[02:56:54] 3Wikimedia Labs / 3tools: Tool Labs: Local webservice should know about HTTPS in original request - 10https://bugzilla.wikimedia.org/53689 (10Liangent) [02:56:54] 3Wikimedia Labs / 3tools: lighttpd redirects URLs of directories without a trailing slash from https to http - 10https://bugzilla.wikimedia.org/64627#c1 (10Liangent) dupe of https://bugzilla.wikimedia.org/show_bug.cgi?id=53689 ? [02:59:31] 3Wikimedia Labs / 3tools: Tool Labs: Local webservice should know about HTTPS in original request - 10https://bugzilla.wikimedia.org/53689#c7 (10Liangent) Seems fixed [03:00:12] 3Wikimedia Labs / 3tools: lighttpd redirects URLs of directories without a trailing slash from https to http - 10https://bugzilla.wikimedia.org/64627#c2 (10Liangent) (In reply to Liangent from comment #1) > dupe of https://bugzilla.wikimedia.org/show_bug.cgi?id=53689 ? seems this bug was accidentially fixed. [03:00:43] 3Wikimedia Labs / 3tools: Tool Labs: Local webservice should know about HTTPS in original request - 10https://bugzilla.wikimedia.org/53689 (10Liangent) 5REO>3RES/FIX [06:15:13] 3Wikimedia Labs / 3tools: lighttpd redirects URLs of directories without a trailing slash from https to http - 10https://bugzilla.wikimedia.org/64627#c3 (10Tim Landscheidt) (In reply to Liangent from comment #2) > > dupe of https://bugzilla.wikimedia.org/show_bug.cgi?id=53689 ? > seems this bug was accident... [06:17:28] 3Wikimedia Labs / 3tools: lighttpd redirects URLs of directories without a trailing slash from https to http - 10https://bugzilla.wikimedia.org/64627#c4 (10Liangent) (In reply to Tim Landscheidt from comment #3) > (In reply to Liangent from comment #2) > > > dupe of https://bugzilla.wikimedia.org/show_bug.cg... [08:44:43] (03PS1) 10Giuseppe Lavagetto: Adding missing classes so that manifests compile. [labs/private] - 10https://gerrit.wikimedia.org/r/130574 [08:59:44] mutante: do you remember the machine name for wikistats? [09:01:16] YuviPanda: wikistats-live since it's in eqiad, was wikistats-01 in pmtpa [09:01:22] aha cool [09:01:54] why are you asking [09:02:01] proxy? [09:02:15] YuviPanda: [09:02:31] mutante: yeah. [09:02:48] mutante: hitting it from tools doesn't work [09:05:18] can tools talk to other labs projects that are not in tools itself? [09:06:50] YuviPanda: do you still need the "upld" group thing? [09:06:56] mutante: yeah. [09:07:14] so it's separated into mw and mobile ..looks [09:07:17] mutante: yeah, it can. changing wikistats.wmflabs.org to wikistats-live fixed it [09:08:11] YuviPanda: ah! cool.. i switched it via web ui when i moved the VM..somewhere [09:08:23] the instance name that is.. [09:08:27] ah, ok [09:08:41] mutante: hashar had a hack somewhere that let wmflabs.org URLs hit the proxy automatically [09:08:50] would be nice if we could port that to toools [09:09:16] i didnt notice any issue because it worked from me from external [09:09:25] after i switched instance name around during migration [09:09:34] nods [09:10:05] YuviPanda: you have this: 'ypanda': group => 'mobileupld' [09:10:14] but you say you want: [09:10:21] group => 'mwupld' [09:10:27] for mw ? [09:10:49] you and brion can upload mobile stuff, csteipp,hashar,mah,mglaser,reedy can upload mw [09:11:00] hi [09:11:04] so, the answer is, yea, needs a patch [09:11:14] in manifests/role/releases.pp [09:11:31] mutante: no, mobileupload. [09:11:40] mutante: no, so I logged in and it wouldn't let me do anything [09:11:43] hashar: hi, sorry for the ping :p [09:11:45] mutante: and groups only returned wikidev [09:11:50] if an instance attempt to hit the URL someentry.wmflabs.org , and someentry is a web proxy [09:11:58] the DNS will gives the public IP of the webproxy [09:12:07] and the instance can't reach it because that public IP is a NAT to another instance [09:12:22] the way I fix it is to use the role::betanatfix class (or something similiar) [09:12:38] @seen dan- [09:12:38] Steinsplitter: I have never seen dan- [09:12:40] aha.. i saw some of those changes to natfix [09:12:45] @seen dan-nl [09:12:46] Steinsplitter: Last time I saw dan-nl they were quitting the network with reason: Quit: dan-nl N/A at 4/30/2014 8:49:34 AM (23m11s ago) [09:12:52] which uses iptables to rewrite the IP destination from the public web proxy IP to the private web proxy IP [09:12:58] @notify dan-nl [09:12:58] I'll let you know when I see dan-nl around here [09:13:02] YuviPanda: that sounds like there is a general issue with mobile upload class [09:13:12] mutante: hmm. but brion was able to. [09:13:15] YuviPanda: do you know by chance if it works for Brion? [09:13:16] but yeah, that sounds possible [09:13:17] modules/beta/manifests/natfix.pp [09:13:18] ah, heh [09:15:59] YuviPanda: brb [09:16:38] hashar: but it'll fail for all wmflabs domains that do not work via the proxy [09:16:41] including betalabs [09:32:34] (03CR) 10Giuseppe Lavagetto: [C: 032 V: 032] Adding missing classes so that manifests compile. [labs/private] - 10https://gerrit.wikimedia.org/r/130574 (owner: 10Giuseppe Lavagetto) [09:34:17] YuviPanda: https://gerrit.wikimedia.org/r/#/c/130577/ [09:34:48] mutante: woo! cool! [09:35:00] mutante: ideally we should rename the ypanda thing to yuvipanda [09:35:04] I'm ypanda nowhere [09:35:47] YuviPanda: i know, second patch :) [09:35:55] apergos said the identical thing, heh [09:36:18] haha [09:52:09] YuviPanda: https://gerrit.wikimedia.org/r/130579 renaming you, ,ok? [09:52:35] notes you also have an account in role/analytics [09:53:13] mutante: yeah. [09:53:17] mutante: +1'd :D [09:56:03] /User[yuvipanda]/groups: groups changed '' to 'mobileupld' [09:56:05] try now [09:56:09] woo! [09:56:12] will do now [10:08:36] mutante: can I get 1 IP for project huggle? [10:15:14] petan: what does it do with it? more than just http? [10:16:08] if it's just a webserver i'd suggest yuviproxy [10:18:29] which you can set up at https://wikitech.wikimedia.org/wiki/Special:NovaProxy [10:18:35] it also gives you https and SPDY for free! [10:18:51] yep, very convenient [10:19:07] didnt have to deal with SSL setup anymore [10:19:12] and could give my IP back [10:19:44] all you have to do is map the subdomain to an instance [10:20:28] mutante: I need http that see IP of users who access it :/ [10:20:38] proxy doesn't seem to relay it [10:21:00] petan: it does. X-Forwarded-For [10:21:04] oh ok [10:21:07] mutante: nvm then [10:21:18] ok, great [10:21:27] :) [10:21:36] YuviPanda: how I retrieve that (in php script) [10:21:55] petan: it is a http header, so however you access the other headers? [10:22:28] I guess using $_SERVER['HTTP_X_FORWARDED_FOR'] [10:22:42] should, yeah [10:22:48] * YuviPanda hasn't written PHP in a long time [10:24:34] yes it works [12:50:36] Damianz: I finished setting up pgsql permissions, you should be now able to access whitelist database directly from any labs instance using host "huggle" user "labs" password "public" [12:50:54] just run select name from whitelist_public [13:00:42] petan: it is possible to transolate