[00:00:05] Ryan_Lane: will you update https://wikitech.wikimedia.org/view/Gerrit or should I? [00:00:17] I will [00:00:50] ok. in a few minutes, second try. [00:01:01] done [00:01:07] oh [00:01:12] I already added him to that project [00:01:13] sorry :) [00:01:18] I was testing to make sure it would work [00:01:30] sumanah: btw, if they are in channel, we have a couple keywords for this [00:01:35] !account-questions | sumanah [00:01:36] sumanah: I need the following info from you: 1. Your preferred wiki user name. This will also be your git username, so if you'd prefer this to be your real name, then provide your real name. 2. Your SVN account name, or your preferred shell account name, if you do not have SVN access. 3. Your preferred email address. [00:01:42] !initial-login |sumanah [00:01:42] sumanah: https://labsconsole.wikimedia.org/wiki/Access#Initial_log_in [00:02:57] Ryan_Lane: my second try will be with a different user. I figured Benny's account was no longer a clean slate to test with. [00:03:25] ah. cool [00:03:27] sounds good [00:03:51] !new-labsuser is add-labs-user --wikiname="" --mail="" [00:03:51] Key was added! [00:03:58] !new-labsuser |sumanah [00:03:58] sumanah: add-labs-user --wikiname="" --mail="" [00:04:00] :D [00:04:23] !ping | petan [00:04:27] @search labsuser [00:04:27] No results found! :| [00:04:31] !access | petan [00:04:31] petan: https://labsconsole.wikimedia.org/wiki/Access#Accessing_public_and_private_instances [00:04:31] @search new [00:04:32] Results (found 1): newgrp, [00:04:35] damn [00:04:37] @search new- [00:04:37] No results found! :| [00:04:38] heh [00:04:43] search breaks with dashes [00:04:48] really? [00:04:50] yep [00:04:58] @search - [00:04:58] Results (found 12): bot, wm-bot, projects, putty, monitor, alert, unicorn, help, instance-json, leslie's-reset, socks-proxy, new-labsuser, [00:04:59] @regsearch new- [00:04:59] No results found! :| [00:05:06] @search instance- [00:05:06] Results (found 1): instance-json, [00:05:10] @search labsuser [00:05:10] No results found! :| [00:05:20] @search -labs [00:05:20] Results (found 1): new-labsuser, [00:05:23] maybe it takes time to index? [00:05:28] @regsearch new- [00:05:28] No results found! :| [00:05:34] @reload [00:05:34] Channel config was reloaded [00:05:36] i just did that ;) [00:05:36] @regsearch new- [00:05:36] No results found! :| [00:05:42] wtf [00:05:45] heh [00:05:49] @regsearch - [00:05:49] Results (found 12): bot, wm-bot, projects, putty, monitor, alert, unicorn, help, instance-json, leslie's-reset, socks-proxy, new-labsuser, [00:05:55] @regsearch new- [00:05:55] No results found! :| [00:05:59] weird [00:06:02] @diaf [00:06:11] 01/14/2012 - 00:06:10 - Creating a home directory for gwicke at /export/home/bastion/gwicke [00:06:20] @search socks- [00:06:20] No results found! :| [00:06:25] @search -proxy [00:06:25] No results found! :| [00:06:30] gwicke: hi. [00:06:30] @search socks-proxy [00:06:31] Results (found 1): socks-proxy, [00:06:35] o.o [00:06:39] oh yes [00:06:40] I know [00:06:45] it's actually correct [00:06:46] hi sumanah [00:06:51] new is a special word? [00:06:57] bot is searching for whole keys or part of value [00:06:58] gwicke: please use the instructions at https://labsconsole.wikimedia.org/wiki/Access#Initial_log_in [00:07:06] it doesn't match part of key [00:07:09] never [00:07:11] 01/14/2012 - 00:07:10 - Updating keys for gwicke [00:07:34] @regsearch new-user [00:07:34] No results found! :| [00:07:36] ugh- password-based login? [00:07:39] @regsearch new-labsuser [00:07:39] Results (found 1): new-labsuser, [00:07:43] here we go [00:09:36] gwicke: eh? for what? [00:09:39] gwicke: the wiki, yes [00:09:43] gwicke: instances, no [00:09:46] sudo, yes [00:09:56] at some point, we'd like kerberos [00:10:00] petan: OrenBochman: hexmode: Status update? [00:10:08] * gwicke is relieved not to receive a shell password per mail [00:10:11] 01/14/2012 - 00:10:11 - Updating keys for gwicke [00:10:11] pt wikibooks imported [00:10:21] site is now beta.wmflabs [00:10:34] johnduhart: should I do something else? [00:10:49] Hold on what? We changed from deployment to beta? [00:10:49] Ryan told me that we finally got new hw for sql, so next week probably would have it [00:10:59] johnduhart: yes [00:11:09] Hold that note, let me check the configuration [00:11:13] johnduhart: robla and hexmode talked about it in #mediawiki a few hours ago I think. [00:11:32] johnduhart: I did it based on hex request who is lead of this, imho :o [00:11:37] Should [00:11:39] Sure* [00:11:43] he came with whole idea of this deployment :) [00:11:49] I just helped him to build it [00:12:27] Sure. [00:14:26] sumanah: got shell, thanks! [00:14:55] great, glad gwicke [00:14:58] hm [00:15:02] formey is kind of dead [00:15:25] petan: Can you please use more descriptive commit messages? https://github.com/johnduhart/deploymentprep-conf/commits/ [00:15:35] yes I can [00:15:40] I just did it really quickly [00:18:00] sumanah: is there a guide on how to access / set up static webspace? If DNS is required for that, then I don't have the rights for it. [00:18:23] gwicke: you are in a room with Ryan SuperLabs Lane and he knows infinity percent more than me, sorry [00:18:39] okay, okay ;) [00:18:49] so, Ryan_Lane ;) [00:18:52] I'm fixing things again :( [00:18:57] gimme a little bit [00:19:03] np- not urgent at all [00:19:11] yeah, when you set up a supercomplicated system sometimes you turn into its janitor :D [00:19:24] well, I'm fixing a production problem right now :0 [00:20:44] :O [00:21:31] svn is down? [00:23:14] gwicke: it is essentially a shell password... it can be used to reset your key. it can be used to sudo [00:23:43] yep, I realized that once I got to that config screen [00:25:34] jeremyb: do those actually work across browsers? [00:25:36] i don't follow [00:25:37] never encountered anybody using it [00:25:43] there's at least a million users i think [00:25:45] jeremyb: formey is being rebooted [00:25:52] (at least the entire US army) [00:26:03] johnduhart: the bot runs on formey? [00:26:08] johnduhart: (wm-bot) [00:26:16] * gwicke never was in any army, the kindergarten I worked at didn't use certs ;) [00:26:22] jeremyb: bots-1 [00:26:31] gwicke: you worked at a kindergarten?! [00:26:43] petan: makes more sense ;P [00:26:59] yeah, good preparation for later tasks in life ;) [00:27:26] gwicke: anyway, the point is every browser under the sun (at least for desktop) surely supports client certs. idk about opera but just about everything else i can think of [00:27:31] gwicke: that's neat [00:28:05] gwicke: also, Ryan_Lane worked for ppl that really, really like client certs in a past life i think. (the army's sibling) [00:28:36] so, we could do client certs [00:28:42] but..... [00:28:50] when someone loses one, what are we going to do? [00:28:50] have to check the Android browser [00:28:58] also, it's local to whatever system you are on [00:29:29] Ryan_Lane: the latter is not unique to the browser variety though [00:29:30] sure. i'm not saying do client certs ;) [00:29:44] what I'd like is two facter auth via sms [00:29:58] eww, i hates teh sms [00:30:09] well, what else would you recommend? [00:30:13] gpg maybe [00:30:30] that's a great way to make sure no one will be able to use labs ;) [00:30:33] would at least allow to send encrypted passwords [00:30:43] ah. for email? [00:30:50] yeah [00:31:02] that means clients need to have gpg keys, though [00:31:33] no easy solution I guess.. [00:31:34] What are we talking about? [00:31:59] Ryan_Lane: either YubiKey or something like http://barada.sourceforge.net/ [00:33:30] yubi key is for shell login, right? [00:33:42] it has a few modes [00:33:49] I'm a fan of sms multi-factor [00:34:02] what's the downside? [00:34:22] Android seems to miss client SSL cert support: http://code.google.com/p/android/issues/detail?id=8196 [00:35:41] Ryan_Lane: certainly better than just a plain-text mail [00:35:56] the plain text mail is just for a temporary password [00:36:07] gwicke: not for the initial password but every log in i think [00:36:21] you are required to change your password when you get it [00:36:24] Ryan_Lane: 1) it's (i think?) not crypted on the wire or in the air. (could be sniffed?) 