[04:04:49] Anyone here that knows anything about wm-bot's db? [04:05:39] I think I've tracked down a bug, but am very tired and want a secind opinion. [04:06:31] There's a channel that's getting 3 redundant rc feed posts for every change. [04:07:12] I think it has something to do with two different people trying to put bot in same channel at same time. [04:08:34] One did @add ##Foo and the other did @add ##foo and there were two bouncers in there at once 3 & 5 the op in the room kicked bothe and brought the bot back a third time. [04:09:32] Now, the db seems to be stuck thinking there are three instances and all should post rc feed [04:10:36] I've had bot join ##Foo and ##foo and dropped it and manually cleared everything I coukd find and reloaded db, but no go. [04:10:56] Anyone think a restart might fix? [04:19:59] I have my doubts, and everyone else must be asleep. [06:02:29] T13|sleeps: which channel [09:38:06] [bz] (8NEW - created by: 2Antoine "hashar" Musso, priority: 4Normal - 6major) [Bug 48501] [OPS] beta: get SSL certificates - https://bugzilla.wikimedia.org/show_bug.cgi?id=48501 [09:38:52] [bz] (8RESOLVED - created by: 2Michelle Grover, priority: 4Low - 6enhancement) [Bug 46621] SSL Support on betalabs - https://bugzilla.wikimedia.org/show_bug.cgi?id=46621 [09:56:24] petan, are you around in a bit for some impertinent personal questions? [09:56:53] ooo [10:51:49] PETAN :O [10:51:53] araound? [10:54:29] @seen petan [10:54:29] T13|needsCoffee: I have never seen petan [10:54:35] @seen petan [10:54:35] T13|needsCoffee: petan is in here, right now [10:54:52] Last post I saw of his was a couple hours ago. [10:55:33] [06:55] <+wm-bot2> T13|needsCoffee: Last time I saw petan they were talking in the channel, they are still in the channel #wm-bot at 8/21/2013 8:59:38 AM (1h55m26s ago) [11:12:47] MartijnH: hm? [11:14:00] !logs [11:14:00] http://bots.wmflabs.org/~wm-bot/logs/%23wikimedia-labs/ [11:41:51] boooyeah, http://wikistream-tools.wmflabs.org/ now actually uses websockets! [11:41:51] addshore, ping [11:41:56] pong [11:42:09] addshore, I see you are hard at work on Addwiki. [11:42:18] Do you see me working on Peachy? [11:44:08] I saw https://github.com/cyberpower678/Peachy/commit/e57c649eba5f83acd305b37339ae4829aca12dd9 :P [11:44:53] addshore, I'm almost done alpha 4 and it's slowly going into the beta stage. [12:17:00] @link [[Meep]] [12:17:00] !ping [12:17:01] !pong [12:18:05] @link [[Meep]] [12:18:06] http://enwp.org/Meep [12:21:10] @link [[meaow]] [12:21:10] http://enwp.org/meaow [12:21:12] :D [12:21:17] @link [[:de:meaow]] [12:21:17] http://enwp.org/de:meaow [12:22:00] test [[meaow]] [12:22:32] @link [12:22:33] http://enwp.org/meaow [12:22:53] @link [[de:meaow]] [12:22:53] https://de.wikipedia.org/wiki/meaow [12:23:00] Steinsplitter ^ [12:24:04] thx :D works great xD [12:24:15] without @link is not possible? [12:24:21] @linkie-on [12:24:21] Links will be automatically translated in this channel now [12:24:29] [[de:Pr0n]] [12:24:29] https://de.wikipedia.org/wiki/Pr0n [12:24:44] [[a]] [[b]] [12:24:44] http://enwp.org/a http://enwp.org/b [12:25:07] wow :):):) thx [12:25:59] petan: default language is en, or is possible to change it? :D [12:26:20] [bz] (8NEW - created by: 2Chris McMahon, priority: 4Unprioritized - 6normal) [Bug 53080] no login error messages on http://commons.wikipedia.beta.wmflabs.org/ - https://bugzilla.wikimedia.org/show_bug.cgi?id=53080 [12:26:36] but great funtion :):) awww [12:26:43] https://bugzilla.wikimedia.org/show_bug.cgi?id=53091 [12:26:48] Steinsplitter read it [12:26:52] that documents it [12:27:08] you can use @configure to change default wiki [12:27:30] awww [12:27:32] Value cs was stored into default_link_wiki to config [12:27:34] i LOVE YOU! ;:) [12:27:40] I LOVE the BOT :):) AWWWWWWW [12:27:40] @link [[bla]] [12:27:40] http://cswp.cz/bla [12:27:55] awwww [12:28:02] Value en was stored into default_link_wiki to config [13:20:22] Steinsplitter: no love for me? [13:20:40] Pfft... see if I request any more new wm-bot toys... [13:21:16] T13|needsCoffee: + love for you [13:22:49] [09:21] <@Revi> [[User:Revi]] [13:22:49] http://enwp.org/Revi [13:22:49] [09:21] <+wm-bot5> https://meta.wikimedia.org/wiki/Revi [13:23:06] petan: ^^^ bug [13:23:29] indeed [13:24:56] This value can't be stored [13:25:13] ^^ another one petan ? [13:26:07] @link [[ko:Example]] [13:26:07] https://ko.wikipedia.org/wiki/Example [13:27:25] @link [[User:Petrb]] [13:27:25] http://enwp.org/User:Petrb [13:27:42] T13|needsCoffee: is that really a bug? what happens when you do that on kowiki [13:28:07] in puppet, i get an error "Error 400 on SERVER: Could not find class ldap::role::client::labs " [13:28:07] i love this fuction, i hope a lot of channel enable it [13:28:07] anyone know why? [13:28:07] * aude thinks i've seen this before and fixed it somehow [13:28:07] but can't remember how [13:28:34] petan: the bug is being unable to store ko as default [13:29:50] Value ko was stored into default_link_wiki to config [13:29:55] [[blah]] [13:29:55] https://ko.wikipedia.org/wiki/blah [13:29:58] works to me [13:32:21] oh, I see... I was using - instead of _ [13:32:33] * T13 trouts hisself. [13:51:54] !ping [13:51:54] *�POOF�* Wadda need? *�POOF�* Wadda need? *�POOF�* Wadda need? [13:51:59] :DDD [13:52:34] petan, why has tools been so... [13:52:40] ...overloaded? [13:52:53] YuviPanda: i'm here now :) [13:53:25] YuviPanda: why are you putting it at wikistream-tools as well? [13:53:40] Cyberpower678 when [13:53:48] A lot of the time. [13:54:03] my guess is NFS, but if you don't tell me exact time I can [13:54:08] 't tell you... [13:54:48] Webserver-01 seems to be constantly running on a high load. [13:54:57] http://ganglia.wmflabs.org/latest/?c=tools&h=tools-webserver-01&m=load_one&r=hour&s=by%20name&hc=4&mc=2 [13:56:17] win 1 [13:56:26] apologies :) [14:00:10] Cyberpower678: Yeah, it's time I add a new server and spread the load. [14:00:22] Cyberpower678: I'm about to do that and two compute nodes today. [14:00:34] Coren, :-))))) [14:00:42] !b 53153 [14:00:42] https://bugzilla.wikimedia.org/53153 [14:00:52] ^ also... &(append|prepend)Text option for index.php action=edit [14:00:52] Coren, what's the status of deleted edits? [14:01:01] Cyberpower678: deleted? [14:01:14] T13, ? [14:01:26] Cyberpower678: It's in Ashar's hand. IIRC, he