the output of @seen? now? [13:00:52] not yet.. [13:01:27] :'-( [14:01:30] I have a project which uses a bot (crawler) for some round checks. Where do I have to put this bot in tool-labs? [14:04:19] note: i do a repo for my tool [14:04:36] / do have a repo [14:05:43] 3Wikimedia Labs / 3Infrastructure: !add-labs-user gone, fix or add docs to link SVN users to labs/wikitech - 10https://bugzilla.wikimedia.org/64596#c1 (10Antoine "hashar" Musso) See also Bug 63028 - Inconsistent GID number (500 or 550) In this case I think jayvdb GID should be changed in LDAP from 550 (prod... [15:54:43] 3Tool Labs tools / 3WMT bots: Fix bot-operator so we can give commands to the bots - 10https://bugzilla.wikimedia.org/60858#c3 (10PiRSquared17) Bump. [16:38:29] Coren: there's nothing in our terms of use forbidding a user having two different labs accounts, is there? [16:38:48] andrewbogott: Good question. I don't believe there is; nor would I worry about it. [16:39:12] andrewbogott: Although, like everything else, too many would be abusive and I'd start wondering wth. [16:39:19] Coren: for users who have a different prod vs. labs username, I may suggest that they just create a new labs account with their prod username and we rely on that to back production accounts. [16:39:49] Messy, but I can't think of any real reason to not do it that way if they're attached to having different uids in prod vs. labs [16:39:59] Yeah, sounds sane. [16:40:05] 'k [16:40:16] Just wanted to make sure I'm not endorsing labs sockpuppetry [16:40:59] andrewbogott: I think it's also considered OK because renaming accounts is messy [16:41:16] i.e. 'just make a new account instead of trying to get your old one renamed' [16:41:23] valhallasw: yes! I tried to rename a user on Monday and now his account is utterly broken and I will never try that again [16:50:40] * andrewbogott emails everyone [16:54:29] andrewbogott: on my wikidata-test instance, the disk space in "/var" is almost full. (it has ~1.8gb) [16:54:38] 1.1 gb is in /var/lib/puppet/clientbucket [16:54:59] 3Wikimedia Labs / 3Infrastructure: beta: Get SSL certificates for *.{projects}.beta.wmflabs.org - 10https://bugzilla.wikimedia.org/48501#c91 (10Greg Grossmeier) 5NEW>3RES/WON Won't Fix'ing this for now. A) We have self-signed certs in place on Beta B) Real certs are expensive C) There hasn't been any te... [16:55:00] is that normal? or is that supposed to be cleaned somehow? (regularly) or needed or what? [16:55:57] aude: I don't know! Let me google a bit. [16:55:57] i see cleaning done https://wikitech.wikimedia.org/wiki/Server_admin_log/Archive_21 [16:56:04] Mind if I log in a poke around a bit? [16:56:06] Tim: on neon: cleaned up /var/lib/puppet/clientbucket/* since it was out of disk space [16:56:10] sure [16:56:18] it's self-hosted puppetmaster [16:57:44] aude: https://ask.puppetlabs.com/question/132/does-filebucket-need-periodic-maintenance-cleaning/ [16:57:53] I don't know why yours has grown so much, but it should be safe to clean [16:57:54] thanks [16:58:00] no idea why [16:58:17] it's looks like "cached" stuff [16:58:36] hiya Coren, I can't create a new instance in the analytics project anymore [16:58:43] are we at our quota or something? [16:58:53] done :) [16:59:31] aude: Apparently there's a way to specify that a particular resource not get backed up there. I don't see an obvious system-wide default for that though [16:59:39] hmmm [16:59:44] not done [16:59:46] ottomata: you can view your quota status; there's