| openstackgerrit | Clint 'SpamapS' Byrum proposed openstack-infra/zuul feature/zuulv3: Slack driver https://review.openstack.org/530521 | 00:29 |
|---|---|---|
| openstackgerrit | Clint 'SpamapS' Byrum proposed openstack-infra/zuul feature/zuulv3: Slack driver https://review.openstack.org/530521 | 00:36 |
| SpamapS | Hrm.. this is an interesting conundrum. I want to use Slack threading to report all the things related to a commit into a single thread. I guess I can keep a commit<->thread_ts cache in the driver. | 00:45 |
| tristanC | SpamapS: congrats on zuulception :-) Did you find a way to auto restart the scheduler? | 01:02 |
| tristanC | perhaps we could add a graceful stop to the scheduler so that we can start another one with the new code... | 01:03 |
| *** xinliang has joined #zuul | 01:29 | |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul-jobs master: DNM: negative test https://review.openstack.org/522438 | 01:43 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul-jobs master: DNM: negative test https://review.openstack.org/522438 | 01:47 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul-jobs master: unittests: make bindep role optional https://review.openstack.org/522255 | 01:54 |
| SpamapS | tristanC: No I didn't. I just do it manually. | 03:00 |
| SpamapS | tristanC: what we really need is a SIGUSR2 or something that means "save queue, re-exec, slurp queue, after all running jobs finish" | 03:01 |
| *** jaianshu has joined #zuul | 04:12 | |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul-jobs master: Add buildset-artifacts-location role https://review.openstack.org/530679 | 04:49 |
| *** bhavik1 has joined #zuul | 05:31 | |
| *** bhavik1 has quit IRC | 05:34 | |
| tobiash | SpamapS: currently I have the scheduler as a statefulset | 05:41 |
| tobiash | And no lifecycle stuff yet | 05:42 |
| tobiash | The plan is to at least save queues periodically on a volume and reenqueue the items on startup | 05:43 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul-jobs master: Add linter job and roles https://review.openstack.org/530682 | 05:55 |
| *** jaianshu_ has joined #zuul | 06:16 | |
| *** jaianshu has quit IRC | 06:20 | |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul-jobs master: Add buildset-artifacts-location role https://review.openstack.org/530679 | 06:20 |
| openstackgerrit | Clint 'SpamapS' Byrum proposed openstack-infra/zuul feature/zuulv3: Slack driver https://review.openstack.org/530521 | 06:21 |
| SpamapS | jhesketh: ^ your config concerns were spot on. There's an updated version that makes more sense I think. | 06:21 |
| openstackgerrit | Clint 'SpamapS' Byrum proposed openstack-infra/zuul feature/zuulv3: Slack driver https://review.openstack.org/530521 | 06:27 |
| SpamapS | tobiash: ^ Also I addressed your review comments there. Thanks for the feedback. | 06:27 |
| SpamapS | tobiash: Ah I'm so used to not having statefulset .. that should work fine. | 06:29 |
| * SpamapS has a super old k8s :-P | 06:29 | |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul-jobs master: Add buildset-artifacts-location role https://review.openstack.org/530679 | 06:32 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul-jobs master: Add linter job and role https://review.openstack.org/530682 | 06:41 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul-jobs master: Fix pep8/yamllint errors https://review.openstack.org/530685 | 06:43 |
| *** hashar has joined #zuul | 08:29 | |
| *** yolanda has quit IRC | 09:13 | |
| *** yolanda has joined #zuul | 09:16 | |
| *** jaianshu__ has joined #zuul | 10:15 | |
| *** jaianshu_ has quit IRC | 10:18 | |
| openstackgerrit | Fabien Boucher proposed openstack-infra/zuul feature/zuulv3: Make Zuul able to start with a broken config https://review.openstack.org/529060 | 11:29 |
| *** jkilpatr has joined #zuul | 12:07 | |
| *** rlandy has joined #zuul | 13:07 | |
| *** rlandy is now known as rlandy|ruck | 13:14 | |
| *** jaianshu__ has quit IRC | 13:33 | |
| *** dkranz has quit IRC | 13:57 | |
| *** maxamillion has joined #zuul | 14:34 | |
| *** pall is now known as pabelanger | 14:45 | |
| *** dkranz has joined #zuul | 14:59 | |
| *** rlandy|ruck is now known as rlandy|ruck|mtg | 15:00 | |
| *** jaianshu has joined #zuul | 15:37 | |
| *** rlandy|ruck|mtg is now known as rlandy|ruck | 16:01 | |
| *** kmalloc has joined #zuul | 16:09 | |
| Shrews | nodepool git history seems to indicate we had issues with the new quota feature. Did that get resolved? | 16:19 |
| clarkb | Shrews: yes I think it did. | 16:20 |
| Shrews | cool | 16:20 |
| *** rlandy|ruck is now known as rlandy|ruck|mtg | 16:21 | |
| clarkb | Shrews: I466e64a999ebf867ada9462a0c00a90de4b70125 and Id488c070df991a554570c5717dc85aec351fed45 were both changes to address the problems we had | 16:21 |