returns from vacation next week. [14:01:35] Asher* [14:01:41] Deleted edits are deleted, as in that is their status? [14:02:31] T13, deleted revisions? [14:02:52] edsu: heya! [14:03:07] edsu: this is running on toollabs, so you won't have to deal with adminstering your own labs instance [14:03:11] !ping del [14:03:12] Successfully removed ping [14:03:15] !ping is pong [14:03:16] Key was added [14:03:23] YuviPanda, ? [14:03:32] Cyberpower678: Deleted revisions are deleted, as in that is their status? <-- better? :p [14:03:39] !ping del [14:03:40] Successfully removed ping [14:03:47] !ping is !pong [14:03:47] Key was added [14:03:53] :p [14:04:08] edsu: plus before this, we couldn't actually do this right, since we all only did... CGI [14:04:19] :) I'm trying to change that, and this was a really good proof of concept, I guess [14:04:40] Coren: did you see http://wikistream-tools.wmflabs.org/? it is running on tools :) [14:04:57] proof of concept, don't bite my head off about giving tools their own domains yet :) [14:04:58] Neat! [14:05:03] T13, I don't get what you mean. An admin can delete an edit or a page. They are marked as deleted in the database. [14:05:15] They are stored in the archive table then. [14:05:46] Cyberpower678: Apparently over your head. :) [14:05:49] * T13 backs away. [14:06:09] Coren: I guess we'll need to replace tools-webproxy with this once it matures. [14:06:15] T13, ?????? [14:08:31] Coren: I was talking to scfe_de about it yesterday a fair bit, and am thinking of relying on identd for auth. I'll write it up a bit more later. [14:09:38] ... that'd work, but only on projects where you can generally rely on the roots (i.e.: not those where project admins are not vetted) [14:10:41] Coren: yeah, this is for the tools specific parts [14:11:05] Coren: there'll be a proxy API, and for general labs it'll be a wikitech interface, and for tools... I'm still trying to figure thatr out [14:12:42] YuviPanda: who deals with admining the labs instance? :) [14:19:35] edsu: That depends on the project; every project has its admins. [14:23:27] Coren: can you help me with puppet? [14:23:38] aude: Probably. What's up? [14:23:57] for wikidata instances (not tool labs) i am getting an error [14:23:58] Error 400 on SERVER: Could not find class ldap::role::client::labs [14:24:11] think i saw that before and fixed it somehow, but don't remember how [14:24:24] do you know about that? [14:24:47] think i'm missing a checkbox on the configuration page [14:25:18] I don't think I've seen that particular error before; I can look into it in a little bit once I'm done setting up my new instances? [14:25:23] ok [14:25:29] i think andrew helped me with it or might know [14:25:37] i can ask him when he's around [14:25:57] Coren: so who is admining this instance? YuviPanda? [14:26:09] edsu: which instance? [14:26:11] I'm probably going to be able to look into it before he gets here, but if he solves your problem before I do tell him I want to know for future reference. :-) [14:26:17] Coren: ok [14:26:42] edsu: oh, http://wikistream-tools.wmflabs.org/ [14:26:44] aude: http://wikistream-tools.wmflabs.org/ [14:26:50] interesting... not you? [14:27:08] aude: i started http://wikistream.wmflabs.org/ :) [14:27:14] yeah, i know [14:27:24] looks like YuviPanda setup the one on toollabs [14:27:37] he's doing stuff with redis there [14:28:28] oh, ok ... wikistream doesn't use redis at all anymore, but maybe he has something planned [14:28:37] oh, really? [14:28:51] aude: yeah, it just streams things and has no memory at all [14:28:56] cool :) [14:29:29] on tool labs, basically we have support from coren etc. on maintaining the overall systems [14:29:47] but then it's less flexible, sometimes or takes more time to get things done [14:29:59] i'm confused doesn't anyone maintain wmflabs.org in general? [14:30:00] it does have access to the "grid" and more resources [14:30:20] edsu: there's labs (all the instances like yours and the wikidata ones) [14:30:34] those need to be self-maintained [14:30:44] tool labs is a project within labs [14:30:57] wo tool labs will monitor services and make sure they are up, etc? [14:31:00] but is special, with staff support and shared environment [14:31:01] s/wo/so/ [14:31:09] edsu: pretty much, yes [14:31:19] so you don't have to setup stuff like redis [14:31:33] setting up redis is trivial, making sure it is still running is work :) [14:31:38] but then for some of my stuff, i want postgresql and postgis (not available yet on toollabs) [14:31:55] so i have to wait or have my own labs project or work elsewhere for that [14:31:56] i see [14:32:11] and i can also help maybe to setup some stuff on tool labs via puppet/gerriot [14:32:14] gerrit [14:32:18] honestly i like isolated environments where things can exist independent of each other [14:32:23] edsu: sure [14:32:41] as long as they aren't so independent that nobody watches them :) [14:32:46] with your own labs instances you can still add people to the project group [14:32:57] well, i think nobody watches them [14:33:06] except might notice if labs as a whole goes down [14:33:13] even the ones in tool labs too it sounds like? [14:33:27] tool labs has more support and monitoring [14:33:30] i guess that's one advantage to a shared environment, more people squawk when things go down [14:33:31] aude: That's correct; labs projects are not supported by ops unless specified (like tools) [14:33:46] mysql is a separate service (separate dedicated boxes) [14:33:50] Although labs /itself/ is considered a production system. [14:33:55] and i assume (likely) is monitored [14:34:00] Coren: yep [14:34:07] Coren: but tools is admin'd just like any other wmflabs project right? [14:34:11] aude: You mean database replication? Yes. [14:34:15] edsu: yes and no [14:34:30] Coren: yes and my project databases [14:34:33] user databases on labs [14:34:37] Coren: you just have a group of admins? [14:34:42] edsu: Not really; unless most other projects tools has staff admins and is officially supported by ops. [14:34:51] on tool labs there is access to replication [14:35:02] oh, that's a big perk [14:35:03] so full admins might have access to private data, so it's more restricted [14:35:18] for general users, they access a subset of the database sans private data [14:35:40] i see, yeah that is