[02:00:08] !log wikitextexp Added BryanDavis (self) as project admin for T211367 [02:00:10] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Wikitextexp/SAL [02:00:11] T211367: Setting up bulk proxies for two multi-wiki mw-vagrant labs vms - https://phabricator.wikimedia.org/T211367 [16:28:36] hey, so here is thifranc ^ he would like to volunteer and write puppet code. that is pretty cool. he asked how to get a cloud VPS to test puppet roles. i said i'd ask, would be nice if we can grant one in maybe the "puppet" project ..or another one. [16:29:21] hi all :) [16:30:03] (or should he request a project.. , i am not sure, but "puppet" sounds right for this) [16:31:24] thifranc, do you have a particular area you want to develop stuff for? [16:32:26] (operations/puppet) [16:32:40] so far I haven't contributed, I'm on to contribute to puppet though [16:33:01] well, we have one of your changes in Gerrit that will merge soon [16:33:05] the first one [16:36:05] I think adding him to the 'puppet' project is fine — best to check with at least one other projectadmin there first though. [16:36:09] anyone running into trouble using mw.Api().saveOption on mediawiki-vagrant? [16:36:22] well, if you are one already then there you have it :) [16:36:26] mw.loader.using( 'mediawiki.api' ).then( function () { new mw.Api().saveOption( 'eventlogging-display-web', '1' ); } ); [16:37:23] it seems to persist if I run it on production wikis, but if I do it on vagrant, it doesn't throw any errors in the logs but seems to not work [16:39:33] doh - user error, wasn't logged in [16:41:07] it throws an error and everything, very nice: (new mw.Api().saveOption( 'eventlogging-display-web', '1' )).fail((e) => console.log(e)); [16:41:28] andrewbogott: ok, thanks. yes, will do that [16:41:37] (check with at least one other and then..) [16:42:05] adding as a member still means ssh (without root) to all existing VMs, right [16:42:37] needs permission to run puppet.. but ideally just on one machine [16:42:57] i guess that would mean new project though, and i dont want to abuse resources. so that seems the conflict [16:43:11] (never sure if "project" actually "costs" a lot or is just a label) [16:43:23] as opposed to the same number in VMs in the same project [16:49:01] mutante: yes, ssh to all machines and probably sudo as well [16:49:18] Depending on the sudo policy for that project, but the default is that everyone gets it [16:49:56] oh.. that too. yea, hm [16:51:17] well, without it no running puppet agent, so it must be [16:52:34] mutante, adding a project member by default in most projects will give them root on all VMs [16:52:40] by sshing in as themselves and running sudo [16:53:33] a project is a sort of concept, it has no inherent cost that I am aware of. the instances that it contains will effectively be cost [16:55:06] I would say don't open a new project request until you've checked with other project members that the existing project is a non-option [16:56:53] from the technical point of view it seems better to "just" create a project and give him a VM inside it ..claiming it is also easy to remove again. but at the same time we do want projects organized by topic area of course and not personal ones. unfortunately i can see both sides [16:57:54] Krenair: ok, it sounds actually like a mail to ops list if anyone has concerns ..maybe [17:01:18] we would need another layer.. commom project by topic area and still separate SSH access by "prefix".. i would use that [17:02:45] mutante, so far you haven't convinced me there needs to be any extra access control on this [17:05:58] Krenair: i am not interested in it per se, but that would remove the need to find consensus between all project members to add one and also it's just the admin in me thinking "minimum necessary to do what they actually want" [17:06:20] now i have to ask everybody and then i could just do it.. that was basically it [18:05:32] !log integration moving integration-slave-docker-1040 to eqid1-r [18:05:34] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Integration/SAL [18:07:07] !log integration moving integration-slave-docker-1037 to eqiad1-r [18:07:08] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Integration/SAL [19:33:35] !log integration moving integration-puppetmaster01 to eqiad1-r [19:33:37] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Integration/SAL [19:45:15] !log integration moving saucelabs-01 and saucelabs-02 to eqiad1-r [19:45:16] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Integration/SAL [20:33:42] !log integration moving webperformance to eqiad1-r [20:33:44] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Integration/SAL [20:36:28] !log integration moving integration-r-lang-01 to eqiad1-r [20:36:29] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Integration/SAL [21:17:14] !log ircd deleting project; abandoned. [21:17:15] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Ircd/SAL [21:18:48] !log nonfreewiki deleting project; abandoned. [21:18:48] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Nonfreewiki/SAL [21:30:11] RoanKattouw: your VM visualeditor-test2.visualeditor has filled its drive — can you do some cleanup there so that puppet has room to move? [21:41:22] !log automation-framework rebooting af-puppetmaster02 to get puppet working again [21:41:24] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Automation-framework/SAL [22:57:30] andrewbogott: I freed up some space [22:57:36] And I think I see what's going on *sigh* [22:58:26] The eslint-plugin-json npm module is pulled in by several services on this box, and it's started pulling in a Unicode package that's 121MB [23:02:27] RoanKattouw: Whoops. [23:02:41] RoanKattouw: Is the box running npm 5+? [23:02:50] I'm now doing rm -rf node_modules ; npm install for Citoid to see if that pulls in something more reasonable [23:04:16] Ugh the host is running npm 1.4.21 and the guest is running 2.15.12 [23:04:26] But the guest isn't able to resolve github.com :( [23:05:01] That's not great. Newer npms can de-dupe.