| corvus | i think there's one thing that's still reverted out -- we now mark requests as failed if they hit an unexpecetd over-quota error | 16:22 |
| corvus | that was also the old behavior | 16:22 |
| Shrews | yeah, was just looking those over | 16:22 |
| corvus | the new quota patches changed that to make it a non-fatal error, but there were structural issues with that, so we just went back to the old behavior | 16:22 |
| corvus | (i'm sure it can be fixed, just needs some thought) | 16:23 |
| Shrews | clarkb: corvus: thx for the catchup | 16:24 |
| *** jaianshu has quit IRC | 17:01 | |
| *** jaianshu has joined #zuul | 17:01 | |
| SpamapS | btw I am finding nodepool to be a real joy to work with. I haven't been part of infra running it, so it's all new to me, and I find myself wishing all my apps were backed by nodepool. | 17:06 |
| SpamapS | I actually think nodepool might be a really great way to run Kubernetes clusters actually. | 17:06 |
| SpamapS | so many actuallies | 17:07 |
| *** rlandy|ruck|mtg is now known as rlandy|ruck | 17:07 | |
| pabelanger | +1 | 17:08 |
| corvus | SpamapS: \o/ i think the request model is going to open a lot of doors in the future | 17:08 |
| SpamapS | Just the idea of actually treating cloud instances like the ephemeral things that they're so good at being... with pools, mins, builders going on in the background. It's a nice experience. | 17:09 |
| corvus | i looked over the todo list, and i'm going to start work on the 'support cross-source dependencies' task. so i'm going to be spending big chunks of time focusing on that. | 17:10 |
| SpamapS | That was the one that scared me the most so I stayed away from it. ;) | 17:10 |
| corvus | i think i'm ready for it. i can do this. i know i can. | 17:11 |
| clarkb | I've not even made caffeine yet | 17:11 |
| corvus | i had some yesterday. that should be enough. | 17:11 |
| clarkb | I found ginger mate tea | 17:12 |
| *** jaianshu has quit IRC | 18:10 | |
| *** harlowja has joined #zuul | 18:12 | |
| Shrews | corvus: If you get a moment, I'm very confused as to why https://review.openstack.org/520664 does not run the new nodepool-functional-devstack-py35 job I'm trying to define. | 18:13 |
| corvus | Shrews: my guess: devstack has no feature/zuulv3 branch so there's no matching devstack job for that branch.... | 18:25 |
| openstackgerrit | James E. Blair proposed openstack-infra/nodepool feature/zuulv3: WIP: Convert from legacy to native devstack job https://review.openstack.org/520664 | 18:25 |
| corvus | Shrews: but we added a neat thing to zuul in december so you can ask it questions like that: ^ | 18:26 |
| Shrews | neat | 18:26 |
| corvus | Shrews: so, erm, we'll have confirmation of that in 40 mins :) | 18:26 |
| Shrews | corvus: fwiw, the fingergw didn't automatically start (was it supposed to?) so i started it by hand on zuulv3 | 18:27 |
| openstackgerrit | James E. Blair proposed openstack-infra/nodepool feature/zuulv3: WIP: Convert from legacy to native devstack job https://review.openstack.org/520664 | 18:27 |
| corvus | Shrews: ^ i dropped all the other jobs so we should get the answer faster | 18:27 |
| Shrews | ++ | 18:27 |
| corvus | Shrews: that sounds right. i like to avoid having puppet auto-start services. but it'll be configured to start on boot. | 18:28 |
| corvus | Shrews: thanks for doing that. does it work? :) | 18:28 |
| Shrews | corvus: looks like we might need to open access to port 79 | 18:28 |
| corvus | oh yep | 18:28 |
| corvus | Shrews: i think that's in system-config if you want to take that | 18:28 |
| Shrews | sure | 18:28 |
| corvus | Shrews: meanwhile, i'll think about how to solve that devstack job problem. | 18:29 |
| Shrews | corvus: thx. i still find the way zuul handles branch jobs perplexing. i should spend some time studying that | 18:31 |
| corvus | Shrews: the results are in on https://review.openstack.org/#/c/520664/ | 18:38 |
| Shrews | corvus: yeah, trying to decipher the results | 18:39 |
| corvus | Shrews: look at the review message zuul left on there (if you're using the web ui, you may need to hit toggleci; if you're using gertty, you may want to use the webui because of the superlonglines) | 18:39 |
| Shrews | "No matching parent for 'devstack'" | 18:40 |
| corvus | Shrews: yeah, and the lines above show the branches where there is a 'devstack' defined. so i think that confirms the hypothesis | 18:40 |
| Shrews | shouldn't that be found in openstack-dev/devstack/.zuul.yaml@master ? | 18:40 |
| corvus | Shrews: it is, but that only matches changes on master, not feature/zuulv3 | 18:41 |
| corvus | Shrews: (so, iow, one way to 'solve' this is to merge nodepool feature/zuulv3 into nodepool master) | 18:41 |
| Shrews | corvus: so if i derive from another job in another repo, that other repo has to have the same branch? | 18:42 |
