mozilla :: #taskcluster

17 May 2017
12:04whimboogrenade|rust-training: https://bheisler.github.io/post/calling-rust-in-python/ fyi
12:50dustinoo rust training, cool :)
12:53dustintx for your help with crashtests btw, whimboo
12:53whimboodustin: i hope it helped a bit
12:54whimbooand sorry that i missed the gfx window part
12:54whimboowhle traveling and having the training I was not fully concentrated
12:57dustinno worries :)
13:04rwood"There was an error executing the Autloland request on try: 'ascii' codec can't encode character u'\u2014' in position 44: ordinal not in range(128)"
13:04rwoodprobably not a TC issue...
13:04garndtI think there is a bug about that
13:05garndtwhere did that error come up from?
13:05garndt(a non-tc bug that is)
13:05rwoodmozreview autoland
13:05rwoodok yeah, makes sense thx
13:05garndtno problem sir!
13:44Tomcat|mtgwcosta: https://treeherder.mozilla.org/logviewer.html#?job_id=99768185&repo=autoland is yours or ?
13:47wcostaTomcat|mtg: yeah, but nothing to do with my patch, let me retrigger decision task
13:49wcostaor maybe it has
13:51wcostadustin: any idea what's going on? it is https://reviewboard.mozilla.org/r/139818/diff/1/#
13:51wcostait worked on try
13:54gerard-majaxwcosta, hello
13:54gerard-majaxwcosta, I have been asked about the status of tc-gh with osx ? :)
13:57wcostagerard-majax: unfortunately migration is drawing all our resources, we didn't start even to talk about it
13:58wcostaTomcat|mtg: it is rather weird failure, as I saw no problem on try pushes, could you please back it out while I investigate what's going on?
13:59garndtgerard-majax: we have a worker that can work on os x, but nothing avaialble in terms of hardware/vms to run it on
14:01Tomcat|mtgwcosta: yeah will do
14:01Tomcat|mtgafter the meeting
14:17Tomcat|mtgwcosta: hmmm its passing in the next run
14:17Tomcat|mtgso should we leave this change in
14:17Tomcat|sheriffdutyyour call :)
14:18wcostaTomcat|sheriffduty: I think so, I tried to reproduce it on try, but it works
14:18wcostahttps://treeherder.mozilla.org/#/jobs?repo=try&revision=7a0e8c4ba278
14:18gerard-majaxgarndt, :(
14:19Tomcat|sheriffdutywcosta: ok
14:23dustinwcosta: did you get it sorted or still need a hand?
14:24wcostadustin: well, it feels like it magically self fixed
14:24dustinoh boy, funzies
14:26dustingawd I hate hg
14:38dustinhttps://irccloud.mozilla.com/pastebin/jczgegy2/
14:38dustinDIAF hg, DIAF
14:49philorthat seems more like hating httplib and our usage of it, both reasonable hatreds
14:51dustinyeah, still, wasting my time
14:52philorinconveniently, I think the hook still checks treestatus even if you push everything everywhere with CLOSED TREE in the commit message
15:04dustingerard-majax: https://github.com/taskcluster/taskcluster-rfcs/issues/14 btw
15:05gerard-majaxdustin, <3
15:05dustinthat should at least track discussions about it
15:55aselagea|builddutygarndt: hello
15:56aselagea|builddutygarndt: I wonder if you&#39;ve had a chance to look at https://bugzilla.mozilla.org/show_bug.cgi?id=1364955#c1? :)
15:56firebotBug 1364955 NEW, aselagea@mozilla.com Implement tooling to monitor queues for taskcluster jobs on on premise hardware
15:57garndtaselagea|buildduty: I sent an email about it, did you not get it?
15:57garndtsent on the 11th
15:57garndtto aselagea@mozilla.com <----is that right?
15:58* aselagea|buildduty checks again
16:01aselagea|builddutygarndt: do you mean the e-mail with the updates after our discussion last week?
16:02aselagea|builddutygarndt: I went through those steps, but it seems I lack pulse credentials
16:03aselagea|builddutyalso hit some errors, so I&#39;m not really sure if these are related
16:08dustinpulse credentials can be had easily from pulseguardian.mozilla.org
16:09dustinaselagea|buildduty: ^^
16:12aselagea|builddutydustin: I&#39;ve reached that page too, but I don&#39;t see how to get an access token
16:18dustincreate a new user
16:18dustinand set the password for it
16:19dustinthe big &quot;+&quot; in the upper right
16:19aselagea|builddutydustin: did that..I was guessing something else was still needed :)
16:19dustinnope
16:20aselagea|builddutygot it
16:20aselagea|builddutydustin: garndt: thanks
16:20garndtsorry, I commented on the bug about that before reading here
16:20garndtthe &quot;My Pulse Users&quot; => &quot;+&quot; is the way to go
16:20dustinI REALLY REALLY REALLY HATE TRY
16:20dustinapparently there&#39;s no way to trigger the mozharness lint job
16:20garndtdustin: that is a lie
16:20dustinjust. no. way.
