[13:49:45] zeljkof: hello :) [13:49:51] zeljkof: so hmm ruby-lint is a good idea [13:49:58] zeljkof: but would need to be packaged first :D [13:50:03] hashar: I saw your e-mail [13:50:16] more work than I was hoping :) [13:50:52] I am waiting for Chris to wake up, let's see what he thinks [13:50:54] zeljkof: it is probably very easy to package it though [13:51:04] hashar: do you need it for yourself? [13:51:12] not really :D [13:54:13] hashar: I would like to get it working [13:54:19] will see what Chris says [13:56:55] hashar: this is also interesting https://codeclimate.com/github/wikimedia/qa-browsertests [13:58:35] zeljkof: that is a software as service to audit your code ? :-D [13:58:45] we should start a company for PHP projects :-D [13:59:06] hashar: the site tracks your code, and when a class goes bad, lets you know :) [13:59:12] I think it is Ruby only for now [13:59:16] bad daughter awake again :/ [13:59:29] php as a similar tool to audit code [13:59:32] bbl [14:38:33] <^demon> hashar: I had to disable the replication again. It was still failing. [14:39:13] :-( [14:39:41] ^demon: any specific message ? [14:39:42] <^demon> Same thing about the key failing. Which was weird, because I was manually able to `ssh -i gerritslave@gallium` [14:40:30] maybe the host verification? [14:40:42] Gerrit would have to accept the gallium fingerprint key [14:40:46] <^demon> Ooh, could be. [14:40:59] <^demon> Lemme try sudoing as gerrit2 first so I get the right known_hosts. [14:41:13] I had a similar issue when configuring the ssh connection for Zuul. It connects to Gerrit as user jenkins [14:41:34] so on manganese you might need to ssh as gerrit2 and attempt sshing to gallium then manually accept the fingerprint [14:42:34] <^demon> Ohhhh yeah, I remember this now, I had this problem with gerrit. [14:43:05] <^demon> You have to manually add the fingerprint to the hosts file. Gerrit won't auto-add it. [14:43:27] securityyyy [14:43:43] you probably want to add that as a comment somewhere in puppet [14:43:46] <^demon> Ok, I'll manually add it back to replication.config and try replicating to gallium. [14:43:52] <^demon> If it works, we'll re-add it in puppet. [14:46:55] gerritslave should be able to write now: [14:46:56] $ ls -ld /var/lib/git [14:46:57] drwxr-xr-x 3 gerritslave root 4096 Nov 22 13:13 /var/lib/git/ [14:49:50] <^demon> Hmm, still failed. Same error. [14:50:00] <^demon> reject HostKey: gallium.wikimedia.org [14:53:58] :( [15:04:39] ^demon: could the replication plugin use a different system than ssh and ~/.ssh/known_hosts ? [15:07:16] <^demon> No, it uses gerrit2's known hosts. [15:07:25] <^demon> Just not the system ssh, mina instead. [15:13:44] <^demon> hashar: Can I abandon 12598 and 17069? [15:21:56] !g 12598 [15:21:56] https://gerrit.wikimedia.org/r/#q,12598,n,z [15:21:59] !g 17069 [15:22:00] https://gerrit.wikimedia.org/r/#q,17069,n,z [15:22:25] 12598 yeah [15:22:27] abandonned it [15:22:42] 17069, I really could use a global dumper [15:24:17] ^demon: I have to leave anyway [15:24:27] thanks for all your help with gerrit perm and replication :-) [15:24:39] <^demon> No problem. I'll try to sort out this host key thing. [15:25:03] <^demon> If we're gonna do 17069, let's add it to eval.php, like I suggested. [15:25:30] yeah should probably make that a nice class [15:26:05] will be back later this evening anyway [15:26:10] meantime, have a good afternoon! [16:17:16] hi qgil [16:48:18] andre__: we're now below 1200 unprioritized according to the saved search "0. Unprioritiesed, non‑enhancement..." [16:49:28] Oh nice. Still that reminds me that I cannot rename other people's saved searches, even if they have typos. ;) [16:50:00] :/ [18:16:52] bsitu: here's the config change fabrice asked for: https://gerrit.wikimedia.org/r/#/c/38736/ [18:17:48] kaldari: okidoki [18:18:59] I left the revert settings as separate from all (even though they're the same), so that other people can see how to set the settings for individual notices (as an example) [18:19:51] okay, I was just about to comment it's not necessary, :) [18:21:26] I should probably add some comments to that effect [18:25:51] ^demon, so why aren't the other public repositories being replicated? [18:26:07] <^demon> I just responded on the bug. [18:26:16] <^demon> I was explicitly