoolCounter one [00:00:56] there could be more such [00:01:05] because it wasn't included [00:02:40] !log deployment-prep [auth] added bugzilla to interwiki.sql and ran it, doesn't seem to be working [00:02:41] Logged the message, Master [00:02:54] !log deployment-prep booting memcache in case that fixes interwiki [00:02:55] Logged the message, Master [00:03:52] don't forget memcached breaks sessions [00:04:09] http://deployment.wmflabs.org/commons_wiki/wiki/Special:RecentChanges [00:04:12] same as on enwiki [00:04:17] !log deployment-prep booting memcached fixed it [00:04:18] Logged the message, Master [00:04:24] petan: you sure? i'm still logged in [00:04:33] it logged me off [00:04:50] i'm very surely logged in... [00:04:54] and broken my import [00:05:02] :( [00:05:06] :( [00:06:19] http://deployment.wmflabs.org/commons_wiki/wiki/Special:Preferences#mw-prefsection-gadgets [00:06:20] yay [00:06:21] so many [00:08:20] petan: could you create an images/timeline dir when creating wikis in the future (asking in case you're scripting it, I'm creating it now) [00:08:33] ok [00:08:35] I will [00:11:02] !log deployment-prep "apt-get install ploticus ttf-freefont" for timeline [00:11:03] Logged the message, Master [00:11:23] !log deployment-prep set config for timeline [00:11:24] Logged the message, Master [00:15:35] petan: have you set up hiwiki yet? [00:15:52] hm... probably not [00:15:58] I will do that tommorow ok? [00:16:01] it's 1 am [00:16:11] :o [00:16:18] sure, go to bed... unless you want to point me to how [00:16:29] or if you import data I can set it up it's not so hard [00:17:01] I will make a guide on auth [00:17:04] I was mostly trying to figure out how you did db set up for now [00:17:13] it's very simple [00:17:22] there is one user mw who has access to all db's [00:17:28] k, go to bed for now unless this is gonna take you 5min ;) [00:17:38] I always create a new db and set up permissions for it [00:17:47] then I copy preconfigured mw to new folder and start installation [00:17:59] then I copy LocalSettings and check extensions I need [00:18:05] you were creating some scripts for set up? where? [00:18:11] I am going to do that [00:18:17] for now I just do cp and cp [00:18:18] :D [00:18:24] k thought you had already [00:18:25] it's about 2 copies [00:18:40] I made a script for extensions deployment from svn [00:18:45] but I need to improve it [00:19:18] oh [00:19:22] hi_wikipedia exist [00:20:51] k, just didn't see dbname for ti [00:21:03] hi_wikipedia [00:21:04] and got lots of errors frm the cli [00:21:14] trying again [00:21:18] http://deployment.wmflabs.org/hi_wikipedia/wiki/Main_Page [00:21:56] ah [00:22:07] was missing a symlink for extensions [00:23:26] yes [00:23:28] running update [00:24:38] hrm... [00:24:41] works [00:24:45] back to interwiki [00:26:42] works [00:26:49] ty jeremyb [08:28:57] PROBLEM Current Users is now: CRITICAL on nova-dev2 nova-dev2 output: CHECK_NRPE: Socket timeout after 10 seconds. [08:33:47] RECOVERY Current Users is now: OK on nova-dev2 nova-dev2 output: USERS OK - 0 users currently logged in [12:17:55] !log deployment-prep restarted apache to fix some problem with config [12:17:55] Logged the message, Master [12:45:50] yay [12:45:55] jeremyb: check out new scripts [14:04:50] !log deployment-prep created temporary memcached instance for data [14:04:51] Logged the message, Master [14:13:54] PROBLEM Current Load is now: CRITICAL on deployment-nfs-memc deployment-nfs-memc output: Connection refused by host [14:14:39] PROBLEM Current Users is now: CRITICAL on deployment-nfs-memc deployment-nfs-memc output: Connection refused or timed out [14:18:54] PROBLEM host: deployment-nfs-memc is DOWN address: deployment-nfs-memc CRITICAL - Host Unreachable (deployment-nfs-memc) [14:49:02] PROBLEM host: deployment-nfs-memc is DOWN address: deployment-nfs-memc CRITICAL - Host Unreachable (deployment-nfs-memc) [15:09:02] RECOVERY host: deployment-nfs-memc is UP address: deployment-nfs-memc PING OK - Packet loss = 0%, RTA = 0.84 ms [15:13:22] PROBLEM dpkg-check is now: CRITICAL on deployment-nfs-memc deployment-nfs-memc output: CHECK_NRPE: Socket timeout after 10 seconds. [15:13:32] PROBLEM Disk Space is now: CRITICAL on deployment-nfs-memc deployment-nfs-memc output: CHECK_NRPE: Socket timeout after 10 seconds. [15:13:42] PROBLEM Total Processes is now: CRITICAL on deployment-nfs-memc deployment-nfs-memc output: CHECK_NRPE: Socket timeout after 10 seconds. [15:14:02] PROBLEM Free ram is now: CRITICAL on deployment-nfs-memc deployment-nfs-memc output: CHECK_NRPE: Socket timeout after 10 seconds. [15:18:52] PROBLEM host: deployment-nfs-memc is DOWN address: deployment-nfs-memc CRITICAL - Host Unreachable (deployment-nfs-memc) [15:23:53] RECOVERY host: deployment-nfs-memc is UP address: deployment-nfs-memc PING OK - Packet loss = 0%, RTA = 1.67 ms [16:07:53] RECOVERY Current Load is now: OK on deployment-nfs-memc deployment-nfs-memc output: OK - load average: 1.20, 1.10, 0.74 [16:08:13] RECOVERY dpkg-check is now: OK on deployment-nfs-memc deployment-nfs-memc output: All packages OK [16:08:23] RECOVERY Disk Space is now: OK on deployment-nfs-memc deployment-nfs-memc output: DISK OK [16:08:33] RECOVERY Total Processes is now: OK on deployment-nfs-memc deployment-nfs-memc output: PROCS OK: 95 processes [16:08:53] RECOVERY Current Users is now: OK on deployment-nfs-memc deployment-nfs-memc output: USERS OK - 1 users currently logged in [16:08:53] RECOVERY Free ram is now: OK on deployment-nfs-memc