a link on the manage projects page [17:00:56] aude: now what? [17:02:08] trying puppet now [17:02:21] yeah, thanks andrewbogott [17:02:23] it says 20/20 [17:02:24] seems ok [17:02:29] did our quota decrease when we moved to eqiad? [17:02:33] we used to have much more than that, i'm sure [17:02:59] ottomata: I don't remember if I duplicated quotas or not. [17:03:10] Anyway I can bump you up if you don't have any instances that can be pruned... [17:04:03] yeah, not really, most are not mine [17:04:09] and we had a lot there because I used it for hadoop cluster testing [17:04:23] ok, hang on... [17:04:23] would appreciate a bump [17:04:33] how are you doing ram and core-wise? [17:05:20] Oh, you're out of cores, not instances [17:06:07] ottomata: try now? [17:07:14] 3Wikimedia Labs / 3Infrastructure: beta: Get SSL certificates for *.{projects}.beta.wmflabs.org - 10https://bugzilla.wikimedia.org/48501#c92 (10se4598) 5RES/WON>3REO This bug seems to be drifted away from the initial comment. (In reply to Antoine "hashar" Musso from comment #0) > We need certificates gen... [17:07:22] worked, thanks andrewbogott! [17:45:28] 3Wikimedia Labs / 3tools: Killed Mysql queries still running - 10https://bugzilla.wikimedia.org/64140#c8 (10kolossos) After 11 days still running: tools.templatetiger@tools-login: MariaDB [(none)]> show processlist; +----------+--------+-------------------+-------------------------+---------+---------+------... [18:37:44] Coren: please take a goold look at my cron tab before it gets migrated, I always get hit with labs bugs [18:51:23] @seen dan-nl [18:51:23] Steinsplitter: Last time I saw dan-nl they were quitting the network with reason: Quit: dan-nl N/A at 4/30/2014 8:49:34 AM (10h1m48s ago) [20:59:14] !log local-wikibugs Deployed a48a000 [20:59:15] Logged the message, Master [21:25:44] 3Wikimedia Labs / 3tools: Killed Mysql queries still running - 10https://bugzilla.wikimedia.org/64140#c9 (10Tim Landscheidt) a:5Marc A. Pelletier>3Sean Pringle Tim, just to be clear: On what DB host is this query running/hanging? Looking at /data/project/templatetiger/public_html/einspielen.sql, this se... [21:56:58] 3Wikimedia Labs / 3tools: Killed Mysql queries still running - 10https://bugzilla.wikimedia.org/64140#c10 (10kolossos) It's running on tools-db. I started it with jsub. I never restarted the query and it has no contact to the official enwiki-db because I import an extract of templates as a text-file into my... [22:15:14] 3Wikimedia Labs / 3tools: /var is full on tools-webgrid-01 due to me spamming /var/log/auth.log with sudo - 10https://bugzilla.wikimedia.org/64683#c1 (10Tim Landscheidt) 5NEW>3ASS p:5Unprio>3Lowest For the time until we fix bug #61102, I have installed a script /home/scfc/bin/cleanup-php-cgis per cro... [22:16:19] scfc_de: BTW, the grid now kills webgrid jobs with SIGTERM instead. [23:13:28] 3Wikimedia Labs / 3tools: Tool Labs: Provide anonymized view of the user_properties table - 10https://bugzilla.wikimedia.org/58196#c33 (10Tisza Gergő) Is the order of rows randomized? Being able to correlate other preferences with, say, fancysig based on row number would be bad. As for timestamp anonimizati... [23:48:12] 3Wikimedia Labs / 3Infrastructure: beta: Get SSL certificates for *.{projects}.beta.wmflabs.org - 10https://bugzilla.wikimedia.org/48501#c93 (10Matthew Flaschen) I can think of three significant problems with self-signed certificates: 1. It trains people to ignore SSL warnings, which means they ignore them...