asked not to replicate some repos. [18:28:54] ^demon, I read that and it prompted me to ask you why [18:29:09] If someone asked you to not replicate them they must've had a good reason [18:29:32] <^demon> Because there's a couple of repos that are being primarily developed in github, and gerrit is more of a copy. [18:29:41] <^demon> This was some stuff in analytics, I believe. [18:29:47] <^demon> Or fundraising. [18:29:49] <^demon> I can't remember. [18:41:13] kaldari: hi! [18:41:24] howdy! [18:41:48] flickr-uploading is turned on for admins and image-reviewers on commons now [18:42:08] BTW, is it true that this is actually the first GSoC code this year to go live? [18:42:21] I was assuming it was the last :) [18:42:58] https://www.mediawiki.org/wiki/Summer_of_Code_Past_Projects [18:43:06] the first was Wikimedia Incubator stuff [18:43:52] kaldari: I sent a note about flickr uploading to wikitech-ambassadors already; I didn't mention the imagereviewers though [18:44:14] that's ok, they wouldn't know what that means anyway :) [18:44:18] okay [18:44:19] :-) [18:44:29] kaldari: is there discussion happening on the Commons vp or something? [18:44:39] image-reviewer is a special user group that only exists on Commons [18:44:47] there was [18:45:01] it's archived now [18:45:23] any feedback in BZ? [18:45:42] kaldari: I wasn't on IRC for some of my time in SF, and then Tues I was traveling and Wed I was sick [18:45:58] that sucks, lots of people sick :( [18:48:22] it's interesting that a lot of the GSoC project were 'almost completed' [18:49:09] maybe next year we should tell everyone to multiply their time estimates by 120% [18:49:27] yeah [18:49:35] also next year I think everyone ought to have 2 mentors [18:49:47] for more frequent code review [18:49:52] that's an interesting idea [18:50:06] did you look at my evaluation of GSoC that I sent in Nov? [18:50:19] no [18:50:22] http://lists.wikimedia.org/pipermail/wikitech-l/2012-November/064498.html [18:50:31] I failed at ensuring that students were scoping [18:50:31] their proposals at about 6 weeks of coding work and dedicating the [18:50:31] entire last month of the summer to bugfixing and code review. [18:50:54] evening [18:51:05] we need to work harder at that next year -- I imagine qgil agrees with me :-) [18:51:21] ha, yeah we had 3 solid months of bugfixes and code review for Flickr-uploading [18:51:39] yes :) [18:52:10] but GSOC is tough no matter what. We have seen very senior projects with very senior mentors struggling to get good rates [18:52:11] but that's mainly because I could only work on it on Fridays [18:58:34] one of the most successful projects for 2012 MediaWiki GSoC was Nischay - he had 2 mentors [19:02:36] https://gerrit.wikimedia.org/r/#/c/18127/ - ori-l do you have any insight into the account creation stuff? [19:04:55] * sumanah emails S Page to ask [19:47:57] howdie [20:13:57] hashar: I'm sure you're tired of this by now, but is there any word on getting wikibugs pulled from git? :) [20:14:16] marktraceur: that is like at the bottom of my job list :/ [20:14:26] marktraceur: I still have to ping Faidon about git::clone() :-D [20:14:31] *nod* a shame, but good to know [20:14:55] marktraceur: maybe we should just sprint a replacement for wikibugs [20:14:58] hashar: Would you mind emailing him, so it's not pending based on a deferred I/O operation? [20:15:08] marktraceur: Mozilla is using a python IRC bot named supybot with a bugzilla plugin. [20:22:39] People on visualeditor have way too much free time now... [20:23:29] csteipp: We're just a fun-loving group :) [20:27:05] qgil: sent you email to get your reaction to two "groups" proposals [20:27:19] chrismcmahon, answering as we chat [20:27:20] csteipp: ask the VE team to set it up on beta so :-] [20:27:44] csteipp: I am sure chrismcmahon would love to have Visual Editor on beta to write some selenium tests for it hehe [20:28:01] chrismcmahon, or do you want to chat over hangout? [20:28:13] qgil: sure [20:28:17] marktraceur: if you want to take the lead on wikibugs, you could rebase https://gerrit.wikimedia.org/r/#/c/26325/ on latest production [20:28:29] hashar csteipp I'd like to have EVERYTHING on beta, but I'm a patient man [20:28:34] marktraceur: the git::clone() calls should use ensure => latest instead