[04:59:56] TimStarling: around? [05:22:55] * ori is pleased with https://gerrit.wikimedia.org/r/#/c/196989/ [13:46:30] 6MediaWiki-Core-Team, 10hardware-requests, 6operations, 5Patch-For-Review: Fluorine needs bigger disks - https://phabricator.wikimedia.org/T92417#1121326 (10yuvipanda) @andrew any updates? [14:25:29] 6MediaWiki-Core-Team, 6operations, 7Wikimedia-log-errors: rbf1001 and rbf1002 are timing out / dropping clients for Redis - https://phabricator.wikimedia.org/T92591#1121388 (10Joe) @chasemp what exactly doesn't work in the hiera change? I see the config on rbf1002 server seem to be correct at the moment. [15:19:46] 6MediaWiki-Core-Team, 6operations, 7Wikimedia-log-errors: rbf1001 and rbf1002 are timing out / dropping clients for Redis - https://phabricator.wikimedia.org/T92591#1121537 (10chasemp) >>! In T92591#1121388, @Joe wrote: > @chasemp what exactly doesn't work in the hiera change? I see the config on rbf1002 ser... [16:20:37] anomie: better? -- https://phabricator.wikimedia.org/T75062#1121750 [16:21:34] bd808: Yes, thanks for not making me do it, since I'd probably have come across poorly. [16:38:59] 6MediaWiki-Core-Team, 10Wikidata-Query-Service: Build tool to update RDF store - https://phabricator.wikimedia.org/T92848#1121825 (10Manybubbles) 3NEW [16:39:06] 6MediaWiki-Core-Team, 10Wikidata-Query-Service: Build tool to update RDF store - https://phabricator.wikimedia.org/T92848#1121833 (10Manybubbles) p:5Triage>3Normal [16:39:19] anomie: Have you seen https://phabricator.wikimedia.org/T91432 yet? Every time I think I understand what JamesA is asking for he restates things in a new way that is just more confusing to me. [16:39:31] * anomie looks [16:39:33] 6MediaWiki-Core-Team, 10Wikidata-Query-Service: Build tool to update RDF store - https://phabricator.wikimedia.org/T92848#1121843 (10Manybubbles) 5Open>3declined a:3Manybubbles [16:39:48] 6MediaWiki-Core-Team, 10Wikidata-Query-Service: Build a tool for synchronizing Wikidata changes into Blazegraph - https://phabricator.wikimedia.org/T89852#1121845 (10Manybubbles) [16:39:58] What I still can't figure out is if we just need some list of users to import into the config or something completely new in the app [16:40:05] 6MediaWiki-Core-Team, 10Wikidata-Query-Service: Build a tool for synchronizing Wikidata changes into Blazegraph - https://phabricator.wikimedia.org/T89852#1121850 (10Manybubbles) p:5Low>3Normal a:5Jdouglas>3Manybubbles [16:41:09] 6MediaWiki-Core-Team, 10Wikidata-Query-Service: Build a tool for synchronizing Wikidata changes into Blazegraph - https://phabricator.wikimedia.org/T89852#1047037 (10Manybubbles) We've now verified that Blazegraph works. Here are some additional requirements: As wikidata changes we need to scrape those chang... [16:45:51] "this is correct" [16:46:46] bd808: His basic complaint seems to be that he wants SecurePoll to query "can vote?" on every single wiki the account is attached to instead of the wiki the user actually came from. No actual reason given, just "we want this" despite it never existing for previous elections. [16:47:13] anomie: Yeah. I gathered this was a new thing they just dreamed up [16:47:59] Dan seemed a bit confused by it all too despite having had a couple meetings with JamesA about it [16:48:29] I may need to break down and call a meeting with JamesA directly [16:49:05] I think this is somehow related to SUL finalization and someone deciding that should change how elections work [16:49:20] hmm? [16:49:34] bd808: The original request was for the basic edit count. Then in T91432#1083556 he realized they don't actually use that anyway (at least not as far as I know), they use a pre-built list for edit counts before a data and between two dates. [16:49:34] legoktm: https://phabricator.wikimedia.org/T91432 [16:50:43] CentralAuthUser has a getGlobalEditCount() method [16:50:58] it's not cached though, and not very efficient [16:55:04] legoktm: That doesn't sound incredibly promising (the inefficient part) [16:55:47] But if I understand this right we just need a one-time list generation for each election so maybe it's not too bad [16:55:53] bd808: When we were doing the SecurePoll stuff, the general consensus was "If they want something more crazy than what the current auto-generation can handle, they can write a script to do it like they always did in the past." So they can write a script that calculates eligibility globally and then dumps those accounts into every wiki's allowedness list. That'll still break if someone