2) what if the carrier's compromised? (idk how secure those systems typically are) 3) means that you must be able to get texts and at least in the US that means pay for them [00:36:39] jeremyb: it would be for every login [00:36:43] jeremyb: and it's two factor [00:36:51] you'd still have to use a password [00:36:53] and it's time limited [00:36:54] ah, then you can't log in without a mobile [00:36:55] yes, i know [00:37:09] correct, can't login without a mobile with two factor sms [00:37:20] Ryan_Lane: still researching yubi but what about those 2 options? [00:37:39] a special shell could also be used to echo a temporary password [00:37:52] that could then be used for the web interface [00:38:20] but how would you access the special shell? ;) [00:38:20] barada requires android [00:38:27] jeremyb: ssh key [00:39:06] gwicke: which essentially means that ssh is the one factor? or you enter the echo'd phrase+your password? [00:39:44] that's a chicken and egg problem [00:39:49] the latter would be safer, but I was thinking of the first [00:40:01] how do you get the initial password to add your ssh key? [00:40:12] you just supply the key [00:40:16] before setup [00:40:24] Ryan_Lane: i'd have to search more but i think it's possible that the same method could be used with both yubi and barada and then ppl could choose which one of the 2 to use for themselvesr [00:40:28] themselves* [00:41:08] gwicke: we want people to be able to sign up without us doing anything [00:41:34] ah [00:41:44] in that case, though, they submit their own password [00:41:50] and they are required to give an email address [00:42:38] without any information from the user, safe communication is hard to establish [00:42:48] Ryan_Lane: also, i think it's very likely that yubi can do web. barada idk as much about [00:43:15] * gwicke should get to parser tests.. [00:43:23] get back, even.. [00:43:37] could do initial auth by reading a fingerprint over phone and also sending some other way [00:43:47] of course it helps if you already know the voice somehow [00:44:26] also you could have more than one level of acct security and add ppl to groups depending on how secure their acct is [00:44:42] some ppl can do 2 factor but not all. some projects can require 2 factor [00:49:53] I'd recommend everyone make proposals [00:49:59] I'm open to better auth [00:50:09] it's hard to track options in irc, though :) [00:50:17] and everyone can demo auth things in labs [00:51:09] Ryan_Lane: well what about ops? is there 2 factor? interest in 2 factor? [00:51:23] I dunno [00:51:25] Labs in labs [00:51:34] things need to be tested [00:51:39] johnduhart: I have labs in labs [00:51:44] how do you think I test things? :) [00:51:52] labception [00:52:02] Ryan_Lane: I thought you prayed alot [00:52:07] I do that too [00:52:12] ah [00:52:16] some things you just can't properly test [00:52:25] yesterday and today have been prime examples of that [00:52:30] hehe [00:52:39] thinking of that [00:52:41] it's time to test dns again!!! [00:52:47] who feels lucky? [00:53:15] Do you feel lucky, punk? [00:53:31] * johnduhart wonders what he's going to work on tonight [00:53:34] we'll see soon enough [00:53:56] then after that is testing ldap being down :) [00:57:34] !initial-login | ezachte [00:57:34] ezachte: https://labsconsole.wikimedia.org/wiki/Access#Initial_log_in [00:57:53] PROBLEM host: labs-mw2 is DOWN address: labs-mw2 check_ping: Invalid hostname/address - labs-mw2 [00:58:06] OH NO! [00:58:50] gosh what will we do without labs-mw2 [00:59:07] johnduhart: take some more stuff down [00:59:23] PROBLEM host: bots-1 is DOWN address: bots-1 check_ping: Invalid hostname/address - bots-1 [00:59:46] well, dns is still a problem [00:59:53] PROBLEM host: master is DOWN address: master check_ping: Invalid hostname/address - master [01:00:13] PROBLEM host: nova-dev1 is DOWN address: nova-dev1 check_ping: Invalid hostname/address - nova-dev1 [01:00:15] Ryan_Lane: yubi has a service you can run to verify OTP tokens against. (which i assume means you could use it for the web. and since there's >=2 DFSG-free implementations of the service so the code/algorithms could be reused if needed) barada has a pam module. [01:00:43] RECOVERY host: bots-1 is UP address: bots-1 PING OK - Packet loss = 0%, RTA = 0.46 ms [01:00:49] Ryan_Lane: anyway, i'm interested in working on it but more research is def needed [01:00:53] RECOVERY host: labs-mw2 is UP address: labs-mw2 PING OK - Packet loss = 0%, RTA = 0.58 ms [01:01:13] RECOVERY host: master is UP address: master PING OK - Packet loss = 0%, RTA = 2.47 ms [01:01:53] RECOVERY host: nova-dev1 is UP address: nova-dev1 PING OK - Packet loss = 0%, RTA = 0.54 ms [01:06:12] 01/14/2012 - 01:06:11 - Creating a home directory for ezachte at /export/home/bastion/ezachte [01:07:11] 01/14/2012 - 01:07:10 - Updating keys for ezachte [01:11:06] ! [01:14:56] * Ryan_Lane is away for a bit [02:10:17] Ryan_Lane: back? [02:13:01] Ryan_Lane: Looks like $wgSiteMatrixSites is wrong on wmflabs/beta [02:13:07] pointing to wmf live urls [02:13:10] hosts [02:13:18] http://labs.wikimedia.beta.wmflabs.org/wiki/Special:SiteMatrix Wikipedia
w
[02:15:50] Hm.. unless those portal are going to be tested, probably ok this way [02:16:52] Krinkle: it would be good to fix it [02:17:07] that's the beta equivalent though ? [02:17:16] Krinkle: what I really want fixed I have been looking for you to do [02:17:32] *drumroll* [02:18:17] Krinkle: http://labs.wikimedia.beta.wmflabs.org/wiki/Problem_reports last two reports here [02:19:13] ok. I already know both solutions [02:19:14] will comment [02:19:18] Also, will try to set the $wgSiteMatrixSites [02:19:46] http://labs.wikimedia.beta.wmflabs.org/viewfile.php?file=mc.php [02:19:57] that's still broken too ;-), th at's what I've been looking to see fixed for a week [02:20:31] Krinkle: do you have a labs account? I could give you access [02:20:39] The source code for viewfile is so awesome [02:20:52] I have 'a' labs account, yes. [02:20:57] 'krinkle' [02:21:20] Reedy: do you know if users were nuked between deployment>beta ? [02:21:20] does it 'work'? [02:21:28] no idea [02:21:35] I'm fairly out of the loop [02:21:43] deployment.wmflabs redirects but not preserving subdomain [02:21:46] anyway, that's old now [02:22:05] Reedy: Yes, my labs account works for testwarm.wmflabs and bastion [02:22:13] This time a year ago I was in SF [02:23:00] I can feel it's a VM. creating an account takes 20 seconds [02:23:12] (between pushing button and seeing the next page that is) [02:23:34] it's cause mysql is virtualised [02:23:46] and that sucks [02:24:47] hm.. selective interwiki ? http://labs.wikimedia.beta.wmflabs.org/wiki/User_talk:Krinkle [02:24:53] mw->live, m->beta, commons->error [02:24:56] Krinkle: only the DNS was changed, users should