deployment-nfs-memc output: OK: 84% free memory [16:28:03] * Damianz gives labs-morebots a muffin [16:45:14] !log deployment-prep moved memcached to deployment-nfs-memc and update global [16:45:15] Logged the message, Master [18:47:12] ok [18:47:26] relative paths can't be used for command line [18:47:35] because you would need to start maint scripts from the given path [18:47:43] hexmode: ^ [18:47:45] :o [18:47:46] !log deployment-prep set up relative paths for config files [18:47:47] Logged the message, Master [18:47:51] I will fix that [18:48:02] petan: ok, so which full path? [18:48:10] yes [18:48:12] full path [18:49:29] petan: yeah, I just meant "which one" since /var and /mnt work [18:49:47] I don't think it matters, I started /mnt [18:50:04] k [18:56:29] done [18:59:14] hexmode: I don't know if this is a good solution we now have to also update all paths in LocalSettings [18:59:26] because it's now in another folder [18:59:46] no... why would you have to do that? [19:00:02] it's looking for "local.php" [19:00:12] but there is no local.php in /mnt/www/global/LocalSettings/ [19:00:28] but it shouldn't be looking in /mnt/www/global [19:00:29] it would either have to be relative path from /mnt/www/global/LocalSettings/ or absolute path [19:00:34] where is it doing that? [19:00:37] it does because it's cli [19:00:46] all maintenance scripts fail [19:00:46] I ran update.php and it worked, though [19:00:52] it didn't to me [19:00:58] where did you start it from [19:01:04] what command are you using? [19:01:23] petrb@deployment-dbdump:/mnt/www$ global/maintenance/updatesites [19:01:37] PHP Warning: include(local.php): failed to open stream: No such file or directory in /mnt/www/global/LocalSettings/auth.php [19:01:43] let me try that to see what you see [19:01:56] just start the same cmd [19:02:29] cli is following symlinks so it's making relative path from target path of symlink [19:02:40] it shouldn't be... 1s [19:03:53] does it work to you? [19:04:38] k, something is wrong here... it works when I run update.php in the directory [19:04:59] try to run update.php from a different [19:05:43] I think it makes a lot of sense... [19:06:02] you start update.php and it open LocalSettings.php and start looking for config.php from current path [19:06:17] if there was absolute path it would work [19:07:09] try now [19:07:27] what did you change? [19:07:47] script changes dir before running maint [19:07:53] hm... [19:07:56] worked like a charm ;) [19:08:06] it probably will work, but it's a bit... unstable [19:08:07] :D [19:08:16] I suggest to remove local symlink [19:08:21] and put absolute path to LS [19:08:35] I think it's same complicated [19:08:51] either having a symlink with absolute path or absolute path in include is same [19:08:56] relative symlink shouldn't have caused it [19:09:07] but relative symlink didn't work at all [19:09:09] for cli [19:09:19] try it, change one wiki to relative [19:09:23] you will see [19:09:32] :| [19:09:36] it does now... 1s I have an idea [19:10:33] enwiki still has relative symlinks [19:10:49] and this solution seems to be working [19:15:00] http://deployment.wmflabs.org/en_wiktionary/wiki/Main_Page [19:15:04] why there is no skin? [19:17:13] hrm... [19:17:19] wth [19:19:37] all newly installed wiki's don't have skin [19:20:04] wonder if it is related to some jquery stuff that stopped working [19:20:25] it's really weird because I didn't change anything related to this [19:21:03] http://deployment.wmflabs.org/auth/wiki/Problem_reports -- bug was being updated via jsonp yesterday [19:21:28] petan: you're still using the same createsite script, right? [19:21:33] yes [19:21:44] but I created some other sites using that today [19:21:51] k [19:22:04] just going to look at that to see how you are creating them [19:23:19] OH [19:23:19] yes [19:25:34] yeah... I'm beginning to wonder about the latest revs in svn [19:25:52] I forgot to update global [19:29:23] how do you initialize the db? [19:29:41] installer [19:29:46] I do it manually for now [19:29:55] you know we have a cli installer now? [19:30:00] I don't know [19:30:02] * hexmode adds it [19:30:07] ah [19:30:18] had to let you know ... I did a lot of the work ;) [19:30:54] heh [19:30:57] not sure how much it is going to help us here, but maybe a little [19:31:33] but what in the worls is causing this [19:32:58] not sure, but I'm looking [19:33:06] let me know if you see something [19:33:14] !log deployment-prep clearing cache [19:33:15] Logged the message, Master [19:33:25] was going to test setting up a wiki and see if that did it. [19:34:07] I think it's a problem with global I did [19:34:18] I forgot to update some settings and it remained cached somehow [19:43:28] k, my testwiki doesn't have a /wiki [19:43:37] that's weird... [19:43:43] how do I get /wiki and /w ? [19:43:49] ./* No modules requested. Max made me put this here */ [19:43:54] that;s what I get from resource loader [19:44:02] same command doesn't work on that wiki [19:44:06] but on other wiki it does [19:44:21] hexmode: /etc/apache/htt* [19:44:21] URL? [19:44:23] see it [19:44:26] http://deployment.wmflabs.org/auth/w/load.php?debug=true&lang=en&modules=sit&only=styles&skin=vector&* [19:44:26] k [19:44:44] just thought you might have that set up :) [19:45:25] oh yes [19:45:27] it works... [19:45:30] that's really weird [19:46:30] k, now I see how you're setting them up [19:47:47] nvm [19:47:54] I found it [19:48:25] https://en.wikipedia.org/w/load.php?debug=true&lang=en&modules=sit&skin=vector&* [19:49:32] http://deployment.wmflabs.org/auth/w/load.php?debug=true&lang=en&modules=sit&skin=vector&* [19:49:34] same [19:50:27] is it b/c you have the aliases set up wrong? [19:50:49] I fixed it [19:50:49] yeah, they're fixed now [19:50:51] k [19:51:00] weird [19:51:28] nah scriptpath was set in localsettings [19:51:40] and the new ones didn't have /wiki [19:51:53] ? [19:51:57] that was a manual part of the setup, right? [19:52:03] setting up the aliases [19:52:05] I don't know, probably yes [19:52:06] yes [19:54:04] there is some problem with load.php [19:54:09] it doesn't work [19:55:13] that's weird [19:55:20] http://deployment.wmflabs.org/de_wikipedia/w/load.php?debug=false&lang=en&modules=mediawiki.legacy.commonPrint%2Cshared|skins.vector&only=styles&skin=vector&* [19:55:28] this is displaying the index.php [19:55:30] o.O [19:55:33] hexmode [19:56:01] nowit works [20:05:12] 01/08/2012 - 20:05:11 - Creating a home directory for johnduhart at /export/home/deployment-prep/johnduhart [20:06:12] 01/08/2012 - 20:06:12 - Updating keys for johnduhart [20:39:10] hm [20:39:26] hm [20:39:26] !log deployment-prep created new group for /mnt/www [20:39:27] Logged the message, Master [20:41:08] petan: Group name [20:41:13] ? [20:41:19] depops [20:41:22] ops is taken [20:41:32] by real ops [20:41:47] ah [20:42:09] Do I just use usermod on the server or is this special LDAP magic? [20:42:34] hm... probably yes [20:42:59] That wasn't really a yes or no question ;) [20:44:52] I added you [20:45:11] Just for reference, how? [20:45:26] by hand [20:45:29] ./etc/group [20:45:30] :P [20:45:36] that's only working way afaik [20:45:38] ah [20:47:23] Still dont' have permissions [20:47:25] permissions set [20:48:21] that's weird [20:48:46] Do permissions not carry over NFS? [20:49:09] that's also fixed but it doesn't seem to work on you [20:49:24] all people are in that group excepting you try to log off and back [20:49:29] I couldn't put you even to git [20:49:37] that's weird [20:50:20] were you trying any other commands before? [20:50:46] Like? [20:51:07] johnduhart@deployment-web:/var/www$ ls -l [20:51:07] ls: cannot open directory .: Permission denied [20:51:17] see, dunno who did it [20:51:18] have you relogged in? [20:51:21] fixed [20:51:25] Platonides: yes [20:51:58] Still can't write though [20:52:02] johnduhart: no idea, let's set it a+w [20:52:06] and ask Ryan [20:52:10] hold on [20:52:16] it's a cosmetic issue and it doesn't matter at all [20:52:30] because no one else than people who are supposed to write there will have shell access [20:53:01] Fine, just a+w it for now [20:53:20] done [20:53:36] let's concentrate on something what really matters [20:53:49] you wanted to change the configs? [20:53:51] * johnduhart grumbles about permissions really mattering to him [20:53:54] Yeah [20:54:12] all people who will ever have access there will be either root or will not have accesss [20:55:07] So first we need a wildcard on *.deployment.wmflabs.org [20:55:43] Ryan is offline now [20:55:50] so I guess it will need to wait [20:56:14] * johnduhart checks for other ops [20:58:35] ok [20:58:41] None at the moment [20:58:48] actually it would be cool [20:58:53] to have that [20:59:03] I gave you some global flags on SUL [20:59:30] ooof, looks like I need to go somewhere. I'll be back soon, if you see ryan ask him [20:59:33] and thanks [21:04:46] Reedy: what is svn path [21:04:54] to 1.19 ex [21:04:54] ...? [21:05:11] /truk/extensions [21:05:18] /trunk/extensions [21:05:20] so head? [21:05:23] Yup [21:05:25] not branched yet [21:05:32] right... [21:19:57] hexmode: done [21:19:58] it's 1.19 [21:20:33] :) [21:21:13] do you know api to get MW space pages? [21:21:14] all [21:21:24] it would really speed imports up [21:22:13] petan: did you look at api.php? [21:22:22] sure but... I am lazy :d [21:22:27] I hoped someone know it [21:22:38] could ask in #mediawiki [21:22:41] ;) [21:22:51] heh [21:25:14] hexmode: I will need to go, try to explain to john that we wanted to have the site working tommorow so maybe it's not best to perform some hard core changes today [21:25:43] actually I like his ideas, but I don't believe we can set it up today [21:26:11] if we start moving domain... :\ [21:33:04] OrenBochman: hi [21:33:11] check https://en.wikipedia.org/wiki/User_talk:Petrb#Two_ideas_for_developers [21:33:21] there is an idea to improve search [21:35:32] petan: johnduhart I think has a good idea but I'll let him know what we're aiming for [21:35:44] yes it's good idea [21:35:46] I like it [21:35:57] but problem is we don't have a time to set it up today :) [21:36:05] Ryan is not here [21:36:06] it's sunday [21:36:18] I doubt he want to mess up with labs config today [21:36:25] sure [21:36:59] johnduhart: are you getting this? we want to make let people know they can test tomorrow [21:37:08] s/make// [21:37:11] I think tommorow is a good day to perform some first bigger tests on simple clone and english clone, then we could place a central notice that we want to perform maintenace on, lets say tuesday and do it that day [21:37:15] there is central notice :) [21:37:17] on all wikis [21:37:34] oh [21:37:42] I need to point to the test plans [21:37:46] ok [21:37:55] in case anyone wants to try them [21:37:55] I don't know what tests plan is [21:38:03] or improve them [21:38:18] https://www.mediawiki.org/wiki/Category:MediaWiki_test_plans [21:38:19] hm... [21:38:41] there are no plans for this [21:38:42] version [21:38:52] going to put that on the problem