who doesn't already have a local account on $wiki goes there [16:55:53] just to try to vote (because SecurePoll's lists work with local user_ids), but I really don't care about such a crazy edge case. [16:56:32] hmmm [16:57:02] and then Deskana|Away forgot this and decided to trick me into giving them more possibly [16:57:35] This sounds like it is going to need a meeting with me, you, Dan and JamesA to sort out sadly [17:12:04] ^d: https://gerrit.wikimedia.org/r/#/c/195033/ [17:14:00] 6MediaWiki-Core-Team, 10Continuous-Integration, 10Incident-20150312-whitespace: add a check for whitespace before leading 6MediaWiki-Core-Team: CirrusSearch: Remove reliance on dynamic scripts - https://phabricator.wikimedia.org/T92852#1122001 (10Manybubbles) 3NEW [17:26:05] 6MediaWiki-Core-Team: CirrusSearch: Remove reliance on dynamic scripts - https://phabricator.wikimedia.org/T92852#1122044 (10Manybubbles) [17:26:28] <_joe_> AaronSchulz: I would need a review on https://gerrit.wikimedia.org/r/#/c/194830/ [17:26:43] <_joe_> I am just configuring it as a read-only dc for now [17:26:52] <_joe_> but I'm not sure I didn't miss anything [17:27:17] <_joe_> I was planning on merging this tomorrow [17:35:49] _joe_: are there any codfw poolcounter servers? [17:36:46] <_joe_> AaronSchulz: not yet, and that was a question for you [17:36:54] legoktm: thanks for https://gerrit.wikimedia.org/r/#/c/196667/ [17:36:56] <_joe_> we need them to be local, right? [17:37:09] np :) [17:37:11] they should be, yeah [17:37:19] I guess for now it doesn't matter if no traffic is there [17:37:23] <_joe_> but still- codfw will likely use eqiad dabases as masters [17:37:26] ideally wgPoolCounterConf would be in one common place [17:37:54] <_joe_> so I had doubts about separating poolcounters [17:38:03] only the actual wgPoolCountClientConf has servers in it [17:38:41] poolcounter does not protect master load (I guess one could imagine that, but we don't use it for that) [17:39:07] s/protect/curtail [17:40:00] did sean already OK the DB part? [17:40:21] using the eqiad masters for masters is OK there of course, I assume the other bits match up [17:46:39] <_joe_> AaronSchulz: yes he did [17:46:47] <_joe_> AaronSchulz: and thanks for reviewing it [17:47:57] 6MediaWiki-Core-Team: Configure 'enqueue' job queue type to be DC-local and in a dedicated runner loop - https://phabricator.wikimedia.org/T92863#1122149 (10aaron) 3NEW [18:08:00] 6MediaWiki-Core-Team, 6operations, 7Wikimedia-log-errors: rbf1001 and rbf1002 are timing out / dropping clients for Redis - https://phabricator.wikimedia.org/T92591#1122221 (10chasemp) relevant https://phabricator.wikimedia.org/T90923 [18:15:54] <_joe_> AaronSchulz: what would you suggest about the common part of poolcounter config et al? [18:27:39] <^d> _joe_: Do something like we do with CirrusSearch. [18:27:52] <^d> Have a common config file that then includes a realm-specific one [18:28:02] <_joe_> ok [18:28:02] <^d> Rather than stuffing everything in realm-specific where it doesn't belong [18:28:10] <_joe_> yeah I'll do this [18:29:47] <_joe_> I hope I'll get someone to review it before tomorrow's swat [18:30:25] <^d> I can help poke this some [18:36:02] <_joe_> ^d: feel free to do it [18:36:21] <^d> I'll clean up the Cirrus config to how I think it should ideally be :) [18:41:08] Tim-away: https://phabricator.wikimedia.org/T24097#263825 do you remember if you ran fixUserRegistration.php on any other wikis besides enwiki? [19:58:13] 6MediaWiki-Core-Team, 10MediaWiki-Maintenance-scripts: Add batching support to fixUserRegistration.php - https://phabricator.wikimedia.org/T92890#1122825 (10Legoktm) 3NEW a:3Legoktm [20:09:38] 6MediaWiki-Core-Team, 10MediaWiki-Maintenance-scripts, 5Patch-For-Review: Add batching support to fixUserRegistration.php - https://phabricator.wikimedia.org/T92890#1122890 (10Legoktm) [20:16:45] anomie: isn't the order by implict or something? I tested it with a batch size of 2 and it worked properly? [20:17:24] legoktm: No. In MySQL you'll get it implicitly if it chooses the obvious index, but there's no guarantee. [20:17:34] ok [20:19:18] https://old-bugzilla.wikimedia.org/show_bug.cgi?id=18638 looks exciting, except I have no idea how to do it [20:35:19] csteipp: You missed a few fregmentings ;) [20:39:23] bd808: You use znc, don't you? I'm trying it out, any tips? [20:40:55] * ori uses znc too. use a client cert to auth with freenode and never worry about nickserv again. [20:41:41] 6MediaWiki-Core-Team, 10MediaWiki-extensions-CentralAuth, 