[01:43:31] <wikibugs>	 10Quarry: Gigantic query results cause a SIGKILL and the query status do not update - https://phabricator.wikimedia.org/T172086#3491791 (10zhuyifei1999) The current storing query status in the same process as the storing query results isn't going to work. SIGKILL cannot be caught, so only the celery master proce...
[19:17:52] <yuvipanda>	 halfak it turns out that if the IR Kernel installer fails, it'll just exit with code 0 and continue on happily :)
[19:18:12] <halfak>	 Oh!  So is there something borked about my "server"?
[19:19:05] <yuvipanda>	 halfak: nope! it's borked for everyone! It works for me because my homedir isn't clean
[19:19:10] <yuvipanda>	 halfak: I should have a fix soon
[19:20:07] <halfak>	 gotcha.  Cool.  Thanks :) 
[19:20:42] <yuvipanda>	 halfak: np!
[19:20:47] <yuvipanda>	 halfak: it's also at Python 3.6 now btw :)
[19:21:55] <halfak>	 Nice!  That'll make installs using wheels much easier. 
[19:22:00] <halfak>	 E.g. numpy and sklearn
[19:50:48] <yuvipanda>	 halfak: were there not 3.4 wheels for those?
[19:50:51] <yuvipanda>	 ah you had to manually upgrade pip first
[19:50:53] <yuvipanda>	 right
[19:50:59] <halfak>	 yuvipanda, there were yeah
[20:21:49] <yuvipanda>	 halfak: cool. Am trying to get automatic push-to-deploys set up for this, so in the future people can just send PRs to github.com/yuvipanda/paws (maybe a different org), and merging to master will automatically deploy
[20:22:23] <halfak>	 yuvipanda, deploy = execute all?
[20:22:30] <halfak>	 Or just make-visible
[20:23:02] <yuvipanda>	 halfak: ah, for the infrastructure itself. So if you wanted a new library in the image, for example
[20:23:15] <halfak>	 Oh!  I see
[20:23:53] <yuvipanda>	 halfak: yeah, so you'd just be able to make a PR to the repo, and then just merging there would autodeploy in 10-15mins
[20:25:20] <yuvipanda>	 halfak: mostly trying to make PAWS get to a self sustaining point, so it doesn't run into Quarry's issues ;)
[20:25:38] <halfak>	 Makes sense. 
[21:30:18] <zhuyifei1999_>	 yuvipanda: in case you get push-to-deploy, I'd be very curious on how it's done
[21:30:52] <yuvipanda>	 zhuyifei1999_: yeah, I think I'll be able to! I'm going to be using travis for it
[21:31:09] * zhuyifei1999_ will look
[21:31:56] <zhuyifei1999_>	 (I only used travis for linting and running tests)
[21:32:25] <zhuyifei1999_>	 also could you clarify 'self sustaining'?
[21:33:17] <yuvipanda>	 zhuyifei1999_: ah, that I'm not in the 'critical path' as much as possible!
[21:33:28] <yuvipanda>	 zhuyifei1999_: if someone is blocked by something, I want to be able to empower others to be able to fix these
[21:33:38] <zhuyifei1999_>	 oh yeah :/
[21:33:38] <yuvipanda>	 zhuyifei1999_: that wasn't entirely the case with current PAWS - it was very custom
[21:33:55] <yuvipanda>	 zhuyifei1999_: but new thing is a 'standard' install of https://z2jh.jupyter.org on a standard install of kubernetes (kubeadm)
[21:34:04] <zhuyifei1999_>	 ah
[21:34:13] <yuvipanda>	 so the possible list of 'upstreams' who can help is a lot larger, and there's very little 'translation' that needs to be done
[21:34:52] <zhuyifei1999_>	 yeah quarry is a ton of custom code, same as many other stuffs I work on
[21:35:36] <yuvipanda>	 yeah
[21:35:45] <yuvipanda>	 I think that's a real problem!
[21:35:51] <yuvipanda>	 (one of the reasons I was suggesting redash :D)
[21:39:45] <yuvipanda>	 halfak: R kernel should be up now
[21:40:23] <halfak>	 WORKS!
[21:40:25] <halfak>	 <3
[21:40:49] <yuvipanda>	 halfak: cool :) should also be a newer version!
[21:52:17] <yuvipanda>	 zhuyifei1999_: so https://github.com/yuvipanda/paws/blob/master/build.py builds the containers that need to be built, and pushes them to a registry. My next step is to have a simple webservice running on the paws cluster that can act as a web hook from travis for deployments
[21:52:35] <zhuyifei1999_>	 k
[21:52:39] <zhuyifei1999_>	 will read