[16:10:43] I'm having trouble enabling the analytics role on vagrant. Does this mean anything to anyone? ==> default: Error: Puppet::Parser::AST::Resource failed with error ArgumentError: Could not find declared class ::cdh::hadoop at /vagrant/puppet/modules/role/manifests/hadoop.pp:45 on node mediawiki-vagrant.dev [16:10:56] I even tried vagrant destroying, and starting from scratch. [16:20:15] earldouglas: are the submodules checked out? [16:24:48] Reedy: how do I check? [16:25:01] I did vagrannt enable-role analytics and vagrant git-update [16:25:09] Is there another step I missed? [16:25:46] I would have presumed it would've done submodules recursively [16:26:09] is there a cdh folder with contents in /vagrant/puppet ? [16:26:36] Nope, but it looks like this did it: git submodule update --init --recursive [16:26:45] Yeah, that should've... [16:27:16] As a general rule, should I do that whenever I enable roles? [16:27:33] bd808: ^ Does the vagrant stuff not recursively checkout submodules? [16:27:34] Could enabling the roles themselves take care of that? [16:27:41] (I know you're a bit busy with other stuff) [16:29:19] Well, tbh, I wouldn't expect it to be role dependent, I'd expect that it would clone the puppet repo and submodules anyway [16:30:07] But I've really not played with our varnish stuff much [16:31:21] * Reedy wonders where Dan is [17:17:41] guillom: "If you are an admin or have other special rights, you now need a strong password. [11]" <-- that hasn't happened yet [17:17:54] ; [17:17:54] // Temporarily set the policy for these roles to the previous WMF setting until [17:17:54] // we communicate the change to affected communities. [17:17:54] $wgPasswordPolicy['policies']['bureaucrat']['MinimalPasswordLength'] = 1; [17:17:54] $wgPasswordPolicy['policies']['sysop']['MinimalPasswordLength'] = 1; [17:17:56] $wgPasswordPolicy['policies']['bot']['MinimalPasswordLength'] = 1; [17:18:26] legoktm: Will it happen this week? [17:18:38] csteipp: ^ ? [17:19:07] It's under "Changes this week" so if it happens this week, it's good. Otherwise, well, I made a mistake. [17:19:29] guillom: I have a timeline I can send you. Staff / Stewards next week, then Checkusers about 3 weeks later. [17:19:34] The Phabricator ticket has the WMF-deploy-2015-06-30_(1.26wmf12) tag. [17:19:51] guillom: yeah, the patch is going out then, but not the config :/ [17:19:57] guillom: That's just the code to do it. Config changes will be pushed over the next few weeks. [17:20:00] ok [17:20:16] Well, it's not really a problem; worst case, people have more advance notice :) [18:48:45] random question: how do you pronounce gerrit? G+air+it or J-eh-rit? [18:49:49] G+air+it [18:50:50] rhymes with Get-rid-of-it [18:51:25] :P [19:05:17] I pronounce it the Italian way. More or less like Jerry + t [19:06:30] [[Tech]]; MR.DAYAT , THE BEST PARANORMAL IN THE WORLD.; /* MR. DAYAT CAN ALL....MR.DAYAT KNOWS ALL... MR. DAYAT CAN ALL... MR.DAYAT KNOWS ALL...20%20%20%20%20%20%20%20%20%. “MR. DAYAT . THE BEST PARANORMAL IN THE WORLD. MOST TRUE DIAGNOSES IN THE WORLD. THE CANDIDAT WHICH WILL GET UP 20 NOBELS ( ********...; [19:06:31] https://meta.wikimedia.org/w/index.php?diff=12537530&oldid=12513113&rcid=6572311 [19:06:45] [[Tech]]; Tegel; Reverted changes by [[Special:Contributions/MR.DAYAT , THE BEST PARANORMAL IN THE WORLD.|MR.DAYAT , THE BEST PARANORMAL IN THE WORLD.]] ([[User talk:MR.DAYAT , THE BEST PARANORMAL IN THE WORLD.