[13:03:03] <zhuyifei1999_>	 @seen jynus
[13:03:03] <wm-bot>	 zhuyifei1999_: Last time I saw jynus they were quitting the network with reason: Quit: Konversation terminated! N/A at 5/26/2018 11:41:10 AM (1h21m52s ago)
[13:04:58] <wm-bot>	 I will let you know when I see jynus and I will deliver that message to them
[13:04:58] <zhuyifei1999_>	 @notify jynus Is there a way to direct a connection to the wiki replica servers which server it should connect to? The problem is with Quarry's SHOW EXPLAIN FOR feature that must be run on the same server as the executed query.
[13:08:41] <chicocvenancio>	 zhuyifei1999_: fixed the userhome PAWS bug!
[13:08:49] <zhuyifei1999_>	 nice
[13:08:57] <zhuyifei1999_>	 how did you fix it?
[13:09:08] <chicocvenancio>	 init-container running as root
[13:09:27] <chicocvenancio>	 it chowns the userhome and exits.
[13:09:54] <zhuyifei1999_>	 oh
[13:10:22] <chicocvenancio>	 https://github.com/yuvipanda/paws/pull/28/files
[13:11:10] <chicocvenancio>	 (merging won't deploy at this point, so I'm waiting for T195217 to merge it
[13:11:10] <zhuyifei1999_>	 nice
[13:11:10] <stashbot>	 T195217: Sane ingress methods for PAWS - https://phabricator.wikimedia.org/T195217
[13:11:40] <zhuyifei1999_>	 just one nitpick: can't command just be the chown without sh -c?
[13:13:03] <zhuyifei1999_>	 I don't have horizon write access to toolforge, unfortunately, so can't help with proxies and security groups :/
[13:13:35] <chicocvenancio>	 I think it can,(if memory serves me right I tried that successfully at one point) 
[13:14:22] <chicocvenancio>	 I kept `sh -c` because all the example in k8s docs use it
[13:14:26] <zhuyifei1999_>	 ok
[13:14:56] <chicocvenancio>	 I don't really have an opinion of what would be better
[13:23:24] <chicocvenancio>	 zhuyifei1999_:  do you think we should drop the sh -c?
[13:23:57] <zhuyifei1999_>	 I have a slight preference for without sh -c, but if the docs are all sh -c, I don't mind either way
[13:25:21] <chicocvenancio>	 the docs are not that comprehensive... in the end initContainers are just containers
[13:54:51] <chicocvenancio>	 removed the sh -c, added a -R
[14:18:03] <zhuyifei1999_>	 chicocvenancio: umm, I'm not sure if I'm very comfortable with -R. consider symlinks
[14:19:09] <chicocvenancio>	 -R and -h?
[14:19:49] <chicocvenancio>	 we could add the specific files that are necessary for jupyterhub operation as well
[14:20:30] <chicocvenancio>	 or just hope that nothing bungles the ownership of those files?
[14:21:54] <zhuyifei1999_>	 I'd go with the necessary files and with -h
[14:23:20] <chicocvenancio>	 we need the .local/share/jupyter/runtime directory to be acessible
[14:23:46] <zhuyifei1999_>	 is it created by the hub?
[14:24:06] <chicocvenancio>	 in general, if no one messes the ownership, it will be created correctly with `tools.paws` as owner
[14:24:25] <zhuyifei1999_>	 I mean, the reason the home dir is root owned is because of docker
[14:24:50] <zhuyifei1999_>	 but that shouldn't apply to any other directories
[14:25:08] <chicocvenancio>	 yes, and the fact that kubernetes hands directory creation for host volumes to docker
[14:25:09] <zhuyifei1999_>	 I don't think paws users can mess with ownership
[14:25:20] <chicocvenancio>	 you're right
[14:26:00] <chicocvenancio>	 ` command: ['chown', '52771:52771', '/home/paws']`then?
[14:26:05] <zhuyifei1999_>	 yeah
[14:28:18] <chicocvenancio>	 zhuyifei1999_:  this also means paws-beta is fully functional for all users as well
[14:28:24] <chicocvenancio>	 not sure what to think about that
[14:28:45] <zhuyifei1999_>	 nice
[14:29:31] <chicocvenancio>	 I thought about creating a notice there, or maybe restricting use a bit (maybe faster culler?, random killing of servers?)
[14:29:48] <chicocvenancio>	 not sure we want it to be heavily used
[14:30:16] <chicocvenancio>	 this bug effectively meant we had to chown the userhome for it to spawn a server for anyone
[14:30:28] <chicocvenancio>	 but now it will just work
[14:30:43] <chicocvenancio>	 albeit with a much smaller cluster than PAWS itself
[14:30:44] <zhuyifei1999_>	 random killing of servers <= as long is it won't kill those who are dev-ing paws itself
[14:31:27] <chicocvenancio>	 we could use a script to kill via the api and spare admins
[14:32:55] <zhuyifei1999_>	 how about using an OAuth consumer that can't edit?