reports page and the fp [21:38:52] 1.19 [21:38:56] ok [21:39:07] no... could use the 1.18 ones [21:39:09] aha [21:39:20] https://www.mediawiki.org/wiki/User:Sumanah/1.18_test_cases [21:40:05] ? [21:40:23] * johnduhart is back from driving [21:41:22] :o [21:41:24] johnduhart: bottom line is: we want to have deployment working tomorrow [21:41:29] that is all [21:41:42] so don't screw it up ;) [21:41:49] johnduhart: if you believe you can move domain today go ahead :) [21:41:54] petan: Can you just spin me up a new web instance so I can work on the new configuration the next few days? It'll be outside of NFS and it won't affect the current webs [21:41:57] I support it [21:42:00] only do things that make it better [21:42:07] sure [21:42:18] * jeremyb is kinda here a little [21:42:48] Thanks, that way I can be more radical in configration changes [21:42:55] johnduhart: it won't be accessible from outside [21:43:09] Radical as in not worrying about breaking the current setup [21:43:16] petan: Sure [21:43:18] feel free to use db... but I suggest not to use simple wiki [21:43:27] I'll stay away from that [21:43:30] ok [21:43:39] but you should definitely use that central sql server [21:43:46] otherwise Ryan will get angry [21:43:46] petan:ypu [21:43:47] :D [21:43:48] yup * [21:44:07] Do you think we should use shorter subdomains instead of the full name? eh. en.wp. isntead of en.wikipedia? [21:44:15] no [21:44:17] full [21:44:20] Sure [21:44:23] maybe make a redirect from short [21:44:38] Until Ryan gets us DNS I'll just play with my hosts file [21:44:40] jeremyb: is there anything you can do that would help? [21:44:48] maybe stand in for ryan? [21:45:04] jeremyb: create my api :D [21:45:13] petan: Deployment-test? [21:45:15] jeremyb: I need to make a list of all MW pages on de wiki [21:45:17] johnduhart: yes [21:45:28] yay [21:46:32] * jeremyb reads [21:47:25] hexmode: how so? [21:47:33] jeremyb: make a list :D [21:47:42] that is very helpfull heh [21:47:55] I need to import mw space of de and en wikinews and wikitionary [21:48:06] so I need to make a list of all pages there [21:48:34] also if you could make all wikis more cute [21:48:35] :) [21:48:43] * Damianz applys kittens [21:48:46] like create some better main page [21:48:47] :D [21:48:51] jeremyb: we're setting up deployment.wmflabs.org for some (hopefully) widespread testing tomorrow [21:48:53] Damianz: that's good idea [21:48:53] * johnduhart waits for deployment-test to respond [21:49:04] johnduhart: labs are really broken today [21:49:09] hexmode: i got that much. what does stand in for ryan mean? [21:49:12] I needed to poke apergos to fix brewster [21:49:19] he didn't even know that's a puppet repo [21:49:32] Poke ryan with a stick! [21:49:33] apergos has root everywhere but doesn't know what is running there :D [21:49:43] jeremyb: petan was saying ryan wasn't around and I thought maybe you could help... I don't know what petan needs, though [21:49:44] Though for one it didn't break cb [21:49:58] petan: that what's a puppet repo? brewster is dhcpd i think? [21:50:04] that's what he said too [21:50:15] but for some reason instances are downloading stuff from it [21:50:30] so when its http doesn't work, puppet doesn't work too [21:50:43] huh... oh, that makes sense. not puppet but apt.wikimedia.org (and maybe ubuntu.wikimedia.org?) [21:50:56] hm, yes [21:50:58] those are straight http [21:51:03] indeed [21:52:05] hexmode: fwiw, i have no more access than petan does. but i do have some knowledge about what runs where. [21:52:16] * hexmode annoys dewiki with an english language notice [21:52:26] haha, about what? [21:52:29] we don't need knowledge we need powers! [21:52:34] <^demon> poor dewiki :( [21:52:35] jeremyb: k... now I know! [21:52:46] dewiki is ok [21:52:50] no problem with them [21:52:54] only on their VP page [21:52:58] always I talked to them it was ok [21:53:08] they sent me english welcome on dewiki message [21:53:09] :D [21:53:11] so not a central notice [21:53:17] :) [21:53:58] johnduhart: once you finish new config we put a central notice so that people know there will be maintenance ok? [21:54:13] Sure! [21:54:20] Once I get an instace :> [21:54:21] I don't like idea that it would get down in a middle of someone's test [21:54:26] of course [21:54:29] working on that [21:54:35] ty [21:54:42] Jan 8 21:53:48 i-000000d8 dhclient: bound to 10.4.0.17 -- renewal in 54 seconds. [21:55:00] ? [21:55:13] PROBLEM host: deployment-test is DOWN address: deployment-test PING CRITICAL - Packet loss = 100% [21:55:43] heh [21:55:47] thanks nagios [21:56:01] I tried it again [21:56:19] my nagios parser is faster than nova [21:56:24] :D [21:56:53] maybe I should make bigger timeout for reload [21:57:05] so that Ryan has more time to get his instances working [21:57:14] :o [22:01:11] hi Ryan_Lane [22:01:19] labs are broken :( [22:01:31] I had to create one instance 10 times before it booted up [22:01:34] hey Ryan_Lane [22:02:00] oh? [22:02:06] well, that's not good [22:02:10] * jeremyb now sees prod SAL [22:02:11] 14:54 apergos: removed old puppet lockfile on brewster, ran by hand [22:02:11] 14:47 apergos: cleared out some very large squid logs on brewster, (basically all of them) plus lighty logs, disk was full. restarted squid manually [22:02:20] ah. brewster [22:02:26] that would cause a problem, yeah [22:02:28] that's too [22:02:29] btw, can someone kick ntpd on wikitech? pleaseeeeee [22:02:52] Ryan_Lane: I can't create instance :| [22:02:57] petan: I looked it over [22:03:00] no? [22:03:05] it everytime stuck on assigning ip [22:03:05] gimme a sec [22:03:12] it has assigned ip but no one can ssh to it [22:03:19] hm [22:03:32] I tried rebooting and even recreating it [22:03:37] Ryan_Lane: firmly planted in SF now? :) [22:03:40] I need to recreate it many times before it works [22:04:06] jeremyb: yep [22:04:10] petan: I'm investigating [22:04:29] https://labsconsole.wikimedia.org/w/index.php?title=Special:NovaInstance&action=consoleoutput&project=deployment-prep&instanceid=i-000000d9 [22:05:17] looks like it worked to me [22:05:22] ah [22:05:25] petan: I know that some wiki's will inform all editor that the page is being edited for a duration once an edit is started. it would be cooler if on each key click the editor would update this timer [22:05:26] yes I can ssh to ip [22:05:29] but not to host [22:05:34] to host? [22:05:35] johnduhart: it works [22:05:35] oh [22:05:44] petan: did you delete and re-create the same instance? [22:05:46] yay [22:05:47] yes [22:05:53] there's an issue with DNS when that's done [22:05:56] ah [22:06:16] the DNS name of the instance gets cached, and the IP address is wrong [22:06:20] johnduhart: you can even mount nfs somewhere can you do that? [22:06:31] or should I [22:06:59] If you could mount NFS that'd be nice, however I'm going to work locally for right now until we're ready ot put it live [22:07:00] Ryan_Lane: we wanted to use dns like en.wikipedia.deployment [22:07:09] johnduhart: no SAN. all compute nodes share and mount their instance storage to each other [22:07:13] or maybe even better would be en.wikipedia.wmflabs [22:07:22] you can do en.wikipedia.wmflabs.org [22:07:25] Ryan_Lane: hm [22:07:28] regarding section - they are not real object as war as MediaWiki is concerned - so what links here reffers to the page not the section - how would it know is the user has scrolled away from the anchor in the uri [22:07:32] actually, you can do better [22:07:35] Ryan_Lane: Yeah, a wildcard DNS entry would be nice [22:07:35] Ryan_Lane: we need all wmf domains like subdomains [22:07:37] Ryan_Lane: Oh? [22:07:45] we can do wildcards [22:07:49] yay [22:07:56] I need to fix something, though [22:07:59] :o [22:08:01] OrenBochman: yes I know [22:08:11] OrenBochman: can you explain it on talk page? [22:08:13] I don't need it right now, I can just toy with my hosts file [22:08:15] sure [22:08:18] thanks! [22:08:27] The test instance doesn't have a public IP anyway [22:08:43] johnduhart: rather than a SAN we use glusterfs since it's distributed, and will scale with the number of nodes we have [22:08:50] actually if we could configure it right now... it would be cool but dunno how hard is it [22:09:07] Ryan_Lane: That's pretty cool [22:09:07] yeh gluster would be cool [22:09:15] Ryan_Lane: that's a temporary nfs [22:09:25] petan: so, yeah, the real issue was brewster dying [22:09:27] I set it because we don't have anythin else yet [22:09:28] petan: Well I didn't want to steal another public IP but if it's no issue :o [22:09:41] petan: yeah, I'm just talking about instance storage [22:09:42] johnduhart: I meant to set it up on -web [22:09:46] ok [22:09:47] I think the volume storage has been ordered [22:09:48] great [22:09:59] IO really suck on VM's [22:10:04] yeah [22:10:07] cpu and ram is cool [22:10:20] I am using nohup on cp :D [22:10:20] yeah, anything heavy IO is going to suck in instances [22:10:23] petan: Well before we do those domains I want to have my new configuration done since, which will be based on the structure [22:10:40] ah... [22:10:43] right [22:10:57] Ryan_Lane: can we register all wmf domains like wikinews.wmflabs [22:11:01] for our project [22:11:21] so we would be able to clone all wikis to scheme lang.project.wmflabs.org [22:11:29] like en.wiktionary.wmflabs.org [22:11:34] Nah, lets not do that [22:11:38] why not? [22:11:38] toollabs will need those [22:11:45] hm... [22:11:53] Er, testlabs [22:11:56] but en.wikipedia.deployment.wmflabs.org is crazy :D [22:12:01] well, what you guys are doing is going into testlabs [22:12:06] it seems to me [22:12:08] hm... [22:12:12] maybe not, though [22:12:17] depends [22:12:24] but it seems like it should [22:12:27] maybe we should merge our project with testlabs? :P [22:12:39] Ryan_Lane: No directly, I hope... [22:12:40] at some point it would make sense [22:12:41] you guys are doing the apache/mediawiki/db part of it right now [22:12:44] * johnduhart looks at file permissions [22:12:46] :P [22:12:54] that's something we can fix over time [22:13:12] actually I already started working on a script to set up a wiki in few commands I think it's in your proposal too [22:13:18] \o/ [22:13:29] petan: yes but that's not a true wikifarm [22:13:31] I'm really glad you guys are doing this. I haven't had time to get to it [22:13:33] Like in production [22:13:43] let's make it one over time [22:13:56] johnduhart: how would you make it? [22:13:58] we need to make some changes in production to make this work correctly [22:14:08] Ryan_Lane: like what? [22:14:08] Testlabs need to be a mirror of production, that's the whole point. This project is for testing new releases [22:14:32] yes but this project also wanted to be a mirror of production but with latest sw [22:14:35] the infrastructure is the same, though [22:14:48] only difference is sw version [22:14:48] hm [22:14:57] maybe a different URL scheme is a good idea [22:14:59] otherwise idea is very same [22:15:18] actually I think we should merge both projects [22:15:33] well, how