a big difference [14:36:06] having access to replicated databases is a big plus [14:36:14] for many tools [14:36:35] (The difference might seem subtle, but there is a big difference with staff project admins and project admins that happen to work for the WMF) [14:36:50] yep [14:36:51] s/with/between/ [14:37:24] there are other labs projects like deployment-prep (e.g. http://en.wikipedia.beta.wmflabs.org/) [14:37:30] that are special [14:37:34] most are not special like that [14:38:12] (looks like beta is testing https) [14:38:15] Right, the general rule is "you admin your project and support it"; it's just that the WMF also uses labs for some projects /it/ supports. [14:38:22] yep [14:45:37] personally I don't mind being admin on wikistream, i'll just have to automate monitoring it [14:46:02] sure, that's fine [14:46:18] it would be nice to make it less of a labs service, except for development, and have it running in "production" [14:46:23] if that concept exists [14:46:33] stream.wikipedia.org would be kinda cool :-) [14:46:46] but i have no idea how to make that happen [14:46:57] or even try to make it happen [14:47:06] if i knew, we'd have osm map tiles already [14:47:14] heh :) [14:47:27] need to find someone in ops to help and get a server or place for it to run [14:47:32] aude: so do you have an osm project in labs? [14:47:48] we have a maps project, but i've not had lots of time for it [14:47:55] and it's not powerful enough [14:47:59] aude: busy w/ wikidata i imagine? [14:48:10] aude: nice work on that btw, it's awesome :) [14:48:14] e.g. if i import all the entire osm data and make tiles, people will scream at me [14:48:17] it will break labs [14:48:31] sometimes that's a good way to move forward :) [14:48:42] i could have tiles maybe for just the united states and that would be okay for development [14:48:49] sometimes it's a good way to make enemies too i guess [14:48:53] heh [14:49:28] tool labs helps bring some dedicated systems for stuff like osm database (but it's not there yet) [14:49:47] and yes, wikidata is busy but we also need maps :) [14:55:14] Coren: so is there an issue with the /data/project space on labs? [14:59:00] * Coren catches up. [14:59:31] define "an issue"? [14:59:31] Coren: i can't cd into /data/project [14:59:37] it freezes [14:59:46] (not toollabs, just labs generally) [14:59:55] I need more info than that; what instance, etc? [15:00:17] i am on wikidata-test-multi at the moment [15:00:22] can see if it happens on others [15:00:41] aude: Looks like gluster is having "fun" again then. [15:00:48] yeah [15:00:58] on other instances, it's slow or not working either [15:01:08] wikidata-testrepo [15:01:09] aude: It's working on bastion [15:01:16] oh, interesting [15:01:22] It's project-local. Lemme go beat up gluster [15:02:57] * Coren mutters really unkind things about gluster. [15:03:03] heh [15:05:44] aude: Should be back. [15:05:48] trying [15:06:12] yes, it works [15:06:14] thanks! [15:07:14] :> [15:07:15] https://tools.wmflabs.org/oauth-hello-world/ is giving me a 500 :/ [15:07:53] legoktm: is it owned by the tool user? [15:08:20] valhallasw: i'm not sure, it's anomie's tool (i think) [15:08:24] ah, ok [15:08:49] hmm [15:09:37] anomie: i'm mainly interested in looking at the source code if that's posted somewhere, i remember there was a link on the tool but I never clicked the link [15:10:03] legoktm: /data/project/oauth-hello-world/public_html/index.php ;-) [15:10:23] heh [15:12:29] It was working yesterday, and I haven't touched it since. Nothing is showing up in the logs I have access to. Ownership on everything looks correct. Coren, petan, YuviPanda? Any of you around and able to help? [15:12:49] anomie? [15:12:54] hi [15:13:02] what exactly should I do :P [15:13:19] petan: https://tools.wmflabs.org/oauth-hello-world/ is returning 500 errors, do you have access to more logs than are in the tool directory? [15:13:33] hold on [15:15:45] tbh I have no idea, it seems that scfs did some things with apache... logs are weird [15:15:56] there are some logs with scfc in name o.O [15:16:19] !toolsadmin [15:16:20] https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/Documentation/Admin [15:16:50] something likely changed but I don't know what it is... [15:17:04] I can see id: local--: no such user in error.log of apache [15:17:11] but that doesn't really say much [15:18:37] people should start documenting their changes otherwise tools are never going to work... [15:18:47] @notify scfc_de [15:18:47] I'll let you know when I see scfc_de around here [15:19:12] Coren: any idea what changed in apache configuration past days? logs seem to be totally useless now [15:19:35] petan: Nothing should have. What do you mean "useless"? [15:19:54] some random errors produced by "id" command with no trace of source [15:20:06] I suppose it's produced by one of error splitters [15:20:15] There is /usr/local/bin/php-wrapper that is splitting php error logs [15:20:26] I suppose that is what produces these `id` errors [15:20:51] petan: It does, AFAIK that has almost always been there. I'll look into it in a little bit, I'm setting up the new instances atm. [15:21:20] Coren: useless as they don't contain useful information, in fact they contains almost no information, server display 500 error, error log is sent to some splitter and likely end up in /dev/null, we should have soome global error log like for access logs [15:21:27] so that I can easily grep any tool from 1 place [15:21:41] without having to search multiple filesystem for corresponding log files [15:21:58] I'll look into it in a little bit. [15:22:02] ok [15:22:04] anomie ^ [15:22:28] btw, Coren exec-01 is still down [15:22:53] petan: I am aware. Something odd with Openstack I will diagnose with Ryan later today. [15:23:00] ok [15:23:18] aude: edsu wikistream-tools.wmflabs.org is *not* another instance. It is running on toollabs, on the grid engine [15:23:27] http://tools.wmflabs.org/?status C-f 'wikistream' [15:23:45] aude: edsu I was using it as a proof of concept to show how you can run web servers on toollabs grid engine, which wasn't possible before :) [15:23:54] YuviPanda: cool [15:24:01] and this also means you can easily run websockets, etc from toollabs, which also wasn't possible before [15:24:09] Internal Server Error [15:24:12] The server