|talk]]) to last version by MyHempBook; https://meta.wikimedia.org/w/index.php?diff=12537531&oldid=12537530&rcid=6572312 [19:08:08] overzealous snitch, you should trim [19:31:53] Reedy, earldouglas: the mediawiki-vagrant clone itself is the one that has those missing submodules (in the puppet dir). vagrant git-update doesn't mess with that top level clone, but the on-wiki instructions tell you to use `git clone --recursive` for the initial clone to ensure that you get the submodules (there are 4 that are shared with ops/puppet). [21:28:03] To get the analytics role working, do I need to do something extra to get oozie on my system? [21:28:09] It's blowing up when I try to provision: [21:28:11] ExitCodeException exitCode=2: gzip: /usr/lib/oozie/oozie-sharelib-yarn.tar.gz: No such file or directory [21:28:18] ==> default: Error: /Stage[main]/Cdh::Oozie::Server/Exec[oozie_sharelib_install]/returns: change from notrun to 0 failed: /usr/bin/oozie-setup sharelib create -fs hdfs://mediawiki-vagrant.dev -locallib /usr/lib/oozie/oozie-sharelib-yarn.tar.gz returned 1 instead of one of [0] [21:31:31] earldouglas: ask ottomata [21:55:40] earldouglas: you can also just try provisioning again. it's very likely the role or its modules are just missing an explicit dependency somewhere [21:56:15] No luck. [22:06:50] is there any DEFAULT value on page.page_content_model on the Wikimedia installation? [22:22:25] saper: NULL on enwiki [22:22:42] same on wikimania2015wiki (ie a new wiki) [22:24:18] Reedy: thanks, I can't reproduce https://phabricator.wikimedia.org/T64780 locally, plwikisource_p has "wikitext", my wiki imports it correctly [22:27:15] saper: now that $wgContentHandlerUseDB = true, page_content_model will be populated, usually as "wikitext" [22:27:46] https://gerrit.wikimedia.org/r/#/c/207353/ will let people change the content model of page from the UI [22:28:42] legoktm: when did that change? [22:29:05] late jan 2015 [22:30:49] ok, ankry reported that in jun 2015, so after $wgContentHandlerUseDB change [22:31:43] saper: the bug was filed Nov 21, 2014 [22:31:59] legoktm: yeah but https://phabricator.wikimedia.org/T64780#1368149 [22:32:33] right, it's no longer failing [22:32:38] just importing improperly [22:33:03] saper: I added https://phabricator.wikimedia.org/T72592 as a blocker [22:33:39] but why the import fails [22:34:19] page_content_model gets set to the revision content model of the latest revision - the latter is set to NULL since this is "this page's default" [22:34:55] The import code for content model stuff is probably rather untested [22:34:59] revision goes via Title and LinkCache to do getContentModel() [22:36:06] Reedy: I deployed proofread page here on git master and it imported fine, oldish 1.26 as well as the current master... [22:36:41] is MW using normal implementation of LinkCache or something else? [22:37:25] uh, should be the same thing [23:03:19] it's the final nulledit that has a wrong content model [23:04:06] MariaDB [plwikisource_p]> select rev_page,rev_id, rev_timestamp, rev_content_model from revision where rev_id=875936; [23:04:09] | rev_page | rev_id | rev_timestamp | rev_content_model | [23:04:12] | 320845 | 875936 | 20150616001339 | wikitext | [23:14:54] sometimes pages show "Templates used onthis page" section when I edit, and sometimes they don't. E.g. https://www.mediawiki.org/w/index.php?title=Help:Templates&action=edit does, but https://www.mediawiki.org/w/index.php?title=Etherpad_Lite&action=edit doesn't [23:15:56] the latter doesn't look to have any templates [23:16:41] Reedy: hi! dang, bad example :), let me see if I can find another. [23:22:21] Reedy: could that be that nulledit selects the last (freshly inserted) revision ID from master, but the followup attempt to get rev_content_model from slave returns nothing (yet)? [23:22:45] Doesn't sound impossible [23:22:54] Wouldn't be the first race condition like that [23:34:46] ah, I thought {{#tag:inputbox}} was a template or module, but it's a parser tag