| corvus | Shrews: if the other repo is branched, yes | 18:42 |
| Shrews | TIL | 18:42 |
| corvus | this doesn't apply to project-config, zuul-jobs, etc | 18:42 |
| Shrews | mordred: fyi, since we were both confused about that native devstack job ^^^ | 18:43 |
| corvus | but it's what makes "this is how to run devstack on pike" and "this is how to run devstack on ocata" work. | 18:43 |
| Shrews | corvus: ok. i can pick that back up when we merge back to master then | 18:43 |
| Shrews | that makes sense | 18:43 |
| corvus | Shrews: we should probably come up with another solution though, just to solve this use case... | 18:43 |
| openstackgerrit | Andreas Jaeger proposed openstack-infra/zuul-jobs master: Fix sphinx publishing https://review.openstack.org/530790 | 18:44 |
| corvus | maybe we need to have a fallback branch when looking for parent jobs. or maybe we need to drop the implicit branch matcher from the master branch of devstack. | 18:44 |
| Shrews | corvus: a 'parent-branch' option? | 18:44 |
| Shrews | to be explicit about it | 18:45 |
| corvus | Shrews: yeah -- that would be basically an explicit fallback branch for parents...we could also consider having an implicit fallback branch... | 18:45 |
| *** jkilpatr has quit IRC | 18:46 | |
| corvus | i'm not coming up with a reason not to do a fallback branch when searching for parents. it needs to only happen if there are no matches (ie, you don't always want to use master if, say, pike matches). but other than that i think it would be okay. | 18:49 |
| Shrews | seems like logical behavior to me, IMO | 18:51 |
| corvus | there is a way we can make this work now, though it requires a (benign) change to the devstack repo... | 18:52 |
| corvus | er, i'd link to the docs, but they haven't been updated due to the bug being discussed in #-infra | 18:53 |
| Shrews | corvus: it's not urgent to get it working now. just some cleanup that eventually should be done | 18:54 |
| corvus | http://logs.openstack.org/05/527805/2/gate/build-openstack-sphinx-docs/67ee106/html/user/config.html#attr-pragma.implied-branches | 18:54 |
| Shrews | i'm just happy to understand the problem now | 18:54 |
| corvus | Shrews: there's the draft docs build for the change i'm thinking of ^ | 18:54 |
| corvus | if we added ['master', 'feature/zuulv3'] to implied-branches in the devstack repo, the job would work. however, i don't think we should do that -- devstack has no business knowing there's such a thing as a feature/zuulv3 branch. | 18:55 |
| corvus | i think adding an explicit branch matcher to devstack that matches master or a negative lookahead which matches everything except stable/* would be okay though. | 18:56 |
| Shrews | corvus: yeah, bad knowledge inversion there | 18:56 |
| corvus | i'm inclined to think that either that ^ or the fallback branch is the way to solve it. | 18:56 |
| Shrews | ++ for fallback (either implicit or explicit or both) | 18:57 |
| corvus | i'll cipher on this a little more, then maybe write up a description | 18:58 |
| *** jkilpatr has joined #zuul | 19:00 | |
| *** JasonCL has joined #zuul | 19:36 | |
| *** jkilpatr has quit IRC | 20:27 | |
| *** dkranz has quit IRC | 20:37 | |
| *** dkranz has joined #zuul | 20:38 | |
| *** jkilpatr has joined #zuul | 20:41 | |
| *** JasonCL has quit IRC | 21:01 | |
| *** jkilpatr has quit IRC | 21:04 | |
| *** jlk` has joined #zuul | 21:05 | |
| *** jlk has quit IRC | 21:06 | |
| *** JasonCL has joined #zuul | 21:13 | |
| *** jlk` is now known as jlk | 21:15 | |
| *** jlk has quit IRC | 21:16 | |
| *** jlk has joined #zuul | 21:16 | |
| *** jkilpatr has joined #zuul | 21:33 | |
| *** dkranz has quit IRC | 21:43 | |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: WIP: Support cross-source dependencies https://review.openstack.org/530806 | 22:21 |
| SpamapS | pabelanger: btw, my memory governor patches conflict badly with recent executor changes, I basically have to rewrite half of it from scratch. Haven't found time to do that just yet. | 22:33 |
| pabelanger | kk, I was going to look into it in the morning | 22:34 |
| pabelanger | so much push up a patchset if I can figure it out :) | 22:34 |
| SpamapS | It's a simple patch. | 22:46 |
| SpamapS | Just that things got moved around a lot and git is confused. | 22:46 |
| SpamapS | Also we need to switch the virtual memory total we use so that we don't count cache/buffer as "used" so we don't get thrashing. | 22:47 |
| SpamapS | (Though I'd argue we should consider having a third governor and not accepting jobs if we're really low on cache/buffer too) | 22:47 |
| openstackgerrit | Clark Boylan proposed openstack-infra/zuul-jobs master: Fix sphinx publishing https://review.openstack.org/530790 | 22:49 |
| *** hashar has quit IRC | 23:05 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!