17:06ochameaudustin: hi, did you saw my pull request for github statuses?
17:07dustinnope
17:08dustinoh, somehow I thought that was for your repo
17:08dustinhm, how is that protected from randomly changing status of anything?
17:09dustinoh, I see you asked that in the PR
17:10* dustin takes conversation there
17:16dustinochameau: also, awesome for the ptach :)
17:23ochameaudustin: it is actually for the devtools repo!
17:23ochameaudustin: do you think we should spawn a RFC right away?
17:23dustinlet&#39;s see what bstack says
17:23dustinwe&#39;d also like to support adding comments to PRs
17:23dustinand uploading files to releases
17:24dustinso i don&#39;t want to make those things awkward with a choice we make here
17:24dustinbut maybe we can think that through quickly between the bunch of us
17:24ochameauactually, I would like to add comment support at the same time
17:24ochameauI don&#39;t need release, for now
17:25bstackI&#39;ll take a look in a bit. The general idea seems great though
17:25dustingood -- permissions on that are a little harder!
17:51camdEli: thanks for the review, man. Fixes applied. :)
18:03ochameaudustin: do you know how to test a post request? I&#39;m testing got.post(url, {body: JSON.stringify({ state: &#39;error&#39; })}); but req.body in an empty object on the implementation side?
18:04ochameauthe only one post request in api.js doesn&#39;t seem to be tested
18:27Aryxwcosta: hi, the gecko decision task failed a few times more, going to back 1365350 out
18:29Aryxfurthermore, do we send everybody who pushed and those gecko decision task failed an email these days (got one)?
18:29wcostaAryx: ok, autoland?
18:29Aryxyes
18:29wcostaAryx: i am not sure how this works, but i got some emails
18:30wcostaI will look what&#39;s going on
18:30Aryxthank you
18:34garndtAryx: I do believe that decision tasks that fail because of a failure or exception email the owner of that task (the push user)
18:35Aryxdoubtful if they will retrigger the job on autoland or inbound. maybe restrict that to try?
18:38garndthrm, it does seem problematic for autoland and inbound eh? I would hope that they would respond to failure emails they get on those branches and ask around
18:38dustinochameau: hm, I didn&#39;t realize we were using got in testing
18:38garndtbut it might not be something they typically do...
18:39dustinochameau: oh, you should be using the generated API client
18:39dustinlike https://github.com/taskcluster/taskcluster-github/blob/master/test/api_test.js#L128 calls the `builds` method
18:40Aryxso far, i have never seen someone asking about that
18:58ochameaudustin: I&#39;m trying to do that, but req.body is still undefined. How does that work for post variables? the generated createStatus function excect three arguments coming from the route, and that&#39;s all
18:58dustindid you define a schema?
19:22ochameaudustin: I don&#39;t know what a shema is, so most likeky I don&#39;t!
19:36bhearsumhow are task dependencies filled out for a task like https://dxr.mozilla.org/mozilla-central/source/taskcluster/ci/push-apk-breakpoint/kind.yml? i see in https://tools.taskcluster.net/task-group-inspector/#/URGuaVTFTO2QAEXZmNsmgg/W4YCfKT1ScesLcKZ0G0o2g?_k=rthuo9 that it depends on two signing tasks, but i have no idea how those get linked up by the taskgraph generation code
19:54akibhearsum: kind-dependencies in taskcluster/ci/push-apk-breakpoint/kind.yml?
19:55bhearsumaki: how do those match up to the actual tasks though? i see &quot;build-signing&quot; there, but i don&#39;t know how that maps to actual tasks
19:55akibhearsum: taskcluster/taskgraph/util/push_apk.py:generate_dependencies()
19:56bhearsumah
19:59bhearsumthanks for the pointer
20:02bhearsumhm, but the tasks actually get passed in by https://dxr.mozilla.org/mozilla-central/source/taskcluster/taskgraph/transforms/push_apk_breakpoint.py#50
20:03akiyeah
20:03bhearsumwhich somehow comes from https://dxr.mozilla.org/mozilla-central/source/taskcluster/taskgraph/transforms/push_apk_breakpoint.py#20
20:04akiit&#39;s the decorator
20:04bhearsumyeah
20:05bhearsumand i think the data being passed in via https://dxr.mozilla.org/mozilla-central/source/taskcluster/taskgraph/generator.py#59
20:05* bhearsum prints his way to victory
20:05akia number of transforms are imported from util.push_apk
20:05akiyeah, i had to do that when i was getting too deep
20:06bhearsumyeah, but it looks like dependent-tasks gets passed into the transforms, which is what i&#39;m interested in
20:06bhearsumanyways, i think i can manage from here - thank yo!
20:06akinp
18 May 2017
No messages
   
Last message: 93 days and 21 hours ago