encountered an internal error or misconfiguration and was unable to complete your request. [15:24:19] Please contact the server administrator, mpelletier@wikimedia.org and inform them of the time the error occurred, and anything you might have done that may have caused the error. [15:24:22] More information about this error may be available in the server error log. [15:24:23] Bleh. [15:24:24] Additionally, a 500 Internal Server Error error was encountered while trying to use an ErrorDocument to handle the request. [15:24:27] Apache/2.2.22 (Ubuntu) Server at tools.wmflabs.org Port 80 [15:24:31] oops [15:24:32] That's almost certainly something I did wrong. [15:25:17] * Coren headdesks. [15:25:21] Yep. My fault. [15:26:01] Sorry, just broke this. Should work again now? [15:26:19] Coren, thanks. [15:26:50] Coren, thanks. [15:26:51] Coren, thanks. [15:26:52] Coren, thanks. [15:26:52] Coren, thanks. [15:26:53] Coren, thanks. [15:26:54] Coren, thanks. [15:26:55] Coren, thanks. [15:26:56] Coren, thanks. [15:26:57] Coren, thanks. [15:27:01] Coren, thanks. [15:27:02] Coren, thanks. [15:27:04] Coren, thanks. [15:27:07] Coren, thanks. [15:27:08] Coren, thanks. [15:27:10] Coren, thanks. [15:27:17] Coren, thanks. [15:27:19] Coren, thanks. [15:27:20] Coren, thanks. [15:27:22] Coren, thanks. [15:27:25] Coren, thanks. [15:27:27] Coren, thanks. [15:27:31] Coren, thanks. [15:27:33] Coren, thanks. [15:27:35] Coren, thanks. [15:27:37] Coren, thanks. [15:27:39] Coren, thanks. [15:27:41] Coren, thanks. [15:27:43] * YuviPanda staples Cyberpower678 in place [15:27:47] Coren, thanks. [15:27:49] Coren, thanks. [15:27:51] Coren, thanks. [15:27:53] Coren, thanks. [15:27:55] Coren, thanks. [15:27:57] Coren, thanks. [15:28:01] Coren, thanks. [15:28:03] Coren, thanks. [15:28:04] Hm. Forgot to ask ops in this channel. [15:28:09] Cyberpower678: Stop this now, please. [15:28:34] thanks coren! [15:28:36] :) [15:28:49] Sorry [15:28:51] edsu: btw, https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/Help has a more detailed description on what toollabs is. You get a grid that is monitored, plus other tools that are monitored - plus access to the dbs [15:28:56] cat on keyboard [15:28:57] !searchlogs [15:29:00] yeah, thanks Coren! :) [15:29:07] !ping [15:29:07] !pong [15:29:14] !logsearch [15:29:15] http://bots.wmflabs.org/~wm-bot/searchlog [15:29:24] Cyberpower678 what the hell was that [15:29:36] petan, I'm not sure. [15:29:39] anomie: no license header on the file? [15:29:51] * YuviPanda hides his license headerless files from legoktm [15:29:54] Coren: you can always type @q Cyberpower678 [15:29:56] @q Cyberpower678 [15:29:57] isn't a LICENSE or COPYING file enough? [15:30:01] @unq Cyberpower678 [15:30:13] XChat just flooded my screen with Coren, thanks. [15:30:13] YuviPanda: there isnt one of those either. [15:30:21] Nothing responded. [15:30:25] legoktm: aaah, okay. then poke, poke! [15:30:33] legoktm: I should fix that. Hmm, which to choose? [15:30:37] @trustadd .*@wikipedia/Coren admin [15:30:37] Successfuly added .*@wikipedia/Coren [15:30:40] anomie: WTFPL! [15:30:44] anomie: PD :) [15:32:39] legoktm: also, http://www.rightscale.com/blog/cloud-cost-analysis/cloud-cost-analysis-how-much-could-wikipedia-save-cloud [15:32:46] apparently all of wikimedia runs from labs :D [15:33:34] heh. [15:33:53] Coren, went back to the logs. Sorry. XChat stopped responding and acted wierd. [15:34:19] Cyberpower678: That it deed. [15:34:38] Yeah, that "analysis" was downright laughable. [15:34:54] Bye everyone. [15:35:37] Coren: would be fun seeing their correction :D [15:36:53] hey nerus [15:37:01] YuviPanda: You presume a lot. Anyone who would take a huge infrastructure of open-soure posixy things that the wmf farm and even *propose* a Windows cloud to replace it is obviously either completely incompetent, quite stupid, or has a service to sell to the gullible. [15:37:02] hey YuviPanda [15:38:08] nerus: wikistream-tools.wmflabs.org [15:38:17] nerus: proof of concept of nodejs + websockets working from toollabs [15:38:26] Naice! [15:38:53] YuviPanda: doesn't work for wikidata? [15:38:58] also, wsgi yet? [15:38:59] what is it streaming? [15:39:00] YuviPanda:WOW what a speed [15:39:07] legoktm: ask edsu [15:39:07] :P [15:39:09] he wrote it [15:39:14] nerus: recent changes? [15:39:28] Steinsplitter: it's edsu's work, I am just getting it to run on toollabs :) [15:39:32] hmmm, that much :O [15:39:42] nice tool [15:39:44] :) [15:39:49] so, i was checking out how virtualenv was setup [15:39:59] i think similar stuff is possible with rvm and stuff [15:40:14] nerus: indeed! [15:40:14] but wouldn't have worked with just apache alone. [15:40:23] nerus: it would, because CGI doesn't care [15:40:31] I see [15:40:40] nerus: we could write our stuff in C, if we want :P [15:40:45] I see ;) [15:41:03] nerus: the only problem with ruby is the fact that so little of us here use / know ruby [15:41:16] i know, you have told me [15:42:02] legoktm: so, no direct support for uwsgi. I think Coren's stopped the uwsgi tyrant work. [15:42:09] i was just checking out as 'possibility' than doing it per se [15:42:17] like curiousity :) [15:42:18] nerus: ah, of course! :) [15:42:28] nerus: we should build something with Ruby soon though. I need to learn it :) [15:42:30] YuviPanda: I have, in the expectation that your more generic solution would solve all ills. :-) [15:43:03] Coren: indeed! I think it already does, as seen by wikistream :) Just needs a new nginx package [15:43:09] which I need to find someone to do ;( [15:43:41] what nginx package? ;) [15:44:05] Coren: i can haz a tools-proxy instance to play with? Since I can't really access ports on tools instances from other labs instances (firewalled) [15:44:38] nerus: so anything we do is to be on puppet, and all software as such debianized - so I'd need a nginx+lua package that's not ancient (as the ones in ubuntu right now are) [15:45:07] how soon do you need it :P [15:45:58] nerus: i can let tools be built using any language as soon as I have that package :) [15:46:06] nerus: I've 4 patches on gerrit waiting for such a package [15:46:48] YuviPanda: Sure. Sometime today? [15:47:18] Coren: sure! how hard would it be to build it? [15:47:23] Coren: I was able to hand-compile it in under a minute [15:47:48] YuviPanda: you could also check if