be find [02:25:05] my account didn't exist anymore [02:25:08] no big deal though [02:26:13] interwikis are broken. they link to .deployment.wmflabs still [02:26:24] lol [02:26:37] and *.deployment.wmflabs redirects to labs.wikimedia.beta.wmflabs [02:27:04] so clicking m:User:Krinkle on http://labs.wikimedia.beta.wmflabs.org/wiki/User:Krinkle goes to http://meta.wikimedia.deployment.wmflabs.org/wiki/User:Krinkle [02:27:08] and back to http://labs.wikimedia.beta.wmflabs.org/wiki/User:Krinkle [02:27:18] lol [02:28:14] 01/14/2012 - 02:28:14 - Creating a home directory for krinkle at /export/home/deployment-prep/krinkle [02:28:20] Krinkle: you should be able to get on now deployment-{web,test,nfs-memc} etc [02:29:00] ok, will check in a minute [02:29:11] 01/14/2012 - 02:29:11 - Updating keys for gwicke [02:29:14] 01/14/2012 - 02:29:14 - Updating keys for krinkle [02:31:39] hexmode: Can you grant me user rights on commonswiki.beta.wmflabs ? [02:32:10] Krinkle: k, any particular group? [02:32:34] is there steward/centralauth in this cluster ? [02:32:48] then I don't need it on each wiki [02:32:51] yep [02:33:25] Krinkle: central auth is new to me... how do I give you steward? [02:33:49] http://meta.wikimedia.beta.wmflabs.org/wiki/Special:GlobalUserRights [02:34:10] I can't see which groups are listed there [02:34:21] but I guess it's there [02:35:37] Krinkle: think you have it now, let me know [02:36:08] yep, got rights commons automatically now [02:36:09] thx [02:37:31] re [02:38:26] OrenBochman: hey [02:38:46] I wonder [02:38:51] what's new [02:39:25] OrenBochman: lqt works [02:39:36] lqt ? [02:40:28] what is the command to get a dir from svn ? [02:40:41] I use an IDE [02:41:00] OrenBochman: liquid threads http://pt.wikibooks.beta.wmflabs.org/wiki/Wikilivros:Di%C3%A1logos_comunit%C3%A1rios [02:41:25] OrenBochman: I'm not sure what you mean by a dir [02:41:38] directory [02:42:02] OrenBochman: do you mean "What is the command to check out a subversion directory?" [02:42:22] yes [02:42:48] OrenBochman: "svn co SVNURL" [02:43:12] 10x [02:43:20] what directory do what to check out? [02:45:01] languages/messages [02:47:06] OrenBochman: if you have svn write access: "svn co svn+ssh://svn.wikimedia.org/svnroot/mediawiki/trunk/phase3/languages/messages" [02:47:39] if not, http://svn.wikimedia.org/svnroot/mediawiki/trunk/phase3/languages/messages/ [02:48:05] not http:// ?I [02:48:11] not https:// ? I meen [02:49:32] that works, too [02:50:15] just checking it out readonly, you can''t commit to it without changing it to svn+ssh [02:50:46] but switching is just an "svn switch" away ;) [02:50:52] sure [02:50:57] just do it right the first time [02:51:03] makes life easier [02:51:04] :p [02:51:35] I don't have check in right in MW just Search [02:52:26] though I did a nice little hack on dumHtml ext [02:54:43] nice icons on that page [02:55:14] you will have more than search [02:55:24] probably just not phase3 and wmf/ [02:55:41] I suppose so [02:56:18] reedy do you do php work mostly? [02:57:25] OrenBochman: reedy does js, too. He is mostly awesome. [02:58:15] Most of the JS I touch is breaking it ;) [02:58:57] do you do server side js too? [02:59:13] You can't really do server side js [03:00:42] why not ? [03:01:01] It's designed to be run in a browser [03:01:02] Node.js ? [03:01:14] Well, yes [03:01:19] but that's a minority usage [03:01:43] server side has been around since a year after it was inveted... [03:02:19] anyhow I wanted to ask how hard it would be to make an extenstion that instruments links [03:02:31] instruments? [03:02:45] adds an ajax call to a tracker when they are clicked [03:02:58] We have a click tracking extension already [03:03:09] that WMF uses for feature related projects [03:03:20] feature ? [03:03:26] like the fund raiser [03:03:38] Moodbar, ArticleFeedback, WikiLove etc [03:03:47] ok [03:03:58] http://svn.wikimedia.org/viewvc/mediawiki/trunk/extensions/ClickTracking/ [03:04:11] 1 # ClickTracking lets you track users' clicks on links/buttons. [03:04:11] 2 # The current implementation tracks clicks on the navigation bar - the [03:04:12] 3 # WikiEditor toolbar makes use of this if present [03:05:49] can you get the user's who clicked too ? [03:06:04] I don't think so [03:06:27] I'd need an option for that [03:06:36] Little of the data tracks stuff to identify the individual user [03:06:47] yep [03:06:50] And that's purposely [03:06:59] I know all about it [03:07:04] ;-) [03:07:05] So for WMF deployment, you're very unlikely to get that approved [03:07:38] I would be sneeky [03:08:01] No, you wouldn't be allowed to do it [03:08:14] c.f. Gollum [03:09:40] actually I have an implimentation in mind that would allow tracking anounymoult [03:09:44] actually I have an implimentation in mind that would allow tracking anounymouly [03:10:29] I think you're ok to track a users habits, as long as you can't say it was X user [03:10:58] that's what I need for the short time [03:11:40] I'd like to track the click stream of a user that follows [03:11:54] 1 a search action [03:12:03] 2. a dismabiguation page [03:12:36] I could use a sample of this kind of data [03:13:36] the reason I want to track user's is to make a personalised search feature [03:14:06] but it could be done useing a cryptographic hash [03:17:12] Krinkle: $wgSiteMatrixSites ?? should be what? [03:17:42] hexmode: I don't know, its currently pointing to the www. portals of wmf projects. we don't have those at labs yet [03:18:40] Krinkle: I'm looking at CommonSettings.php ... will start looking at the documents [03:41:39] Krinkle: http://labs.wikimedia.beta.wmflabs.org/wiki/Special:SiteMatrix [03:41:55] shouldn't webfonts handle burmese [03:42:26] dunno, I think it has an exception for lists like these [03:42:30] ? [03:43:06] my.wikipedia.beta.wmflabs.org/wiki/ [03:43:09] ugh [03:43:35] http://ta.wikipedia.beta.wmflabs.org/wiki/ works, though [03:43:42] filing a bug [03:44:34] what bug ?. [03:44:45] oh burmese=my [03:44:45] I don't think webfonts does burmese [03:45:02] it's also not in the module registry on that page [03:45:23] http://labs.wikimedia.beta.wmflabs.org/wiki/Special:Version [03:45:23] https://my.wikipedia.org/wiki/ it *should* though [03:45:25] it's not installed [03:45:31] webFonts [03:46:17] None of the fonts are listed as burmese [03:46:22] https://my.wikipedia.org/wiki/Special:Version?uselang=en yep [03:46:31] http://svn.wikimedia.org/viewvc/mediawiki/trunk/extensions/WebFonts/fonts/ [03:46:52] not installed, but I'm seeing blocks instead of chars! Fix it webfonts! [03:47:19] Reedy: http://svn.wikimedia.org/viewvc/mediawiki/trunk/extensions/WebFonts/fonts/Mymr/ [03:47:25] that one should work [03:47:34] Hmm [03:47:36] Myanmar [03:47:39] unobvious much? [03:48:00] Myanmar == Burmese ... keep track of your nasty world politics ;) [03:48:11] Myanmar == Burma [03:48:14] lol [03:56:41] Reedy: http://my.wikipedia.beta.wmflabs.org/wiki/ -- works! [03:57:06] all I did was set wmgusewebfonts true :) [04:09:35] no webfonts for dz :( [04:10:10] or to/tibetan [04:13:20] or bo [04:25:36] is there a way to ask apt-get what packages are installed [04:25:53] dpkg -l [04:26:13] OrenBochman: dpkg -l | grep ^i [04:26:45] OrenBochman: or, if you're looking for a particular one, then "dpkg -l pkg" [04:27:05] or "apt-cache show pkg" -- it will say installed [04:27:51] thanks [04:30:51] Krinkle: still getting mw.util problems on http://en.wikipedia.beta.wmflabs.org/wiki/Main_Page?usedebug=true when not logged in [04:33:28] and when logged in [04:33:39] and using debug=true or just no debug [04:55:30] hexmode: can you check the OAIrepository auth user and password [04:55:44] or tell me where to find them [04:55:49] sure, how do I do that? [04:55:55] is it a config var? [04:56:03] should be [04:56:08] it is an extention [04:56:25] !OAIrepository [04:56:57] looks like OAI settings are in CommonSettings.php [04:58:30] is there a pass word in there [05:00:05] I can't see one [05:01:19] that probably meens it is not secure... [05:01:43] OrenBochman: all I see: http://pastebin.com/19Cm4k4F [05:01:49] add one? [05:02:20] can you check what is happening in production ? [05:02:23] PROBLEM Disk Space is now: CRITICAL on puppet-lucid puppet-lucid output: DISK CRITICAL - free space: / 14 MB (1% inode=35%): [05:02:59] I believe one is needed - when I tested it yesteday it requested auth [05:03:37] tested in wikipedia [05:06:18] OrenBochman: Do you know what I should change? [05:07:23] PROBLEM Disk Space is now: WARNING on puppet-lucid puppet-lucid output: DISK WARNING - free space: / 40 MB (3% inode=35%): [05:08:57] p.s. nothing about this in the extension docs [05:21:17] OrenBochman: glancing at the code it looks like OAI is just using the HTTP creds [05:24:20] I will now try to index again [05:25:54] is there a command to allow me to run the command so that it will go on if I disconnect ? [05:26:06] I saw petan using it [05:42:48] dammmmmmmmmmmmmn [05:55:16] OrenBochman: screen [05:55:57] hexmode: there are more and more problems with this installation [05:56:03] :P [05:56:04] :( [05:56:19] * hexmode tries to find the emoticon for crying [05:56:32] :'( [05:56:33] ? [05:56:50] OrenBochman: i'm quite behind on scrollback... what's the current obstacle? [05:56:53] I had suggested we put in an empty MedaiaWiki installation on the server so that the setup scripts will work [05:57:58] and then gradually modifing the setup to be more like production [05:58:18] insetead I believe we put the wagon infront of the horse [05:59:05] the current directory layout is nothing like production and the same goes for the server names [05:59:21] and ofcourse we are missing all the scripts [05:59:40] ? shall i repeat my question? [06:00:08] which should be sanitized by ops who don't know how things realy work [06:00:16] I'm almost done [06:00:30] who *don't*? [06:01:17] yes who don't know how what the search in the script means [06:01:28] and what files are necessary [06:01:49] the production keeps working as long as no one messes with it [06:03:01] try running sudo /usr/local/search/ls2/build [06:03:24] (deployment-search) [06:03:53] now it cannot find InitialiseSettings.php [06:03:54] * jeremyb looks [06:04:38] config never worked because it could not find the mainatence dirs [06:05:40] deployment-wmsearch? [06:05:51] correct dude [06:10:25] and the problematic setting comes from the lsearch-global.conf [06:11:49] by the end this setup is going to be quite different from production... [06:30:56] OrenBochman: so, there's currently an NPE here: https://svn.wikimedia.org/viewvc/mediawiki/trunk/lucene-search-2/src/org/wikimedia/lsearch/config/GlobalConfiguration.java?revision=82929&view=markup#l840 [06:32:01] NPE? [06:32:09] null pointer exception [06:33:08] I don't think so [06:33:19] i'm certain? [06:35:21] no no no [06:35:34] the problem is that it can find the maintence dir [06:35:47] huh? [06:36:08] do you have a screen running that i can attach to? [06:36:28] which it assumes is god knows where [06:38:36] OrenBochman: look at ls2-build.out in your home [06:38:50] ok [06:42:46] yep [06:42:56] buit is a GIGO situation [06:45:32] i don't know about that... [06:46:01] certainly it would be a good idea to have some docs for that config file format [06:46:05] idk if any exist [06:46:29] there are some - only they cover the previous version [06:47:27] which has a simplet format [06:48:47] simpler [06:49:19] anyhow the real problem is that it has no good error reporting [06:49:29] i've seen worse [06:54:16] that's 200 lines of code [07:03:43] jeremyb: can you get a copy junit3.x into /search/ls2/lib on search-test [07:04:03] there is one in deployment-search [07:16:23] jeremyb: I cannot create a directories in /home - any ideas? [07:16:53] why would you need to? [08:54:03] PROBLEM Current Load is now: CRITICAL on swarm1 swarm1 output: Connection refused by host [08:55:03] PROBLEM Current Users is now: CRITICAL on swarm1 swarm1 output: Connection refused by host [08:55:23] PROBLEM Disk Space is now: CRITICAL on swarm1 swarm1 output: Connection refused by host [08:56:03] PROBLEM Free ram is now: CRITICAL on swarm1 swarm1 output: Connection refused by host [08:57:43] PROBLEM Total Processes is now: CRITICAL on swarm1 swarm1 output: Connection refused by host [08:58:23] PROBLEM dpkg-check is now: CRITICAL on swarm1 swarm1 output: Connection refused by host [10:07:35] OrenBochman: hey [10:07:48] need me? [10:07:57] brb [10:57:23] PROBLEM Disk Space is now: CRITICAL on puppet-lucid puppet-lucid output: DISK CRITICAL - free space: / 39 MB (2% inode=35%): [12:09:00] petan: could you make me admin on cswiki labs? [12:09:15] and importer [12:09:27] username? [12:10:13] Danny_B|backup: ^ [12:10:24] Danny B. [12:10:28] mind the dot [12:11:08] done [12:11:10] and there should be some local bureaucrat there who could add admins if necessarily [12:11:18] done too [12:11:19] :D [12:11:22] ah [12:11:24] good [12:11:26] thx [12:11:28] ;) [12:12:37] in case of any config change or extension installation request, poke me [12:12:57] I think there was a talk related to AFT or something on czech wiki [12:14:16] well, shouldn't i be able to do it myself since having the access to labs? [12:14:36] ok, right [12:15:04] I meant in case you didn't know how [12:16:56] well, it depends... ;-) we'll see, first i gotta find out how to get to shell via putty, since there is documentation only for linux... [12:17:10] i assume you don't use win [12:21:42] no [12:21:54] @search putty [12:21:54] Results (found 1): putty, [12:21:56] !putty [12:21:56] how to tunnel - http://oldsite.precedence.co.uk/nc/putty.html [12:27:51] very nice [12:39:36] hexmode: around? [14:15:20] https://github.com/johnduhart/deploymentprep-conf/commit/02c08fd8e2110d43785b538bef4a4d8b442fac57 [14:15:40] Why was InitialiseSettings touched [14:19:01] hexmode: ^ [14:19:38] johnduhart: sorry. should have logged it. I enabled webfonts for mywiki [14:20:14] hexmode: Don't touch InitialiseSettings, ever. That's suppose to match production. [14:20:33] Secondly, if it's needed, why isn't it on in production? [14:20:38] sure, also created a ticket to enable webfonts in production [14:20:50] hexmode: can you update ep [14:20:58] did you receive my email :o [14:21:10] petan: haven't checked email yet [14:21:13] ok [14:21:25] johnduhart: what's problem with enabling stuff which isn't on prod [14:21:32] that's why we creted this project [14:21:43] it's not for replicating prod, but to test sw before it's deployed to prod [14:22:10] or he inserted it to InitialiseSettings? [14:22:17] in that case I get it :) [14:22:20] petan: If we're going to do that then, Do it correctly, log it, and document it somewhere [14:22:26] sure [14:22:43] we should create a change log on wiki so that people can track changes [14:22:43] http://labs.wikimedia.beta.wmflabs.org/wiki/File:Tmp4.xcf - new feature in this releasea [14:22:59] petan: the irc log is on the web [14:23:08] I know but we didn't link it from wiki [14:23:21] SOFIXIT ;) [14:23:26] also people probably