[14:33:12] <zhuyifei1999_>	 (or is that already the case? /me looks)
[14:33:24] <chicocvenancio>	 nope, it has the same permissions as paws
[14:35:12] <chicocvenancio>	 we could do that, but it would limit a bit our ability to test things, and PAWS can be used without oauth (or with a different OAuth consumer)
[14:36:23] <zhuyifei1999_>	 hmm
[14:37:03] <zhuyifei1999_>	 random culling isn't too bad I guess
[14:37:59] <zhuyifei1999_>	 the test install of v2c has its logic mostly broken (without messing with the return url)
[14:38:26] <chicocvenancio>	 v2c?
[14:40:03] <zhuyifei1999_>	 video2commons, the web tool I made from scratch
[14:41:16] <zhuyifei1999_>	 other web tools of mine are mostly forked or simple (as in CGI) or broken
[14:41:25] <zhuyifei1999_>	 or taken-over
[16:09:46] <chicocvenancio>	 zhuyifei1999_: do you want help with v2c? 
[16:24:09] <zhuyifei1999_>	 chicocvenancio: yeah, definitely
[16:24:57] <zhuyifei1999_>	 you helping?
[16:31:06] <chicocvenancio>	 Sure! Add me to the project, point me at what needs attention 
[16:56:43] <zhuyifei1999_>	 matanya, andrewbogott: could you add chicocvenancio to the video project?
[16:57:32] <andrewbogott>	 zhuyifei1999_: sure, one minute...
[16:58:20] <chicocvenancio>	 Chico Venancio is the ldap account
[16:58:23] <zhuyifei1999_>	 chicocvenancio: I invited you to https://github.com/orgs/toolforge/people
[16:58:48] <zhuyifei1999_>	 (I'll get the toolforge tool membership in a sec)
[16:59:10] <andrewbogott>	 zhuyifei1999_: is there a reason you're not a project admin there?
[16:59:22] <zhuyifei1999_>	 idk
[16:59:33] <zhuyifei1999_>	 because I never asked for it?
[16:59:48] <andrewbogott>	 ok — you and chico are both users now.  I suggest you ask Matanya to make y'all admins if you're going to be active there :)
[17:00:28] <zhuyifei1999_>	 hey, half of v2c lives there. of course I'm active :)
[17:02:23] <zhuyifei1999_>	 chicocvenancio: I added you as maintainers of all three tools in toolforge
[17:03:31] <zhuyifei1999_>	 the backend code is trash :( some translated from php TimedMediaHandler code
[17:04:13] <zhuyifei1999_>	 I was in a middle of a rewrite (for resume-able tasks) and mostly gave up
[17:06:43] <zhuyifei1999_>	 I'll see if I can redo that rewrite next week. Could you review?
[17:07:29] <chicocvenancio>	 👍
[17:08:29] * zhuyifei1999_ goes zzz. needs to fix some pywikibot madness tomorrow
[18:57:48] <legoktm>	 I'm wondering what directory layouts people use when deploying PHP tools
[18:58:06] <legoktm>	 AIUI it's supposed to be in public_html/, so people symlink that to another directory?
[19:01:07] <legoktm>	 hm that's what bd808's tool does, so I'll copy that :)
[19:01:34] <chicocvenancio>	 legoktm: kubernetes or grid?
[20:38:30] <wikibugs_>	 (03PS1) 10Krinkle: Add slice-level timing for revision matching [labs/tools/guc] - 10https://gerrit.wikimedia.org/r/435653 (https://phabricator.wikimedia.org/T195515)
[20:42:59] <wikibugs_>	 (03CR) 10Krinkle: [C: 032] Add slice-level timing for revision matching [labs/tools/guc] - 10https://gerrit.wikimedia.org/r/435653 (https://phabricator.wikimedia.org/T195515) (owner: 10Krinkle)
[20:43:30] <wikibugs_>	 (03Merged) 10jenkins-bot: Add slice-level timing for revision matching [labs/tools/guc] - 10https://gerrit.wikimedia.org/r/435653 (https://phabricator.wikimedia.org/T195515) (owner: 10Krinkle)
[20:53:05] <legoktm>	 chicocvenancio: k8s probably
[21:13:47] <legoktm>	 I think I've figured it out
[22:31:30] <legoktm>	 https://tools.wmflabs.org/coverme/ it works ^.^