there is a ppa available [15:47:50] YuviPanda: Not very, I expect. [15:48:07] valhallasw: I was told to not use PPAs. [15:48:30] valhallasw: plus the PPAs I checked didn't have the options I wanted (nginx+lua). [15:49:16] YuviPanda: get the debian/rules from the PPA and use that to roll your own? [15:49:18] YuviPanda: if I recall correctly, i have to write a spec file and build debs for different architectures [15:49:31] i have done that but that was like 4 years back [15:49:41] valhallasw: I would, but am lazy - plus there are plenty of other people who can ;) [15:49:57] nerus: I think there's something like git deb or something now, that lets you put that on a git repo or... something? [15:50:09] I've been stuck in this horror called OS X for a year now so do not know :( [15:51:51] YuviPanda: if it can take a couple of days or so I can do it :) [15:52:06] nerus: i'll poke you if I can't get Coren to do it :) [15:52:15] hey scfc_de [15:53:10] YuviPanda: Hello. [15:53:48] scfc_de: Coren has offered to package up nginx for me today, so yay :) [15:53:59] ... wait what? [15:54:06] scfc_de: I'm having puppet troubles with the current test instance though, so can you help take a look? [15:54:17] Coren: oh, wait - you were offering to get me tools-proxy? [15:54:19] and not package nginx? [15:54:24] * Coren nods. [15:54:29] Coren: aah, okay! [15:54:57] Coren: okay, but I will need sudo on tools-proxy for now, then. We can destroy and re-create it later :) [15:55:11] YuviPanda: That means less work for me?! [15:55:34] scfc_de: well, apparently Coren did *not* offer to package nginx, so I still need to get that done [15:55:35] YuviPanda: around? [15:55:41] looks like :P [15:55:43] ... what a stupid question on my part ... [15:56:13] addshore: 'sup? [15:56:24] YuviPanda: I need a few more minutes to be "fully" online, I'll ping you then. [15:56:30] scfc_de: hehe, okay! :) [15:56:43] pm ;p [16:00:48] [bz] (8NEW - created by: 2Aude, priority: 4Unprioritized - 6normal) [Bug 53158] setup devrepo on new, larger instance - https://bugzilla.wikimedia.org/show_bug.cgi?id=53158 [16:01:16] reason to use puppet [16:26:20] Coren: tools-exec01 is having what problem in openstack? [16:27:12] Ryan_Lane: if you haven't seen it yet, wikistream-tools.wmflabs.org is the proxy proxying to a tool that's running on a random port in grid engine :) [16:27:13] Ryan_Lane: It appears to... not exist. Since it didn't restart, I tried going to its console and... " domain with matching name 'i-00000604'" [16:27:25] not found. [16:27:28] yeah [16:27:33] because that isn't it's id :) [16:27:34] YuviPanda: Ready to go. [16:27:37] *its [16:28:02] OS_TENANT_NAME=tools nova list [16:28:07] scfc_de: let me add you to the project [16:28:19] ...? According to the instance list it is; and I tried the same with i-0000064c and /that/ worked. [16:28:20] OS_TENANT_NAME=tools nova reboot tools-exec-01 [16:28:24] anomie: Do you still have server errors? [16:28:38] Coren: this is on wikitech, or on virt0? [16:28:44] scfc_de: No, working now [16:29:27] Ryan_Lane: I mean with virsh on virt10 [16:29:47] scfc_de: what's your wikitech name? [16:29:54] oh [16:30:00] Coren: don't use reboot like that [16:30:17] anomie: Okay. For future reference, error logs are and have always been accessible only to roots on the webserver instances in /var/log/apache2/error.log. [16:30:19] always let openstack do the reboot [16:30:23] Ryan_Lane: heh. I was confusing. It didn't reboot (on wikitech) so I wanted to go see its console (on virt10) [16:30:41] Ryan_Lane: But on virt10, the instance doesn't seem to exist at all. [16:30:41] YuviPanda: "Tim Landscheidt" [16:30:48] it's listed as running [16:31:08] via virsh list on virt10 [16:31:13] Ryan_Lane: I know, which is why I'm confused. :-) I can connect to the other node's consoles, but not that one. [16:31:41] you guys should see http://www.rightscale.com/blog/cloud-cost-analysis/cloud-cost-analysis-how-much-could-wikipedia-save-cloud :) [16:31:44] ... no it's not. I don't see 604 in virsh list [16:31:49] scfc_de: added [16:32:02] oh [16:32:06] it is not running [16:32:13] scfc_de: instance proxy-project on project project-proxy [16:32:19] scfc_de: apt-get install nginx-extras fails [16:32:20] it's not on virt10 [16:32:48] it's on virt9 [16:32:54] OS_TENANT_NAME=tools nova show tools-exec-01 [16:33:05] Oh! Silly camper. I had presumed that it was with every other instance. :-) [16:33:12] But it was clearly a coincidence. [16:33:12] scfc_de: I've an nginx instance installed on /usr/local/nginx. [16:33:23] Ignore me, then. :-) [16:33:28] scfc_de: so, 1. any idea why I can't install that? 2. can you help make a deb? :D [16:33:33] there's no guarantee as to where an instance is going to land [16:33:40] YuviPanda: Aren't there debs already for nginx? [16:33:42] you always have yo check :) [16:33:45] scfc_de: too old [16:34:05] Ryan_Lane: I though it was some sort of RR so that if X and X+2 are one one, X+1 also must be. [16:34:09] scfc_de: I need nginx + the lua module, and the ones that are anywhere don't have a new-enough version of the lua module [16:34:20] YuviPanda: :-( [16:34:39] scfc_de: building it was rather trivial (get archives, configure, make, install), but unsure how to debianize [16:35:09] YuviPanda: Is the version you're looking for packaged in later Ubuntu releases? [16:35:24] scfc_de: later versions of ubuntu have a new enough nginx, but not a new enough lua module [16:35:49] Coren: it's based on number of instances running on a host [16:35:52] YuviPanda: That sounds like work. [16:36:03] scfc_de: which? [16:36:04] so, based on deletions and such... [16:36:31] Ryan_Lane: Yeah, yeah, I get it now. :-) [16:37:20] YuviPanda: Not just copying the work of someone else :-). Can the new nginx deb be installed or are there conflicts? (I. e., do we need to package one or two debs.) [16:37:23] Ryan_Lane: And my stuck instance was the same issue: fail boot makes grub force interactive and wait for return. We *really* gotta disable recordfail in our images. [16:37:39] Coren: I'll puppetize that work this week [16:37:44] the image creation work [16:37:45] scfc_de: not sure, how do I check? :D [16:37:52] Coren: I can show you how to make new ones, though [16:37:53] scfc_de: currently just running apt-get install gives me an error [16:37:56] if you'd like to