would you specify version when hitting en.wikipedia.wmflabs.org? [22:15:35] so that we could use one big cluster with deployment version and production version of sw [22:15:54] en.wikipedia.wmflabs.org/deployment/wiki [22:15:58] en.wikipedia.wmflabs.org/wiki [22:16:21] that could work, but it would be different from how production is configured [22:16:28] probably [22:16:30] would it be problem? [22:16:33] likely better to do it in the url [22:16:37] yes [22:16:38] err [22:16:40] in the domain name [22:16:41] think of the cookies please [22:16:49] yeah, that actually helps the cookies :) [22:16:50] I'd prefer them to have the same number of components in the path [22:16:57] +1 [22:17:05] domain name is likely best [22:17:06] ok [22:17:09] haha, think of the cookies [22:17:17] what domain name would you guys like? [22:17:33] *.deploy [22:17:42] maybe just d [22:17:47] en.wikipedia.d.wmflabs [22:17:48] That would work [22:17:53] well, it's going to be a shitload of domains :) [22:18:06] do you guys have an IP yet? [22:18:08] why not just enwiki.wmflabs instead of making en.wikipedia ? [22:18:09] Ryan_Lane: wildcard [22:18:09] yes [22:18:13] petan: no [22:18:15] Ryan_Lane: deployment.wmflabs [22:18:19] that's waht we have now [22:18:22] johnduhart: huh? [22:18:26] We're not doing this to that right now [22:18:32] what you mean? [22:18:38] ah [22:18:38] I need to put together the new config before we point the new domains at it [22:18:41] sure [22:18:52] of course I meant we will move deployment to it later? [22:18:58] yeah [22:19:23] Platonides: then the apache rules will be harder [22:19:24] btw Ryan_Lane I still think we could merge it using en.wikipedia.d.wmflabs and en.wikipedia.wmflabs [22:19:33] dabatabases could be shared [22:19:34] petan: agreed [22:19:42] * data :D [22:19:48] databases can't be shared... [22:19:49] Ryan_Lane, the domain would match db names [22:19:53] hm... ok [22:19:58] the software version would be differnt [22:20:00] so at least logins to wikis could be shared :D [22:20:11] Could I get an IP that points to deployment-test and have *.d point to that, so I work on the new setup? [22:20:20] Platonides: it's best to keep it as close to production as possibl [22:20:22] sure [22:20:22] *possible [22:20:30] they can _mostly_ be shared, but sometimes a schema change on one side would break the another version [22:20:33] Ryan_Lane: can you tell me how to register wildcard? [22:20:41] and give us one ip [22:20:53] petan: I don't think we can do *.*.d.wmflabs.org [22:20:59] ah... [22:21:08] Ryan_Lane: Just *. should work [22:21:08] but we can register more domains? [22:21:12] to same ip [22:21:18] I can. you guys can't. [22:21:19] I've done that before [22:21:52] I'll just create all the domains [22:22:01] ok [22:22:03] ok [22:22:07] wait [22:22:19] would it be possible to move them all later to our current instance? [22:22:24] yes [22:22:27] ok [22:22:38] the wildcards get added to a floating IP [22:22:44] great [22:22:45] then you can move the floating IP at will [22:22:56] Yup [22:22:59] sounds ok [22:23:11] heh. this is going to be a ton of domains [22:23:18] Ryan_Lane: Are you really sure you can't just do one wildcard? [22:23:31] you can do *.foo [22:23:33] how would that work? [22:23:37] but *.*.foo... [22:23:41] right [22:23:44] done [22:23:56] Reedy: I just *.local.johnduhart.me [22:24:06] test.local.johnduhart.me works [22:24:08] te.st.local.johnduhart.me works [22:24:13] ah. right [22:24:40] I made a bunch for production since we needed IPs for each top level [22:24:46] a single one will likely work here [22:25:02] Wildcards in SSL certs are different then than DNS wilcards, if that's how you're confusing this [22:25:17] Ryan_Lane: Yeah, that's for SSL, we can do without that [22:25:36] One domain is fine [22:25:38] sucks to be without SSL, but we can do a dirty hack for ssl anyway [22:25:44] :p [22:25:53] PROBLEM host: deployment-test is DOWN address: deployment-test CRITICAL - Host Unreachable (deployment-test) [22:27:10] :| [22:27:12] petan: Sort of need a webserver on test ;) [22:27:18] sure [22:28:01] done [22:28:08] bah [22:28:13] I wanted to install also squid [22:28:15] well, I fucked that up [22:28:17] is it puppetized? [22:28:28] Ryan_Lane: :p [22:28:30] what [22:28:45] I made *.deployment.wmflabs.org [22:29:11] while trying to make *.d.wmflabs.org [22:29:15] petan: I'll get squid installed after we get the new config working [22:29:20] * johnduhart claps for Ryan_Lane [22:29:35] there's a bug with deleting * records too, so now I need to delete it via LDAP [22:29:43] ok [22:30:04] hexmode: so many bugs :) [22:30:23] in one day :D [22:31:11] well, brewster wasn't a bug :) [22:31:16] it was a downed service [22:31:20] petan: One more thing, can you mount NFS on /mnt for me? [22:31:21] true [22:31:25] sure [22:31:29] not on /mnt [22:31:31] After that I'm waiting on Ryan_Lane and I'm good to go [22:31:40] Ryan_Lane: don't worry [22:31:43] I'm not sure a single letter (*.d. ) is descriptive enough [22:31:43] /mnt already has a filesystem mounted [22:31:49] I know [22:31:55] we can leave as *.deployment, then [22:31:57] it's working [22:31:57] that's why it's /mnt :D long story [22:32:18] well, kind of working [22:32:56] johnduhart: done [22:33:01] Cool [22:33:18] Ryan_Lane: I used mnt on apache because it was big, then I decided to use nfs so I had to umount it remount as backup storage :D [22:33:31] so now I have 2x20gb on one instance [22:33:37] one for backups and one live [22:33:49] Ryan_Lane:http:/ test.deployment.wmflabs.org doens't point to the right instace? [22:34:03] no? [22:34:04] johnduhart: wait a moment [22:34:08] 208.80.153.215 [22:34:11] it's not configured I guess [22:34:12] it uses that IP [22:34:38] Ryan_Lane: we wanted another public ip [22:34:42] Ryan_Lane: I needed that and a new public IP for deployment-test, sorry :x [22:34:54] why another public IP? [22:35:03] after it's configured we delete current public ip and switch to new one [22:35:08] why? [22:35:12] Ryan_Lane: Because the current configuration is a mess [22:35:20] And I'm reworking it [22:35:26] Once it's ready we can switch [22:35:32] because we want to use that site tommorow for testing and we don't have a time to fix configuration today, so john would fix it and meanwhile we would be using old one [22:35:42] once new one is configured we switch to it and drop old one [22:35:53] I don't see why you need a new IP for that [22:36:00] to go to a new instance? [22:36:00] oh [22:36:02] heh [22:36:06] heheh [22:36:15] it's not so easy to test wildcards on proxy [22:36:18] ok. as long as we'll kill off the other one soon [22:37:03] yup, should be ready on wednesday [22:38:59] johnduhart: do you have a TODO list for this new instance? i.e. is there something i can do? [22:39:18] I'm having a strange issue on my local wiki - the stat page times out [22:39:31] OrenBochman: you mean your laptop? [22:39:51] jeremyb: Sorry, don't have a TODO at the moment [22:39:51] on my desktop in the office [22:40:32] I increased php time out to 60 seconds [22:41:34] johnduhart: well i can try to take something if you want to offload. but i'm also working on some RL distractions too so no huge commitments [22:41:47] jeremyb: I'll let you know, thanks :) [22:41:55] johnduhart, petan: ok. I added a new IP, and a wildcard entry [22:42:00] great [22:42:01] Great. [22:42:09] it isn't associated with an instance [22:42:21] Failed to allocate new public IP address. [22:42:27] *some* dns names will still point to the old IP, till the cache purges [22:42:28] how do I insert it? [22:42:33] petan: it's already there ;) [22:42:36] ah [22:42:48] Ryan_Lane: Do you have shell access btw? [22:42:58] to your instances? [22:42:59] yeah [22:43:03] Ryan has shell access even to your router johnduhart [22:43:06] I mean general cluster [22:43:08] petan: :o [22:43:10] :D [22:43:11] heh [22:43:15] I have access to everything [22:43:17] yes [22:43:29] * Ryan_Lane is on ops [22:43:36] Just stick a keylogger on your instance, get Ryan_Lane to fix something and you own wikipedia. [22:43:39] :P [22:43:46] OrenBochman: seen https://www.mediawiki.org/wiki/Manual:Performance_tuning ? [22:43:55] Damianz: nah, we don't use passwords in the cluster [22:43:57] yes [22:44:04] hm. weird. the * entry doesn't show up [22:44:11] Ryan_Lane: what should I register as domain? [22:44:13] nothing? [22:44:13] Ryan_Lane: Buut we have a cluster, brute forcing your key won't take that long...... [22:44:17] :P [22:44:19] host name [22:44:19] OrenBochman: also, https://www.mediawiki.org/wiki/Manual:How_to_debug [22:44:20] petan: it's already registered [22:44:26] ok, so I keep it empty? [22:44:30] Damianz: I use a different key for labs [22:44:40] awww [22:44:44] johnduhart: it should work [22:44:44] and I don't forward my production one anywhere except production [22:44:45] Ryan_Lane: Heh, I might have a few questions about the steps between main.conf for apache and CommonSettings.php [22:44:53] petan: works [22:44:56] ok [22:45:00] :o [22:45:00] johnduhart: ok [22:45:05] I probably should but I'm lazy, at least my ldap and key passwords are different [22:46:00] btw Ryan_Lane could I join test labs project? [22:46:03] RECOVERY host: deployment-test is UP address: deployment-test PING OK - Packet loss = 0%, RTA = 0.72 ms [22:46:11] Ryan_Lane: So, in main.conf each project has a DocRoot like this [22:46:12] DocumentRoot "/usr/local/apache/common/docroot/wiktionary.org" [22:46:15] I am actually pretty interested in cthat production clone [22:46:26] or is it only ops project now? [22:46:30] Does this mean there's a sepercate copy of MW for each project is the a symlink? [22:46:35] is that* [22:46:48] no [22:46:52] we alias in mediawiki [22:47:03] the docroot has things like robots.txt, favicon, etc [22:47:08] Ah! [22:47:19] I see now, that always threw me off [22:47:25] many don't have anything set, and a default is set for them [22:47:39] you don't have shell on production, do you? [22:47:49] do we publish our apache config? [22:47:50] Ryan_Lane: No [22:47:57] Ryan_Lane: http://noc.wikimedia.org/conf/ [22:48:07] ah. good. we do [22:48:28] Ryan_Lane: there's pieces missing [22:48:37] why doesn't that surprise me [22:48:48] Ryan_Lane: like how are initializesettings/commonsettings loaded? [22:50:01] jeremyb: the config we have on our site is actually taken from noc but rewriten [22:50:10] so that's how it's loaded, I guess [22:50:21] Common is instead of Local [22:51:06] petan: i'm pretty sure there's multiple files missing... (i.e. in the same svn with the stuff from noc.wm.o/conf but not published there) [22:51:15] anyway, bbl [22:51:20] sure [22:51:40] Ryan_Lane: Okay question two, what is in /usr/local/apache/common/docroot/mediawiki/w? Is that a whole MediaWiki instace? I ask because MWMultiVersion changed a few things [22:51:54] Ryan_Lane: is testlabs available for regular devs or it's restricted for now? because I see only ops there... [22:51:54] yes, whole mediawiki instance [22:52:16] petan: restricted to ops