don't want to read irc logs to get summary of changes [14:23:35] having all on one place would be good [14:23:42] community people don't know what git is [14:23:48] petan: whenever someone does this, it goes to a seperate log [14:23:53] maybe having a page like "extensions we are testing" would be good [14:23:59] I want a page of deviations of production and why on labswiki [14:24:07] !log deployment-prep enabled webfonts for mywiki [14:24:08] Logged the message, Master [14:24:30] petan: those log messages are saved in a seperate place from the irc log [14:24:39] oh really? [14:24:50] I thought it goes to wt [14:24:54] oh [14:24:56] labscolnsole [14:25:11] ah sure [14:25:16] true [14:25:20] !log deployment-prep Reverted change to wmf-config/InitialiseSettings.php [14:25:21] Logged the message, Master [14:25:21] yes, but now that I say, I don't know where exactly [14:25:31] johnduhart: what about making it to Deploy [14:25:32] !log deployment-prep Updated wmf-config/InitialiseSettings.php from production [14:25:32] Logged the message, Master [14:25:40] johnduhart: why are you reverting it? [14:25:40] :o [14:25:55] petan: Yes, just document it elsewhere from the SAL [14:26:01] hexmode: there is override file, probably it should have been moved there [14:26:12] ah [14:26:16] no idea why john didn't move it rather than revert though [14:26:32] nope, webfonts isn't on mywiki [14:26:36] :( [14:26:46] http://my.wikipedia.beta.wmflabs.org/ [14:26:49] You're all here, I have to run [14:26:55] hexmode: yes because he mercilessly reverted :D [14:26:55] k [14:27:04] instead of moving it to another file [14:27:07] :D [14:27:11] petan: now to find the override [14:27:22] InitialiseS*Deploy* [14:27:29] same folder [14:27:38] got it [14:28:07] I would like to have it for Common too [14:28:31] Just do it correctly and document it out of SAL [14:29:04] johnduhart: despite the annoyance thanks for keeping us all on best practices :) [14:29:26] yw [14:30:17] !log deployment-prep enabled webfonts for mywiki properly in IntialiseSettingsDeploy.php [14:30:18] Logged the message, Master [14:30:35] petan: you would like to have what for common? [14:30:38] webfonts? [14:30:57] separate config [14:31:15] I just created it I give it like 20 minutes before john revert me too :D [14:31:30] just don't ping him, maybe he would over look this ^^ [14:32:34] petan: don't forget to log it, Otherwise he may revert it [14:32:42] :) [14:32:55] !log deployment-prep separated common to own deployment file [14:32:56] Logged the message, Master [14:33:38] I also logged a couple tickets to get webfonts coverage for some other languages... I guess they don't have free fonts, though [14:33:46] ok [14:58:31] https://github.com/johnduhart/deploymentprep-conf/commit/25b08c8df0d96d5c90d8c4adafca3c6fe82f15c4 [14:58:38] petan: It doesn't work like that [14:58:50] You don't think there's other changes in that file? [15:00:03] And you lumped an related, undocumented change along with that commit [15:00:11] Undocumented as in not in the commit message [15:00:23] of course I know it [15:00:27] it's still better than before [15:00:43] anything what can be moved off that file is supposed to be moved [15:00:58] better than copy pasting it next time [15:01:06] where should I document it [15:01:45] petan: Document what? [15:01:54] And you lumped an related, undocumented change along with that commit [15:02:01] ah [15:02:04] what change? [15:02:17] The change to common/wmf-config/InitialiseSettingsDeploy.php [15:02:28] didn't notice that... [15:02:36] hexmode: document it :) [15:02:48] No... [15:02:53] Undocumented as in not in the commit message [15:03:16] Don't lump unrelated things together in a commit, that's what I'm getting at [15:03:45] If you're doing that because you don't know how to use git, learn to. [15:29:56] * Beetstra forgot who was working on perl utf8 encoding problems yesterday .. [16:11:18] * Beetstra utters bad word [17:02:12] hexmode: :o [17:02:15] around? [17:02:35] I want to enable some extensions which are waiting for review, actually most of them, right now [17:10:40] re [17:26:59] petan: Don't, wait unil they get reviewed. [17:27:07] why [17:27:13] We're focusing on testing 1.19, not new extensions [17:27:30] We don't want to add new variables into this test. [17:27:32] this place could be used for testing them as well, people need to test on production like env [17:27:43] The current focus is 1.19 [17:27:57] Once we're done with 1.19, then we can test your extensions [17:28:05] mine? [17:28:16] I am talking about extensions in review queue, not only mine [17:29:58] for same reason I could say we should use only extensions which are currently on prod and not head of svn, since most of them are not reviewed as well [17:30:08] and won't be deployed in same time as 1.19 [17:36:20] Uh, no [17:36:49] Extensions that are currently in use on WMF will get their trunk version deployed [17:36:58] They get branched with 1.19wmf1 [17:37:16] And they are all getting reviewed before deployment [17:38:05] anyway I still see no harm, if there was problem we can disable them anytime [17:38:23] it would be enabled for test wiki anyway for now [17:38:41] people are going to test on other wikis [17:40:50] I don't think you understand my concern. When doing an experiment you should be changing one variable, so you can clearly see the results of that change. The variable we're changing is using the latest code instead of 1.18wmf1. Adding new extensions is not part of that. Secondly once 1.19wmf1 gets branched other extensions won't be able to get tested as they won't be part of that branch. [17:54:49] johnduhart: I understand your concern pretty much, however I believe that this is only production like environment we have and it's great wasting of resources if we dedicate only for this one purpose and that to test mw 1.19 and nothing else, there are more devs who would benefit from testing other sw there, actually anything what is going to be deployed one day on wmf sites could be tested there [17:55:31] petan: Wait for testlabs. Any how do you intend to test extensions that aren't in 1.19wmf1? [17:55:35] And* [17:55:43] I see no hard from enabling few extension for some wiki, other wikis (we have 300+ of them) would be unaffected [17:56:05] johnduhart: I was going to test extensions which are waiting for deployment to wmf site [17:56:37] Okay, and when 1.19wmf1 gets branched how are you going to handle that? [17:56:53] handle what [17:57:13] I think you should look at how our branching system works. [17:59:43] definitely I don't think we can't use this site after 1.19wmf is branched, it still can be used of any other deployment of anything else [18:00:07] it's not called 1.19deployment [18:00:39] What? [18:00:49] I am talking about project [18:00:51] on labs [18:01:04] "definitely I don't think we can't use this site after 1.19wmf is branched, it still can be used of any other deployment of anything else" doesn't make sense [18:01:43] ah right I mean I see no reason why we shouldn't use this site after 1.19 is branched [18:01:55] for other sw we are going to deploy [18:02:23] Okay, so for software that isn't in the 1.19wmf1 branch, *HOW* will that be done [18:02:38] that sw is in /live even now [18:03:03] it's checked out /trunk [18:03:04] petan: When 1.19wmf1 gets branches. we will svn switch to it. [18:03:52] right, we can always checkout extensions from trunk to live just to test them, I don't think we should be limited by svn [18:04:27] anyway once 