make that change [16:38:03] YuviPanda: *I* don't know :-). Let me see. [16:38:09] :D [16:38:44] Ryan_Lane: It's not so urgent that it can't wait until the all-staff when we get real hacking time; but we have to make sure to not forget it. Should I open a bz? [16:40:24] YuviPanda: Hmmm. Just installing aptitude: "Errors were encountered while processing: nginx-extras": "ln: failed to create symbolic link `/etc/nginx/sites-enabled/default': No such file or directory". *Argl*. [16:40:35] scfc_de: yup! [16:42:10] YuviPanda: "remove nginx-extras" => "install nginx-extras" => Same error. Can we start with another clean instance? [16:42:32] scfc_de: this has unpuppetized stuff, so we can create another instance - but don't delete this :) [16:42:35] YuviPanda: Is this for Tools only or for the Labs proxy? [16:42:44] YuviPanda: No, no, I don't intend to. [16:42:46] scfc_de: both, really. [16:42:56] Coren: yeah, open a bz [16:43:11] scfc_de: there'll be tools-proxy, which will be tools-only and replace tools-webproxy. there'll also be project-proxy-proxy-project which is labswide :) [16:43:57] YuviPanda: Okay. Should I create an instance project-proxy-2 as a puppetmaster::self? [16:44:11] yes [16:44:55] YuviPanda: Security group "proxy"? [16:45:32] scfc_de: hmm, not sure [16:45:42] looking [16:45:46] it needs port 80 open to everyone [16:45:56] scfc_de: yeah, that sounds about right [16:45:59] 'proxy' [16:46:21] [bz] (8NEW - created by: 2Marc A. Pelletier, priority: 4Unprioritized - 6normal) [Bug 53159] Disable recordfail on labs instances - https://bugzilla.wikimedia.org/show_bug.cgi?id=53159 [16:47:10] YuviPanda: "channel 0: open failed: administratively prohibited: open failed" => I have to enable NFS and wait for a Puppet run + reboot? [16:47:20] I think so [16:47:32] though I didn't have that [16:47:34] Coren: Ryan_Lane ^? [16:47:47] YuviPanda: You can "ssh proxy-project-2.pmtpa.wmflabs"? [16:48:16] nope [16:48:31] Okay, then I'll enable NFS, wait 30 mins and reboot. [16:48:35] security group changes take a little bit [16:48:44] rebooting should 'just work' no? [16:48:46] for puppet [16:51:08] YuviPanda: So we need nginx/1.5.3 and what Lua module? [16:51:24] scfc_de: 0.8.6 lua [16:51:34] scfc_de: though anything past 0.6 is okay, but yeah, 0.8.6 :) [16:51:44] "released on 6 August 2013" :-). [16:51:45] at least nginx 1.4 [16:52:06] Coren or Ryan_Lane can you please increase the quota for wikidata project? we need more cores in order to create a new instance [16:52:06] but yeah [16:52:27] * aude needs to rebuild one of our instances on a new, bigger instance [16:52:28] aude: Sure thing. Hang on. [16:52:30] thanks [16:53:35] scfc_de: http://wiki.nginx.org/HttpLuaModule#Installation [16:53:35] aude: How many more do you need? [16:53:40] Ryan_Lane: What's the WMF policy regarding packaging later versions of Ubuntu packages? Naming "wmf-nginx" and "Conflicts: nginx"? [16:53:55] scfc_de: it should depend on luajit (not liblua) [16:54:12] Coren: at least enough to make another medium instance [16:54:29] then i'll delete the current instance after i rebuild [16:55:02] aude: Quota bumped. [16:55:11] thanks [16:55:46] hmmm it still fails [16:56:06] RAM: 75776/76800 [16:56:10] maybe that's the problem [16:57:38] YuviPanda: Looking at http://packages.ubuntu.com/search?keywords=nginx+lua, could it be that the Lua module is part of nginx? Let's see. [16:57:53] scfc_de: it's a 3rd party module, so... [16:57:57] doubt it? [16:58:04] aude: You now have moar ram!!1! [16:58:19] scfc_de: though possible, yeah. I looked at 13.04, nginx-extras [16:58:27] thanks [17:00:00] YuviPanda: http://packages.ubuntu.com/raring/nginx-extras says it depends on liblua5.1-0? [17:00:08] YuviPanda: NB: raring. [17:00:23] Coren: it worked :) [17:00:25] scfc_de: see, they're interchangeable. liblua and luajit. I want to build with luajit since it is orders of magnitude faster. [17:00:36] scfc_de: also, brb - heading to a doc now (something painful in my ear) [17:01:54] YuviPanda: Okay, I'll dig deeper. Get well soon! [17:02:15] scfc_de: but yeah, getting it with *anything* is a good start, but for 'prod' use we'll want luajit [17:03:02] YuviPanda: Oh, nginx-extras is nginx-extras_1.1.19-2wmf1_amd64.deb -- so some WMF's been packaging nginx already. [17:03:13] scfc_de: yeah, and it's super, super old :P [17:03:36] okay, really oging to doc now [17:17:04] Coren: need more help [17:17:26] in one of our instances, our root partition is 100% full [17:17:46] i am wondering which directories are stored there? [17:17:57] and trying to find some stuff to delete [17:18:07] (already deleted some log files) [17:29:51] aude: Most likely in /var (/var/log, /var/cache). You might want to (cd /var; du -sh *) to get a good idea where [17:30:36] everything in /var is in the root partition? [17:30:51] /var/www appears to be in the larger 40 gb partition on /mnt [17:31:34] aude: What does "mount" say? [17:34:02] http://dpaste.com/1352089/ [17:38:16] aude: Then everything should be in /. Is /var/www a symlink? [17:39:09] scfc_de: yes it is [17:39:19] at least the stuff i deleted [17:40:01] What was the main culprit? [17:40:28] i don't know yet [17:40:38] but probably our ibdata1 (mysql) has become too large [17:40:59] i wonder what the recommended practice is for where /how to put mysql storage [17:41:03] for labs [17:46:49] yep, we have 10 gb root [17:46:55] aude: On tools-db, /var/lib/mysql is a symlink to /mnt/mysql. [17:47:11] 2.7 in /usr 2.9 in /var and 3.0 in /srv [17:47:26] i think we can do that when we rebuild to make the symlink [17:47:31] and use puppet [17:49:08] i don't think there is much i can delete to "fix" this [17:50:25] aude: You can shut down the mysql server, move the data to /mnt (if you mount sdb there) and symlink the old location to the new one. [17:51:29] ok [17:51:32] i can try [17:51:37] it's already shut down [17:52:34] *Argl*. It's project-proxy-3.pmtpa.wmflabs, not proxy-project-3.pmtpa.wmflabs. I'm an idiot. FTR: "channel 0: open failed: administratively prohibited: open failed" means bastion cannot find the hostname. *Argl*^2. [17:52:38] scfc_de, petan: for future reference, please test installing a package on an exec host before you put in in puppet; if there are conflicts to resolve, the resulting puppet config will get stuck and fail. This allows you to