of ensure => [20:28:39] marktraceur: I think that will sort it out [20:28:59] chrismcmahon: we need to evangelize the other teams to start using beta for their tests :-] [20:28:59] hashar: I'd be happy to help, sure [20:29:13] marktraceur: so rebase, ensure => latest ping faidon and I about it :-) [20:29:29] marktraceur: sorry I am not filling like spending a couple hours solving wikibugs :/ [20:29:43] marktraceur: too many things to handle already, so wikibugs is really low priority for me :( [20:29:53] * hashar needs a second brain or to hire someone. [20:38:40] <^demon> hashar: I forgot. It only creates on the remote end when you're actually *creating* a new repo (not for repos that already existed and are just now being replicated). [20:38:52] <^demon> Easiest solution is to just rsync the repos over probably. [20:44:28] ^demon: that is interesting. [20:44:40] ^demon: might be worth opening a bug against the Gerrit replication plugin [20:44:54] ^demon: are you syncing ? [20:44:56] <^demon> Yep. [20:45:14] doh [20:45:37] <^demon> About 25% done [20:46:02] we need root to delete the old /var/lib/git/integration dir which belongs to root:root (it was used to install Zuul, now it is in /usr/local/src/ :) [20:46:05] nice [20:58:06] ^demon: mutante removed the /var/lib/git/integration dir that belang to root:root :-] [20:58:26] <^demon> cool, thanks. [20:59:21] with that I guess we will be able to release tar balls for any repo whenever a tag is pushed :-] [20:59:33] ^demon: are the private repos replicated ? [20:59:43] <^demon> The one private repo, yes. [20:59:49] <^demon> labs/private [21:00:10] <^demon> operations/private is deprecated. [21:00:35] can we filter out that repo ? [21:00:47] New patchset: Merlijn van Deen; "Test extensions in multiple branches" [integration/jenkins-job-builder-config] (master) - https://gerrit.wikimedia.org/r/38790 [21:02:25] <^demon> operations/private was just deleted, so we only have labs/private. [21:02:41] qgil: groups are officially proposed, thanks for the fine-tuning. [21:02:46] New review: Hashar; "I will review and test that next week. Thanks a ton for your help!" [integration/jenkins-job-builder-config] (master); V: 0 C: 0; - https://gerrit.wikimedia.org/r/38790 [21:03:07] <^demon> hashar: Once we rsync it this time, we'll just delete it. gerrit won't replicate it since it already exists, remember? ;-) [21:04:01] chrismcmahon, good! Please apply the templae http://www.mediawiki.org/wiki/Groups/Proposals/Proposal_template whenever you have a chance [21:04:59] chrismcmahon, and then announce each one at wikitech. There is also a template to save you work there: http://www.mediawiki.org/wiki/Groups/Proposals/Email_template [21:06:20] ^demon: if you say so :-] [21:13:01] qgil: what is the difference between a "Promoter" and a "Member" of a group? [21:13:20] chrismcmahon, you are a promoter :) [21:13:44] qgil: harsh / me wanted to ask. Can I be a member of the Ahmedabad group, or does that make no sense/ [21:13:44] chrismcmahon, promoters are the ones pushing the group [21:13:44] ? [21:14:00] YuviPanda, will you participate in their activities? [21:14:05] qgil: yes. [21:14:09] then yes [21:14:16] sweet. [21:14:31] YuviPanda, why the question? isn't the answer obvious? [21:14:59] qgil: just confirming. Not sure how 'local' local groups are supposed to be [21:15:42] YuviPanda, the groups are local, but they are open to anybody willing to participate in their activities [21:15:46] no password requested ;) [21:15:50] :D [21:15:50] I mant passport [21:15:53] sweet! [21:15:55] m e a n t [21:15:58] :P [21:16:05] after effects of the holiday party? :P [21:16:45] er, shoore [21:16:48] YuviPanda, and I was wondering whether you had seen http://www.mediawiki.org/wiki/Groups/Proposals/Marketing [21:17:10] <^demon> hashar: Done rsync'ing. Doing a full replication now to test. If it all goes fine, we'll re-enable in puppet. [21:17:31] qgil: isn't that usually called 'outreach' than 'marketing'? [21:17:53] i was initially confused as to what marketing meant in this context [21:18:08] YuviPanda, I'm glad you ask. Your opinion is welcome at http://www.mediawiki.org/wiki/Talk:Groups/Proposals/Marketing [21:19:02] YuviPanda, "marketing" is an accurate word, but it seems to be disliked by some in an environment