1.19 get branched 1.20 will be in trunk [18:04:45] so I think we should rather keep svn on 1.20 to prepare for next deployment [18:05:00] actually this site should be always running svn head [18:05:13] that's best way to catch bugs asap [18:05:20] petan: When 1.19 gets branched it won't get immediately deployed, and we should continue to follow it until deployment and a bit after [18:05:26] sure [18:05:36] I am talking about situation after deployment happen [18:05:49] uh huh [18:06:51] btw it's possible to checkout revision on 1.19wmf to live and then checkout head of certain extensions from /trunk only [18:07:07] without changing anything in svn [18:07:25] you can switch repo on level of folder [18:07:42] Uh, no [18:07:57] Don't think so [18:08:14] at least you can checkout different rev in other folders [18:08:32] Okay, it's possible [18:12:47] I love how you went and did this mid discussion https://github.com/johnduhart/deploymentprep-conf/commit/fe607bac3bb3e07ac3802ae40e15939aad3ca82d [18:18:17] actually I did it before discussion, I commited it mid [18:19:05] before discussion it wasn't false though [18:19:36] What wasn't false? [18:20:08] that option in Inital [18:21:44] !log deployment-prep Removed myself from the project. [18:21:45] Logged the message, Master [18:21:47] I'm done. [18:22:08] huh? [18:22:51] hexmode: You and petan are on your own, I've had enough. [18:23:12] enough of what, you know I changed it before the discussion? [18:23:18] so I could hardly know you would disagree [18:23:45] actually I disabled it because of what you said [18:25:53] The fact that the goal of this project constantly changes, and that I'm constantly cleaning up stuff after people just doesn't do anything to help me. During the last several days I've been much more angry then any other involvement with a MediaWiki project, and I'm not okay with that. And so, I'm walking away. [18:26:54] I think that goal is same as it was always was to build a test cluster to test software on before deploying it [18:27:47] regarding stuff you disagree with, perhaps if you let the other people know that you disagree with that we could try to find a compromise [18:49:22] !log deployment-prep enabled global blocking [18:49:24] Logged the message, Master [19:04:04] johnduhart: btw I wanted to tell you that you have done a great job on that project, I don't think we would be able to configure it so well as you did, so if you changed your mind let me know and I will happy to put you back [19:27:04] !log bots started LiWa3 on bots-2 - parsed revid parsing [19:27:05] Logged the message, Master [19:27:22] !log bots started XLinkBot on bots-2 - utf8-encoding seems fixed [19:27:23] Logged the message, Master [19:27:58] johnduhart: just got back from a wedding, thanks for all your help till now :) [19:31:28] petan: just catching up on the discussion from the past few hours [19:32:25] petan: johnduhart is right, this particular project isn't for testing new, unreviewed extensions. After we get 1.19 tested, then yes. Till then, please disable them. [20:07:11] stopped pdns on virt1 [20:07:29] seems the recursors in production hadn't restarted when I changed their config via puppet [20:07:32] so, it should work this time [20:20:17] seems to be working this tine [20:20:19] *time [20:21:57] Ryan_Lane: hey! :) [20:22:18] Ryan_Lane: you wouldn't be able to do any squid work would you? [20:22:27] on a saturday? [20:22:35] unless it involves SOPA, no [20:22:43] SOPA! [20:23:02] it's an end run around the DNS blocking of sopa [20:23:22] I already know what work needs to get done for SOPA ;) [20:23:32] :) [20:25:03] !account-questions | hyperon [20:25:03] hyperon: I need the following info from you: 1. Your preferred wiki user name. This will also be your git username, so if you'd prefer this to be your real name, then provide your real name. 2. Your SVN account name, or your preferred shell account name, if you do not have SVN access. 3. Your preferred email address. [20:33:09] petan: johnduhart: updated with scope for clarity http://etherpad.wikimedia.org/DeploymentPrep [20:33:17] continuing to work on it [20:35:09] 01/14/2012 - 20:35:09 - Updating keys for hyperon [20:35:12] 01/14/2012 - 20:35:12 - Updating keys for hyperon [20:36:35] !log testswarm swarm1 failed to build - deleting [20:36:36] Logged the message, Master [20:36:44] ah. shit [20:37:09] 01/14/2012 - 20:37:08 - Updating keys for hyperon [20:37:11] 01/14/2012 - 20:37:11 - Updating keys for hyperon [20:37:57] seems I forgot to update the scripts so that virt0 would be used as the build host [20:55:14] ok. i'm off for a bit [20:55:41] johnduhart: petan: opinions on scope? Agree that we can make an annoucement? [20:57:06] hexmode: I never enabled any extension [20:57:19] I will check it [20:58:06] hexmode: that's what I meant, after test of pre deployment we could it to test other sw we wanted to enable on cluster later [20:58:26] * could use [20:58:29] petan: k, I saw OnlineStatusBar and I don't think that is in prod... but I think we're pretty stable now [20:58:39] where did you see it [20:58:44] it's not enabled anywhere [20:59:11] petan: InitSettingsDeploy has something about it. But this isn't a huge deal [20:59:25] We've done a lot of great work [20:59:32] and I don't want to lose it [20:59:39] yes it contains default=false that mean it's not enabled [21:00:01] I think waiting to announce now just takes wind out of the sails [21:00:11] and I'd like to keep going [21:01:57] hexmode: ok so what are we waiting for now? [21:02:03] I have no idea when search is going to work [21:02:13] nothing? me to write an announcement? [21:02:16] ok [21:02:41] petan: I'll work on the announcement (short and sweet) ... do you know anything about JS? [21:02:49] almost nothing [21:03:08] brion vibber is probably a good person for that [21:03:12] hrm.... k [21:03:33] or krinkle or roan or trevor [21:03:40] but none of them are here :) [21:03:44] no biggie [21:04:10] http://labs.wikimedia.beta.wmflabs.org/wiki/Problem_reports -- last issue on there is why I mention this [21:04:20] ok, now to write a short announcement [21:04:52] petan: oh, mail was not working, right? [21:05:03] I think Ryan_Lane did that on purpose, though [21:05:09] no it's not allowed on labs [21:05:56] k, makes sense, just need to note that in the announcement and on problem_reports [21:08:33] right [21:09:04] hexmode: what you plan to do after deployment happen, just delete everything or use it for something else? [21:09:41] petan: after 1.19? Then we can start on making sure this is all puppetized [21:09:51] and ready for 1.20 [21:09:54] :) [21:09:57] right, configuration and scripts are easy to be puppetized [21:09:59] but db not [21:10:30] also you will always have to set up all vm's nfs, memcached, sql server... [21:10:44] it's not going to be easy to set up still... even if we puppetize it [21:10:51] yeah, we're getting a QA guy who is supposed to be starting in a couple of weeks.... I'll work with him and maybe hand this off to him [21:11:00] right [21:11:04] petan: yeah, automate it all! [21:11:27] that's a question if it's even possible... we can't automate creation of instances ;) [21:11:38] but yes probably most of that is possible [21:11:45] also you will always have to set up all vm's nfs, memcached, sql server... [21:11:51] that's automateable [21:11:54] why not? push this button for a farm [21:12:01] :) [21:12:43] petan: could you svn up and run updatesites? [21:12:53] johnduhart: I didn't find any puppetized