figure out if you have to pin versions, etc, beforehand. [17:53:07] scfc_de: ... that error message is a fail. [17:53:43] * Coren is currently fixing the odd dependency issues around libvips-dev. [17:54:46] Coren: BTW, Doc/Admin says re new instances, run "apt-get update" or something. Does that collide with the "ensure => present/latest" issues? [17:55:14] Coren: I put those on puppet, btw. I just did an apt-cache show to check that the packages existed [17:55:28] I... suppose I can try installing them on toolsbeta [17:55:32] It does; and the update is unavoidable anyways since puppet does one anyways. [17:55:55] scfc_de: back because I couldn't really find a clinic. Going to head to bed and see if I can sleep. [17:55:56] the present/latest issue is a Hard Problem on tools because of the dynamic nature of the environment. [17:56:21] YuviPanda|away: Okay, I'll look around some more now that I found out how to log in :-). [17:56:51] scfc_de: ty! [17:57:08] (I.e.: we normally shouldn't be using latest in a semi-production system, but using present can cause version drift since we don't install all the packages at the same time nor create all instances simultaneously) [17:57:30] Coren: Or I meant "apt-get upgrade". One of those causes Ryan headaches IIRC. [17:57:58] Coren: So you're suggesting using latest in Tools? [17:58:00] Yeah, apt-get upgrade is also both a source of problems and the solution to version drift. Fun times. :-) [17:58:15] no apt stuff causes me headaches :) [17:58:25] except for upgrading between major versions [17:58:55] No, because latest has issues of its own. I don't think I've decided what the best thing is atm; but we'll have to keep an eye out for version drift. [18:00:16] Coren: I'm going to head off now, but can you still get me tools-proxy? :) ty! [18:00:36] YuviPanda|bed: Yeah, will do. [18:01:46] Coren: It would just be nice to have one procedure written down; at the moment, it's not very determinate with what versions we end up with. [18:01:58] (Or collisions.) [18:02:37] scfc_de: Yeah, it's not an easy problem to solve. I'm thinking we might simply want to do present with a regular safe-upgrade [18:06:49] ... I just noticed I spent an age and a half fixing things for libvips-dev in exec environ... why was this put in exec and not just dev? [18:10:07] Coren: Misunderstanding ? [18:11:45] [bz] (8REOPENED - created by: 2Daniel Schwen, priority: 4Unprioritized - 6enhancement) [Bug 52717] libvips-tools, libtiff etc install - https://bugzilla.wikimedia.org/show_bug.cgi?id=52717 [18:21:39] YuviPanda|bed: how could it not be another instance on wikistream? [18:21:46] YuviPanda|bed: of wikistream I mean [18:21:53] oops, just noticed "bed" :-) [18:22:00] nighty-night ... [18:46:01] Coren: it worked to move the mysql data directory [18:46:35] i had to change my.cnf to tell it the new location and edit apparmor mysql config also [18:46:55] thanks for the suggestion! [18:47:16] aude: I think you could have gotten away with a symlink :-). [18:47:22] i tried [18:47:35] Oh. [18:47:36] without the app armor thing, it had insufficient permissions [18:47:50] now the symlink would work but also works to change my.cnf [19:39:46] Is there a way to get fully valid SSL certificates for a Labs instance? [19:40:01] We need to provide SSL, but it seems that the certificates in puppet are not actually valid [19:45:16] FastLizard4: got some sample URL? cert validity is in the eye of the trusted certificate store beholder:) [19:50:08] saper: I'm not sure what you mean. [19:50:56] I'm trying to use certificates::star_wmflabs_org [19:55:03] did you have an instance set up where I can connect to via SSL? no time to play with git now [19:57:55] [bz] (8RESOLVED - created by: 2Chris McMahon, priority: 4Unprioritized - 6normal) [Bug 53080] no login error messages on http://commons.wikipedia.beta.wmflabs.org/ - https://bugzilla.wikimedia.org/show_bug.cgi?id=53080 [20:00:52] saper: https://acc.wmflabs.org [20:01:17] I've checked the apache config a few times, pretty sure that's not where the problem is [20:23:40] what's up with the webserver no longer executing Python scripts as Python? [20:27:40] Anyone there? [20:29:53] Nettrom: You mean Tools? Which URL? [20:30:32] scfc_de: http://tools.wmflabs.org/suggestbot/quality-metadata.py [20:31:02] also http://tools.wmflabs.org/suggestbot/link-recommender.py [20:31:16] worked up until earlier today, I think? [20:31:33] Also, why does labs only use 1024-bit RSA keys :P [20:32:19] Nettrom: Coren set up a new tools-webserver-03 today, and it serves those bots now. Let me check where the config is different. [20:33:33] Hmmm. Can't log into tools-webserver-03. [20:33:44] scfc_de: Oh? Odd. [20:33:52] * Coren goes check that as well. [20:33:59] Perhaps something I forgot to puppetize. [20:34:13] Coren: Through bastion; ssh'ing to -login and then ssh tools-webserver-03 works. [20:34:34] scfc_de: Ah! That's normal. [20:35:32] It worked for me for tools-webserver-01. [20:35:43] That, otoh, isn't. :-) [20:35:48] Can anyone answer a question about Labs SSL certificates? [20:35:57] FastLizard4: Yes. [20:36:28] Coren: Is there a way to get SSL certificates for HTTPS that are signed by an authority recognized by default in at least most modern browsers/operating systems? [20:36:44] The default SSL certs in puppet seem to be issued by the Labs CA, which is of course not recognized [20:37:17] I'm in the process of beginning to move a project from Toolserver to Labs, but SSL is a requirement and not having valid SSL certs blocks the move [20:37:35] FastLizard4: ... on a non-tools project? [20:37:44] Yes [20:37:58] Specifically, [[WP:ACC]] [20:37:58] https://ko.wikipedia.org/wiki/WP:ACC [20:38:09] ...On the English Wikipedia, of course (silly wm-bot) [20:39:06] /etc/apache2/sites-enabled/100-webserver.conf and /etc/suphp/suphp.conf are identical; hmmm. [20:40:09] FastLizard4: That's... an interesting question. Normally, we wouldn't allow Wikimedia-purchased certificates to live on volunteer projects without anyone possibly having root being at least under NDA (cuz the key); YuviPanda is, however, working on an nginx proxy for general labs use that can do SSL and which ACC could then use. [20:40:30] Last I heard, it was almost ready for general use. [20:40:42] (I.e. days aways) [20:40:50] Ahh [20:41:10] So the