for now [22:52:20] ah, ok [22:52:20] well, as root anyway [22:52:38] Ryan_Lane: And what is LocalSettings loading? [22:52:59] I'm starting to think we don't necessarily need a testlabs project [22:53:06] heh [22:53:07] johnduhart: CommonSettings.php [22:53:19] :P [22:53:27] well, if we handle each of the pieces in a different project, we can add people to those things individually [22:53:35] johnduhart: that's actually how the current config is loaded [22:53:40] settings.php is Common [22:53:43] I know [22:53:45] I renamed it and rewriten [22:53:56] Ryan_Lane: speaking of which... we should move some of that into public version control and allow ppl to submit patches for it (maybe even in gerrit). i think last time i brought that up someone said it's a dev thing and it's up to them where to put it/move it or not. (i.e. ops didn't want to impose on devs) [22:54:12] jeremyb: yes. it's up to the devs [22:54:22] so, talk to them about it [22:54:31] though, really, we kind of need it for labs [22:54:37] exactly [22:54:57] also, hard to know what would be involved in splitting to public+private parts without being able to see it [22:55:15] anyway, running away [22:55:34] Ryan_Lane: I'm asking that beacuse in CommonsSettiongs.php is seems to require MWMultiVersion being loaded before CommonSettings.php is [22:55:53] I dunno [22:55:58] I'm not an expert on the mediawiki confgi [22:55:59] *config [22:56:24] Okay [22:56:26] Roan would be good to ask. or Tim. [22:56:57] I'll ask some questions when he's around tomorrow. [22:56:59] * johnduhart gets to work [22:57:03] You'd be better asking Aaron about the hetdeploy esk things [22:57:08] Seeing as he wrote 99% of it [22:57:14] heh [22:57:22] johnduhart: what's dn? [22:57:25] for test [22:57:30] johnduhart: pls hilight when you ask so i can read about it too :) [22:57:34] *.d.wmflags? [22:57:48] wmflags :) [22:57:50] labs [22:57:55] petan: http://hi.petan.deployment.wmflabs.org/ [22:57:59] ah [22:58:04] so it's deployment... [22:58:11] I hoped it would be d [22:58:12] it should be d [22:58:14] so it's shorter [22:58:18] wait [22:58:26] no. it's deployment. heh [22:58:31] I thought we agreed it was deployment along with Platonides [22:58:34] yeah [22:58:36] For clarity instead of shortness [22:58:41] d isn't descriptive enough [22:58:46] hm... [22:59:34] ok johnduhart don't forget to make a log or notices of what you have changed somewhere, on wiki :) [22:59:39] Sure :) [23:00:00] I was actually thiking of writing some manual, but if you are changing it, it probably doesn't matter [23:00:13] because everything is going to change [23:00:26] I don't know if we should use mww or auth wiki [23:00:37] because our db is not so stable and with no backup [23:00:44] so maybe mww would be better [23:00:53] mww, auth wiki? [23:00:59] mediawiki.org [23:01:03] http://deployment.wmflabs.org/auth/wiki/Main_Page [23:01:03] ah [23:01:04] auth [23:01:12] we need to bring up a wmflabs.org wiki [23:01:14] it's called auth because I created SUL there [23:01:16] petan: I'm off for today [23:01:27] hexmode: ok [23:01:37] Ryan_Lane: hm... that's a good idea [23:01:44] maybe create a project for it :D [23:01:48] heh [23:01:54] Ryan_Lane: /usr/local/apache/* is scaped and not NFS, right? [23:02:00] yes [23:02:02] * johnduhart is the man of many questions [23:02:07] heh [23:02:12] Ryan is a man of answers :D [23:02:53] Ryan_Lane: btw didn't you want to merge wikitech with labs? [23:02:59] petan: Would you accept etherpad as a work log :o [23:03:01] though for this I'll likely often be a man without answers [23:03:13] johnduhart: that would be cool too [23:03:16] johnduhart: why not !log and wikitech? [23:03:27] or !log and labsconsole [23:03:42] !logging [23:03:42] To log a message, use the following format: !log [23:03:43] :o [23:03:53] Ryan_Lane: Well think of it like this, it's a work log and semi-readable documentation [23:03:53] etherpad works too, but it would be nice to move it to a real wiki at some point [23:04:00] It will be [23:04:03] * Ryan_Lane nods [23:04:07] http://etherpad.wikimedia.org/DeploymentPrepReConfig [23:04:33] !etherpad is http://etherpad.wikimedia.org/$1 [23:04:33] Key was added! [23:04:51] !ep alias etherpad [23:04:51] Successfully created [23:05:22] !epad is multiplayer notepad [23:05:22] Key was added! [23:05:26] :) [23:05:35] Add ALL the aliases [23:05:52] can we play games in a notepad? [23:05:53] like XO [23:06:03] lol [23:06:14] not if you want me to be productive ;) [23:06:18] :) [23:06:29] * Damianz throws petan a ball to keep him entertained [23:06:34] yay [23:06:36] ball! [23:06:41] I need to sleep [23:06:41] bouncy ball! [23:06:48] I have to wake up at 6 am [23:06:50] it's only midnight! [23:06:54] and it's midnight :D [23:07:09] I need to book my train tickets :( [23:21:15] !log deployment-prep Enabled rewrite rules on deployment-test [23:21:16] Logged the message, Master [23:22:03] johnduhart: thank you for fixing that I need to go [23:22:08] np [23:22:36] it's good to know more people work on that [23:23:21] night [23:23:53] PROBLEM HTTP is now: WARNING on deployment-test deployment-test output: HTTP WARNING: HTTP/1.1 404 Not Found - 482 bytes in 0.006 second response time [23:26:18] http://en.wikipedia.deployment.wmflabs.org/Hello [23:26:19] :D [23:33:17] "If no other wiki is already using the MW version then you probably should stop what you're doing." [23:33:18] heh [23:45:13] !log deployment-prep Installed php5-cli on deployment-test [23:45:14] Logged the message, Master