like this :) [21:19:08] yes, that too [21:19:16] a negative connotation :) [21:19:53] YuviPanda, sonds too FOSSoldschool to me, but this is not a battle I will pick. If people prefer another term so be it [21:20:05] ^demon: you are awesome :-] [21:20:32] ^demon: is there any synonym for "awesome" ? Cause I keep using that word [21:20:37] qgil: added my 2c on the naming bikeshed, and added my nam eto volunteers :) [21:20:38] YuviPanda, just leave your opinion in the talk page please. And if you like the charter please sign up or endorse [21:20:54] (maybe that is just because everyone around is awesome and there is not any better definition floating around) [21:20:59] YuviPanda, you're fast. Thank you! [21:21:23] qgil: :) [21:21:50] hashar, amazing, da one, greeeat.... :P [21:22:11] you are alternatively allowed to use 'chad' [21:22:53] YuviPanda, I reserve the "ch" word for very special occasions... (ok, and now back to work) [21:23:29] qgil: thanks :) [21:23:50] qgil: and sorry about the hate mail I sent you this week. I was getting upset to get a myriad of bug notifications :-] [21:23:50] hashar, I don't think I gave you any good advice. I would avoid those words. :) [21:24:08] qgil: thanks for your nice reply. That helped me understand why you did that and I discovered the metrics you are building (much helpful) [21:24:18] hashar, no worries. It help me decide to change the approach. [21:25:12] geez, you two are being too nice to each other :P [21:25:16] * YuviPanda heads to bed [21:25:29] hashar, now I'm sending emails like that one explaining why it is a good idea not to let "Nobody" and wikidata-bugs to e assignees of FIXED bugs [21:25:47] YuviPanda, we are nice - and we can't avoid it [21:25:56] and now... I really go back to work :) [21:31:00] qgil: I am wondering if that could be enforced in bugzilla (prevent resolution of bug whenever the assignee is wikibugs / nobody) :-D [21:31:45] hashar, I still think the basic principle applies: if it takes less the 5 seconds, do it. If it takes more, don't bother. :) [21:32:17] hashar, we have more important problem, but this one is so easy to solve each time someone assigns or fixes a bug. [21:35:22] qgil: yeah will remember about it as much as I can :-] [21:56:41] <^demon> hashar: Oh btw, I added a plugin to jenkins yesterday that lets you do some cool filtering with views. It let me create https://integration.mediawiki.org/ci/view/All-Failing/ for example :) [22:16:58] ^demon: ohhh [22:18:59] ^demon: I ultimately want to get a view that show all repositories which are failing tests. That is more or less the idea for the regression jobs at https://integration.mediawiki.org/ci/view/Regressions/ [22:19:12] ^demon: I run the PPHUnit test suite on core after a change has been merged. [22:19:15] works fine for master [22:26:32] marktraceur, I still have your browser tab up showing VE's problem with z-ordering on the Caputh page. Did you note the problem? Do you want me to? [22:26:59] spagewmf: Oh, sorry, I meant to reply--I filed the bug, we're good now [22:27:08] np [22:29:00] <^demon> hashar: Is something up with jenkins? I've still got no results on https://gerrit.wikimedia.org/r/#/c/38798/ [22:30:11] ^demon: krinkle told me about it [22:31:03] restarted zuul [22:31:11] maybe the replication broke the Gerrit stream events ? :) [22:31:41] <^demon> Dunno. [22:32:33] grmblbl [22:33:41] I will have to look at the 300k lines of debug logs for zuul [22:35:35] hashar: what is the gain with doing replication locally and pulling from there? [22:36:11] Krinkle: Gerrit does not support git archive [22:36:18] which I want to use to generate tarballs [22:36:39] archives for what? we check out change sets to run tests [22:36:53] that will let me remove a huuuuge hack in the jenkins jobs . I currently have to git clone --mirror from gerrit then git archive from the local checkout [22:37:14] for extensions? [22:37:24] among others yes [22:37:41] i also need bare repositories for something else, can't remember right now, it is too late :-) [22:37:45] but this is only for mediawiki core right? Others will keep the repo in the workspace and apply the change set (done by zuul) [22:38:03] <^demon> Krinkle: The gain is we don't have to do clones over the wire, we can clone from on-disk. [22:38:34] but