and well configured sql server so far, that would be a good thing to start with [21:13:51] petan: you can start on that now! (on another instance, of course ;) ) [21:14:10] I don't want to say it's impossible, just it's not close to be so easy right now, it's gonna take some time unless we set it up, also there are other things you will never automate [21:14:25] how you get users to use that site, you can't puppetize users [21:15:00] when you announce this, we get some people who will probably use this site and find a lot of bugs [21:15:28] when you delete it, we lost them, and later if you create a new testing site using puppet, it will be empty and you will have to find testers again [21:15:42] that's why I think making some "more permanent" project would make sense [21:16:04] puppetizing is definitely a usefull stuff [21:16:15] petan: I've nothing against using the same dns name for future tests [21:16:22] but setting up clone of production for every single test? do we even have resources for that? [21:16:27] nothing in "beta" about 1.19 [21:16:37] petan: not for every one [21:16:55] That's what testlabs is for [21:17:06] but I'm sure everyone doesn't need a whole cluster [21:17:06] johnduhart: what you mean? [21:17:26] anyway afaik testlabs are for devs only and shouldn't be accessible for regular users [21:17:37] I thougt this was supposed to be replacement of prototype server [21:17:57] petan: When testlabs is completed developers will be able to create wikis for testing things [21:18:04] johnduhart: I know [21:18:14] petan: No, users will be able to access wikis hosted on testlabs [21:18:38] that is exactly something I wasn't talking about, I was talking about a site where people can test stuff before it get on production [21:18:43] petan: this particular one is, yes. And if they want larger tests they can request non-SOCKS access. [21:19:07] petan: That *is* what this is [21:19:11] labs si [21:19:13] is [21:19:14] ugh [21:19:18] typing is hard [21:19:18] actually not only devs participate on development there is a feedback from users too [21:20:21] private cluster closed for devs only is usefull but what is difference between testing on your wiki installed on desktop and cluster? I could move production config to wiki on my own pc, main reason I ever moved it to server was that other people could test and have feedback [21:20:27] why set up a new cluster every single time? [21:20:34] petan: I'm confused. What makes you think they couldn't participate on labs? [21:20:55] hexmode: because testlabs are not going to be accessible just as John said right now [21:20:58] also, mail will be enabled at some point [21:21:02] ah [21:21:02] mark and I are working on it [21:21:03] sorry [21:21:08] I didn't see comma [21:21:15] (sorry for the gratuitous ping ma rk) [21:21:17] petan: No, users will be able to access wikis hosted on testlabs [21:21:21] I thougt he said no users heh [21:21:22] :D [21:21:27] oh [21:21:42] and it's possible to automate instance creation [21:21:43] but... [21:21:56] Ryan_Lane: There's really no need to [21:22:05] openstack doesn't currently support all of the things we are doing in openstackmanager [21:22:23] andrewbogott is working on adding those things to openstack [21:22:33] I also don't think we need to automate that [21:22:37] I agree [21:22:41] I don't think we do either [21:24:07] it could be interesting to have snapshots of a copy of the infrastructure for testing [21:24:18] where a test could run, then it would rollback to a snapshot [21:24:22] hm that would actually make it easiest [21:24:49] you would just recover a snapshot and you would have it working in few minutes [21:24:59] yeah [21:25:06] we need to switch to the OS API [21:25:13] I haven't even started that dev work [21:26:23] ok right now, let's make clear one thing this deployment-prep is one time project or are we going to use it even in future? [21:26:45] I'd imagine we'd reuse it [21:26:48] petan: I plan to use the work we've done for future work [21:27:03] at minimum we should use it for puppetization [21:27:11] and documentatio [21:27:26] So in 6 months or so this should be gone and we'd be using testlabs, correct hexmode? [21:27:52] I don't think that nuking it and making everything from scratch before next time, actually I think we could use this working (and accessible) site and keep it running head of mw 1.20 [21:27:56] hrm... this is testlabs, or am I missing something [21:27:56] untill next deployment [21:28:10] maybe we could catch bugs earlier than month before it [21:28:10] he means the testlabs project [21:28:27] deployment-prep isn't testlabs [21:28:32] until its possible to work without root, you guys need to keep using this project [21:28:54] right [21:28:58] it will be never possible you will need to create some roles [21:29:03] we need to reload apache etc [21:29:09] we always will need to do that [21:29:16] well, I want to have scripts that will do this stuff for us [21:29:22] btw how is sql server going to be designed [21:29:33] petan: Like it is in production. [21:29:34] I want it to be an openstack project [21:29:34] every project will have own server instance on some port I guess? [21:29:34] so I'm not familiar with what you all mean by testlabs so... back to writing [21:29:41] hey everyone [21:29:44] johnduhart: I am talking about labs not testlabs [21:29:49] ah [21:30:06] hyperon: ! [21:30:30] hexmode: hi [21:30:40] for the sql servers, I'd like to have a database as a service, where people can create databases through the labsconsole interface [21:30:46] Ryan_Lane: I can't imagine projects would share same instance of sql server, database names would be similar [21:31:00] and we give people the "dbadmin" role in project [21:31:02] *projects [21:31:10] like here we have prod names of databases and testlabs would probably have same names of db too [21:31:11] petan: why not? [21:31:21] I can imagine different projects using the same sql database [21:31:32] Ryan_Lane: what if two projects wanted db named same? [21:31:38] they are SOL [21:31:58] or, we have it named by project [21:32:06] so prefix it [21:32:06] SOL? [21:32:17] prefixes... blah [21:32:21] Ryan_Lane: That's a production deviation! [21:32:22] but either way, we want to be able to share sql between projects [21:32:23] ;) [21:32:32] SOL = Shit Out of Luck [21:32:36] sharing is good [21:32:51] but keeping separate instance of sql server where people could even manage users etc could be usefull [21:33:10] without touching others db's [21:33:12] or users [21:33:17] or whatever [21:33:18] yeah, I'd like to have a "Manage databases" interface [21:33:19] Ideally I think testlabs should have priority for dbnames, so other projects shouldn't be creating enwiki or centralauth db names. [21:33:31] where dbadmins can create/delete databases [21:33:36] and also add/remove users [21:33:39] Ryan_Lane: As long as there's a API interface for mass registering projects. [21:33:42] and manage user's permissions [21:33:46] johnduhart: I think best would be to have completely different instance of sql server for each project [21:33:56] johnduhart: yes. the idea is for this to be an openstack project [21:34:03] which means it's API driven [21:34:11] Ryan_Lane: Mass registering db names, sorry [21:34:15] * Ryan_Lane nods [21:34:17] yeah [21:34:20] ah [21:34:35] I always consider the interface use first [21:35:12] as it helps me figure out how the API should look ;) [21:35:21] right [21:35:58] this is the proposal for this, btw: http://www.mediawiki.org/wiki/Wikimedia_Labs/Create_shared_sql_service_for_all_projects [21:36:03] it's basically empty right now [21:36:15] Ryan_Lane: Question, instead of having a script on testlabs for creating wikis would an extension on labsconsole be better? [21:36:40] well, if we made an API, it could be both [21:36:50] right [21:37:02] so the first step is to make the REST API [21:37:14] first step is to define the API :) [21:37:22] Ryan_Lane: define/make [21:37:25] yeah [21:38:48] I think we should start with mariadb as the assumed db [21:39:04] openstack services are generally written with support for "drivers" [21:39:08] not MySQL? [21:39:32] so, we'd write a mariadb driver [21:39:38] and a MySQL driver [21:39:39] maria support ldap, but mysql has a better support still [21:39:40] mariadb has support for PAM auth [21:39:44] which gives ldap [21:40:27] mariadb is a drop-in replacement for mysql [21:40:37] so 99% of the mariadb driver will be the mysql driver [21:40:56] but nonetheless, in the end it should be ready to support both. [21:41:07] yeah [21:41:14] so we should make sure that 1% is dealt with [21:43:02] I guess it's easier to start with mysql [21:43:12] we can add the mariadb specific stuff after [21:43:17] yes, since thats what i'm familiar with :P [21:43:58] Ryan_Lane: is the sql server going to be managed by puppet? [21:44:09] yes, everything is :) [21:44:20] yes but we can only push to testlabs [21:44:24] so the API should contain ways to create/delete schemas and databases, ways to control access, etc. [21:44:24] but only from the perspective of installing the software [21:44:30] those will be in the testlabs group [21:44:33] err [21:44:33] what is etc.? [21:44:34] branch [21:44:52] that's mostly it [21:45:07] that's what i thought [21:45:15] we should put this onto the proposals page [21:45:22] and if we think of other things, we can addit [21:45:26] *it [21:45:27] so the access controls would be easier with mariadb as it has PAM authentication support [21:45:32] yeah [21:45:40] you can list a user as an external user [21:45:41] I already wrote there that server is going to be managed by puppet [21:46:03] is it going to be in labs nagios? [21:46:03] petan: technically, you guys can push to production too [21:46:06] yes [21:46:09] ok [21:46:13] Ryan_Lane: technically? [21:46:14] how [21:46:22] but, only ops can merge either testlabs or production [21:46:36] I didn't know we can push to prod [21:46:44] you shouldn't though [21:46:50] there isn't really a reason to do so [21:47:07] even ops, in the long run, shouldn't be [21:47:17] everything should go through test, then to production [21:47:22] now it makes sense, later if configuration of mediawiki on wmf wikis was managed by gerrit it would be cool [21:47:37] agreed [21:48:35] there is not anything regarding config we can improve when we can't see to servers on prod, so pushing to prod now isn't really needed for us, I just didn't know it's even possible [21:49:13] other than cases where I would see someone made a fatal mistake merging something and I wanted to revert that :D [21:49:27] heh [21:49:47] so for the db API we'll need a way to create/delete schemas and databases, and control access [21:49:49] okay [21:50:11] ok. I need to leave the computer :) [21:50:17] back online later maybe [21:50:35] same [21:51:50] * aharoni waves at MaxSem and Ryan_Lane [21:52:06] Ryan_Lane: MaxSem suggested asking you for a labs account. [21:52:36] :o [21:52:43] so i'm willing to work on the REST API for database creation [21:53:04] should i create an instance for myself where i can do all the dev work, or should i use an exsisting one? [21:53:29] Ryan_Lane: can you insert me to open stack when you have time [21:53:52] I found it a bit hard to work on code of that when I don't have where to test it [22:56:45] !account-questions | aharoni [22:56:45] aharoni: I need the following info from you: 1. Your preferred wiki user name. This will also be your git username, so if you'd prefer this to be your real name, then provide your real name. 2. Your SVN account name, or your preferred shell account name, if you do not have SVN access. 3. Your preferred email address. [22:57:06] petan, hyperon: do both of you need access to the openstack project? [22:57:13] PROBLEM Free ram is now: WARNING on bots-2 bots-2 output: Warning: 19% free memory [22:57:19] there's some work we'll need to do for this [22:57:26] since you guys will need an LDAP server [22:57:36] hm [22:57:37] maybe not [22:57:44] I guess this doesn't necessarily require one [22:57:54] you can just use keystone with a mysql driver [23:09:56] Ryan_Lane: 1. amire80 2. amire80 3. amir.aharoni@mail.huji.ac.il [23:11:50] !initial-login | aharoni [23:11:50] aharoni: https://labsconsole.wikimedia.org/wiki/Access#Initial_log_in [23:13:10] 01/14/2012 - 23:13:10 - Creating a home directory for amire80 at /export/home/bastion/amire80 [23:14:11] 01/14/2012 - 23:14:10 - Updating keys for amire80 [23:15:09] 01/14/2012 - 23:15:09 - Updating keys for amire80 [23:17:13] RECOVERY Free ram is now: OK on bots-2 bots-2 output: OK: 20% free memory [23:22:22] petan: johnduhart: around? [23:22:37] can I get you guys to glance at my proposed announcement? [23:22:43] http://etherpad.wikimedia.org/119-beta [23:30:51] Ryan_Lane: Mind if I PM you? [23:31:58] sure [23:45:13] PROBLEM Free ram is now: WARNING on bots-2 bots-2 output: Warning: 18% free memory [23:50:55] Ryan_Lane: i will need access [23:51:35] ok [23:52:05] 01/14/2012 - 23:52:05 - Creating a home directory for hyperon at /export/home/openstack/hyperon [23:52:16] you should make your own instance [23:52:34] and use devstack: http://devstack.org/ [23:52:35] so there's an openstack project? [23:52:50] which means you should use an oneric instance, not a lucid one [23:52:54] devstack wouldn't happen to be puppetized? :P [23:53:03] no. it's a script [23:53:06] 01/14/2012 - 23:53:06 - Updating keys for hyperon [23:53:11] and you may have a few issues with it [23:53:18] specifically, it fucks up the apache config [23:53:25] and it may complain about ganglia [23:53:35] when you get to that point, I'll help you [23:53:46] so what's its advantage? [23:54:09] it's a development environment for openstack [23:54:18] it sets everything up [23:55:00] wow, add instance looks better since i last saw it [23:55:08] heh [23:55:13] ignore the puppet stuff for now [23:55:25] I really need to make that collapsed by default [23:55:35] in fact, I'm gonna do that now [23:55:40] what's the naming convention? [23:56:02] nova- [23:56:09] so, maybe nova-daas1 [23:56:14] database as a service [23:56:25] i like that [23:56:45] what size instance? [23:56:48] is small good? [23:56:50] hexmode: That looks good [23:56:52] small is likelt fine [23:57:06] johnduhart: why'd you remove yourself from deployment-prep? [23:57:24] johnduhart: good! I already started the bot ;) [23:57:29] just now [23:57:35] http://toolserver.org/~pathoschild/crossactivity/?user=EdwardsBot [23:57:50] Ryan_Lane: The fact that the goal of this project constantly changes, and that I'm constantly cleaning up stuff after people just doesn't do anything to help me. During the last several days I've been much more angry then any other involvement with a MediaWiki project, and I'm not okay with that. And so, I'm walking away. [23:58:02] ah [23:58:07] sorry to hear that [23:59:29] Ryan_Lane: a lot of it is my fault for learning how to manage a project like this on the fly... but johnduhart's contributions have been great despite that