problem might end up being entirely moot shortly unless the self-signed certificate breaks testing. [20:41:43] Unfortunately, in a few days I'm starting the first of two back-to-back vacation trips lasting about three weeks, and I'm currently the only person with the capabilities to perform the move :P [20:42:09] We're also at the point where we're ready to go live with the move :P [20:43:19] (That's the reason why I'm in a bit of a rush; this move has been a while in the making and I wanted to get it done before I went on vacation) [20:44:17] Hm. Annoyingly enough, due to timezones, YuviPanda|bed is asleep currently. I can certainly tell him that you'd be a very good first victim of the new proxy. You want to open a ticket asking for the certificate ASAP though, it can take a day or two to get that. [20:45:26] Heh, okay [20:45:30] Where should I file the ticket? [20:48:41] This is awfully short notice though. I hope we can squeeze everything in time. Open a bugzilla in labs/infrastructure, I'll do the follow up. [20:49:03] (afk for a few) [20:49:46] Coren: I don't want you to rush too :P [20:49:53] I'd rather things be done right than quickly :P [20:50:15] Value wikitech was stored into default_link_wiki to config [20:50:16] Coren: If it can't be done before I go on vacation, I think we can live with being on the Toolserver for a few more weeks :P [20:50:23] But if they can be done quickly and right, that is good :p [20:50:49] mods-available/suphp.conf seems to be lacking suPHP_AddHandler application/x-suphp-cgi. Let's try that. [20:50:52] @link [[Xxx]] [20:50:52] https://enwp.org/Xxx [20:51:11] Coren: But yeah, I'll go file that ticket and send you a link when I've submitted it. Thanks for your help! :) [20:51:25] What's the interwiki prefix for wikitech? [20:52:39] Needs a restart to take effect. [20:53:11] Nettrom: Does it work for you now? [20:56:36] What's the interwiki prefix for wikitech? Or isn't there one? [20:57:03] Technical_13: "wikitech" :-). [21:04:42] [bz] (8NEW - created by: 2Andrew "FastLizard4" Adams, priority: 4Unprioritized - 6normal) [Bug 53175] SSL certificate for ACC project (acc.wmflabs.org) - https://bugzilla.wikimedia.org/show_bug.cgi?id=53175 [21:04:52] Coren: https://bugzilla.wikimedia.org/show_bug.cgi?id=53175 and thanks again :) [21:11:39] @link [[wikitech:Main page]] [21:11:39] https://enwp.org/Main_page [21:11:46] Hrmm. [21:13:25] Technical_13: It works on the actual enwp :-). [21:14:42] That is a new plugin for wm-bot so still ironing little kinks out. [21:15:56] scfc_de: my test-script is successful now, thanks for getting it fixed quickly! [21:16:17] Technical_13: Does it pull the interwiki table, or are the prefixes hard-coded? [21:16:25] Nettrom: np [21:17:00] Not sure, I had thought it was going to pull from table but don't know how petan actually implemented it. [21:17:27] Technical_13 yes it does [21:17:44] I mean it pull from extracted table which is in file [21:17:46] :P [21:17:56] that one you provided in bugzilla [21:17:59] So wikitech isn't in there. [21:18:02] Oh... [21:18:21] @link [[hmb:Brain]] [21:18:21] https://helpmebot.org.uk/wiki/Brain [21:18:27] Ha.. okay. [21:18:45] @link [[Something:Blah]] [21:18:45] https://enwp.org/Blah [21:18:47] aha [21:18:49] that is a bug [21:18:52] meh [21:18:55] Should probably use wmf table [21:19:12] petan: that is with an invalid default set. [21:19:29] This value can't be stored [21:19:43] Doesn't return currently set... [21:19:47] mhm [21:19:54] But I had set it to wikitech [21:20:03] Which isn't on that table. [21:20:26] Value en was stored into default_link_wiki to config [21:20:32] [[Something:Blah]] [21:20:32] http://enwp.org/Something:Blah [21:20:36] aha [21:25:34] [[wikitech:Nova Resource:Tools]] [21:25:34] http://enwp.org/wikitech:Nova_Resource:Tools [21:25:53] Well, it reaches the target :-). [21:34:30] [bz] (8RESOLVED - created by: 2Aude, priority: 4Unprioritized - 6normal) [Bug 53158] setup devrepo on new, larger instance - https://bugzilla.wikimedia.org/show_bug.cgi?id=53158 [22:11:12] Has there been a recent change preventing the compute cluster from accessing enwiki.labsdb? [22:15:58] Hasteur: There shouldn't be/have been. [22:16:25] Hasteur: Do you know what the failiure is? (I.e. credentials, network, etc) [22:17:24] Hrm... I can access my bot's DB from the tools-login, but when I jsub the job off, it croaks after about 30 seconds complaining it can't get to the mysql db [22:17:43] * Coren checks. [22:18:29] Trivial tests seem to work. [22:18:31] IP forwarding on the new exec nodes? [22:19:01] scfc_de: I just flushed those just in case. Hasteur: try again? [22:19:03] Coren: Did you see my Bugzilla ticket? [22:19:20] Coren: I sent you the gist of the data [22:19:42] FastLizard4: I did. [22:20:03] Coren: Awesome. Thanks again :) [22:21:20] Hrm... The purge of the nat appears to have solved it, or I got assigned to a different compute node [22:21:55] where does qstat report your job is? [22:22:13] task@tools-exec-01.pmtpa.wmfla [22:22:34] Hm. [22:23:04] That's not a new node, so we... oh! I *knew* I did it for the new nodes but I just put -01 back online after an outage and I forgot THAT one. :-) [22:23:14] Wallah! [22:23:35] * Coren files a bug to track adding this to puppet so it doesn't bite us. [22:23:54] Oh, my tool's zipping around just fine, and I'm going to go ahead and delete that secret data I passed to you. [22:24:32] [bz] (8NEW - created by: 2Marc A. Pelletier, priority: 4Unprioritized - 6normal) [Bug 53181] Make the iptables initialization occur at boot time - https://bugzilla.wikimedia.org/show_bug.cgi?id=53181 [22:24:33] Hasteur: Wise move. [22:25:03] Coren: What about the plan to put that in DNS proper? [22:25:33] scfc_de: After NFS. [22:25:42] [bz] (8RESOLVED - created by: 2Tim Landscheidt, priority: 4Unprioritized - 6normal) [Bug 52527] tools-webserver-01 seems to be out of memory from time to time - https://bugzilla.wikimedia.org/show_bug.cgi?id=52527 [22:26:13] Coren: k [22:26:24] scfc_de: With luck, sometime next week. [22:26:51] Thank you Coren [22:27:37] Coren: Without the stalls, NFS has become a very non-visible problem :-). [23:58:54] greg-g and robla thought this sounded like an interesting idea and I need to lear how to use labs so: https://wikitech.wikimedia.org/wiki/New_Project_Request/WMF_Bouncer