that's only the first time a new repo is monitored, the others are just small fast-forwards [22:38:51] na the repository are cloned on each build [22:38:55] the workspaces are wiped [22:39:07] hashar: but not mediawiki core? [22:39:09] (right now0 [22:39:16] I am not sure for core [22:39:23] but for extensions workspaces are wiped out [22:39:34] well, it most certainly doesn't reclone the entire repo each change, that takes longer [22:39:44] I doubt it does it for extensions right now [22:40:12] <^demon> Anyway, replication is pretty snappy. Don't really have problems with lag. [22:40:31] Krinkle: ^demon: I am not sure what is happening between Gerrit and Zuul. Seems Gerrit might not have sent all its event via stream-event. I have restarted Zuul and it seems to be happy (test: https://gerrit.wikimedia.org/r/#/c/38800/ ). [22:40:34] logging that [22:41:10] <^demon> We've rebooted gerrit a bunch of times today, it could've very well lost something that was supposed to stream. [22:41:25] possibly [22:42:02] ^demon: will we be able to enable V+1 and V+2 on Gerrit/Zuul next week ? [22:42:22] I am on vacation at the end of next week so that is the last window for me :-) [22:42:51] <^demon> Yeah, we can do that Monday. We can bikeshed over the weekend on what to rename categories to :) [22:43:04] <^demon> VRFW-1 is "Failed, and +1 is "Verified", but we'll need 4 terms now. [22:44:10] V-1 -> "Go fix your code" [22:44:21] V+1 -> "You are not paying attention, lint failure" [22:44:38] V+2 -> "Jenkins is happy. Tests pass, now get a human to review the code" [22:44:45] V-2 -> "Epic failure." [22:44:49] <^demon> V+1 "Passes style/lint checks" [22:44:59] <^demon> V+2 "Passes all tests" [22:45:22] or we could use something simpler: V+1 "Error 0x309" V+2 "Panic." [22:45:30] thought that is not going to help [22:45:58] <^demon> Do we actually need -2? [22:46:02] !g 38802 [22:46:02] https://gerrit.wikimedia.org/r/#q,38802,n,z [22:46:10] <^demon> It could just be -1..+2 [22:46:11] ^demon: I don't think -2 is going to be useful [22:46:30] we just want to prevent merges aren't we ? [22:46:47] though we could make V-1 to be overridable by a manual V+1 [22:46:54] and V-2 to be un overridable. [22:47:18] <^demon> Well it's still obey the same Lowest-Score-Veto rules, and normal remove reviews permissions. [22:47:19] or unacceptable change even by a human [22:47:48] -1..+2 is probably good enough so :-) [22:48:17] 2012-12-14 22:47:58,822 ERROR gerrit.GerritWatcher: Exception on ssh event stream: [22:48:19] bahh [22:48:29] Zuul is not happy with a received JSON object [22:48:39] ah [22:48:40] error: [Errno 111] Connection refused [22:48:41] <^demon> Recently, or earlier? [22:48:50] just now, like a minute ago [22:49:04] (I am tailing the Zuul debug log gallium: /var/log/zuul/debug.log [22:49:20] <^demon> Puppet just ran on manganese. [22:49:27] <^demon> I think the replication change just went out. [22:49:32] that might be it [22:50:07] <^demon> Yup, that's it. [22:50:28] <^demon> Ok, so that's all settled in puppet now so we can take that off the todo list :) [22:50:31] <^demon> I'll close the bug [22:52:12] one less bug!! [22:52:50] ^demon: on the receiving host, do we have a way to find whether the replication is actually working ? [22:53:52] <^demon> Timestamps of when files are updated? Incoming ssh connections from manganese? [22:54:00] <^demon> There's not a log or anything. [22:54:21] that is what I thought :-) [22:55:04] that is good enough for now anyway. Thanks again for your help on setting up the replication. [22:55:11] that is going to make my life a bit easier :D [22:56:23] ^demon: I had to write a crazy shell script to be able to use "git archive" : http://dpaste.org/LUQ7H/ ; ) [22:56:49] now I will simply git archive --remote=/var/lib/git/mediawiki/core master [22:56:51] \O/ [22:59:12] anyway bed time for me [23:02:27] <^demon> Not a problem. I'm gonna call it a night. I've gotten everything done on my todo list for this week and I don't want to get pulled into something new :p [23:14:21] spage: https://gerrit.wikimedia.org/r/#/c/38805/ [23:16:23] ^demon: I am wondering if puppet could be restarting Gerrit again [23:18:10] I will monitor zuul this weekend [23:18:16] more than time for me to sleep a bit [23:18:23] thanks again *waves*