5Patch-For-Review: No valid null revision produced during global rename - https://phabricator.wikimedia.org/T76975#1122996 (10Legoktm) 5Open>3Resolved [20:42:42] ori: I've been using SASL for that for a while now, I don't think they had cert support yet when I set that up. Already got that working with the 'sasl' module ;) [20:43:19] anomie: Mine is pretty "stock" I think except for the module I load to send pings to my iphone [20:43:30] which won't be of much use to you I imagine [20:43:42] (it's iphone specific) [20:43:50] 6MediaWiki-Core-Team, 10SUL-Finalization: Figure out what to do with non-existent wikis that are still connected to SUL - https://phabricator.wikimedia.org/T89482#1123015 (10Legoktm) [20:43:57] bd808: I thought about figuring out something like that for my Android phone, but then I thought "Do I really want pings coming to my phone when I'm off?" [20:44:04] 6MediaWiki-Core-Team, 10SUL-Finalization: Run MassMessage to contact user talk pages of all affected accounts - https://phabricator.wikimedia.org/T90820#1123017 (10Legoktm) [20:44:20] Yeah. That's a downside [20:44:33] but I'm pretty good at scanning and then ignoring [20:44:35] 6MediaWiki-Core-Team, 10SUL-Finalization, 10Wikimedia-Site-requests: Run migrateAccount.php --auto on all usernames with unattached accounts - https://phabricator.wikimedia.org/T89770#1123018 (10Legoktm) [20:44:45] and I don't keep my phone around when I go to bed [20:48:24] * anomie set up certauth, chansaver, savebuff, sasl, and simple_away [20:48:38] * ori recommends fail2ban [20:57:39] anomie: why does SecurePoll have a SecurePollContent which just extends JsonContent and does nothing else? [20:58:23] legoktm: I forget. There was a reason at the time. [20:59:21] oh, I guess you need it for the custom handler [20:59:39] That sounds reasonable [20:59:41] yeah, that's probably why [21:02:14] bd808, ori: Is there a module to have znc not buffer the bots? [21:02:28] hmm... not that I know [21:10:57] 6MediaWiki-Core-Team, 5Patch-For-Review: Fix job runner service availability during redis outages - https://phabricator.wikimedia.org/T91835#1123175 (10bd808) 5Open>3Resolved [21:11:05] 6MediaWiki-Core-Team, 5Patch-For-Review: Investigate/prevent master revision query on enwiki edit form - https://phabricator.wikimedia.org/T86862#1123179 (10bd808) 5Open>3Resolved [21:24:35] 6MediaWiki-Core-Team, 10MediaWiki-Maintenance-scripts, 5Patch-For-Review: Add batching support to fixUserRegistration.php - https://phabricator.wikimedia.org/T92890#1123277 (10Legoktm) 5Open>3Resolved [21:54:14] I told myself I was going to leave this meeting "15 minutes" into it, but dangit organizational discussions.... [22:22:13] _joe_: make a PoolCounter-common.php with the rate limiting settings and have PoolCounter-eqiad/codfw.php for the server config [22:29:53] 6MediaWiki-Core-Team, 10Deployment-Systems: Can't update l10n cache - https://phabricator.wikimedia.org/T92900#1123630 (10bd808) [22:32:20] 6MediaWiki-Core-Team, 10MediaWiki-extensions-SecurePoll: Set up mini wikifarm in Labs which has SecurePoll on it - https://phabricator.wikimedia.org/T88725#1123643 (10bd808) p:5High>3Normal [22:34:10] 6MediaWiki-Core-Team, 10Deployment-Systems, 10Incident-20150312-whitespace, 5Patch-For-Review: scap's check_php_syntax() should check for text before ' KSmithTPG1: Will you be going to the scrum of scrums meeting (did anyone tell you about that one?) [22:52:01] KSmithTPG1: Wednesday at 10:30 SF time in R31 Elder [23:02:02] bd808: I was just asking the TPG folks about that [23:02:43] bd808: I'm inclined to try going. Do you think I definitely should, definitely shouldn't, or "meh"? [23:02:58] KSmithTPG1: +2 must do :) [23:03:04] thx [23:24:33] TimStarling: I looked at a random mw host (mw1001) and the l10n cache files there are all owned by mwdeploy which makes sense because that user runs rsync. I think maybe all that needs to happen on tin is for the l10nupdate:l10nupdate files to be chown to www-data:www-data. [23:25:12] There may be a bit more to it than that in the new branch code. I think it chmods these dirs to l10nupdate in the first place [23:25:32] I'm already working on a fix in scap [23:25:37] excelent [23:25:53] it will be basically the same as what LU does [23:25:54] * AaronSchulz waits for the Mr Burns gifs [23:26:21] ah, make a tmp dir and then copy things over when done? [23:26:24] yes [23:27:03] AaronSchulz: I'll go with this one -- http://aintnogod.com/ipb/uploads/gallery/album_17/gallery_298_17_1836.gif