| *** jamesmcarthur has quit IRC | 01:09 | |
| *** bhavikdbavishi has joined #zuul | 02:56 | |
| *** bhavikdbavishi has quit IRC | 03:21 | |
| *** bhavikdbavishi1 has joined #zuul | 03:21 | |
| *** bhavikdbavishi1 is now known as bhavikdbavishi | 03:23 | |
| tobiash | SpamapS: I've done only a proof of concept so far with aws but probably will get back to it later this year | 03:46 |
|---|---|---|
| tobiash | We'll need real quota handling for aws which will be totally different and challenging because of quota per instane type | 03:47 |
| tobiash | The static driver is a similar challenge innthis regard because of multi label support | 03:49 |
| SpamapS | tobiash: I have some half-done code that will use reservations for quota management. | 04:10 |
| SpamapS | tobiash: the idea being that at least we catch instance quotas up front, and not-zuul can't take what we need. | 04:12 |
| SpamapS | Also, wtf AWS for not having an API for limits. | 04:13 |
| SpamapS | Also kind of want to add spot price capability | 04:28 |
| SpamapS | seems like it could save a lot of money since most of the time jobs will be entirely interruptible. | 04:28 |
| *** bjackman_ has joined #zuul | 04:30 | |
| *** bjackman__ has joined #zuul | 04:35 | |
| *** bjackman_ has quit IRC | 04:37 | |
| *** bjackman_ has joined #zuul | 04:39 | |
| *** bjackman__ has quit IRC | 04:41 | |
| *** quiquell|off is now known as quiquell|rover | 05:43 | |
| *** pcaruana has joined #zuul | 06:30 | |
| *** snapiri has quit IRC | 06:33 | |
| *** quiquell|rover is now known as quique|rover|brb | 06:41 | |
| *** quique|rover|brb is now known as quiquell|rover | 07:13 | |
| *** jangutter has quit IRC | 07:15 | |
| *** jangutter_ has joined #zuul | 07:15 | |
| *** SpamapS has quit IRC | 07:15 | |
| *** shanemcd has quit IRC | 07:16 | |
| *** shanemcd has joined #zuul | 07:16 | |
| *** SpamapS has joined #zuul | 07:17 | |
| *** sshnaidm|off is now known as sshnaidm|pto | 07:23 | |
| *** gtema has joined #zuul | 07:23 | |
| *** themroc has joined #zuul | 07:46 | |
| *** jpena|off is now known as jpena | 07:47 | |
| *** frickler has quit IRC | 07:58 | |
| *** frickler has joined #zuul | 07:59 | |
| mhu | Hello | 08:09 |
| *** johanssone has quit IRC | 08:18 | |
| *** johanssone has joined #zuul | 08:24 | |
| *** zbr has quit IRC | 08:28 | |
| *** hashar has joined #zuul | 08:54 | |
| *** bhavikdbavishi1 has joined #zuul | 09:28 | |
| *** bhavikdbavishi has quit IRC | 09:29 | |
| *** bhavikdbavishi1 is now known as bhavikdbavishi | 09:29 | |
| *** electrofelix has joined #zuul | 09:41 | |
| bjackman_ | mordred, Any chance you could take a look at https://review.openstack.org/#/c/649900/ ? | 09:42 |
| bjackman_ | It's the fix for the gerrit<->Zuul trigger bug I was looking into last week | 09:43 |
| *** yolanda_ has quit IRC | 09:47 | |
| *** bhavikdbavishi has quit IRC | 10:35 | |
| *** zbr has joined #zuul | 11:08 | |
| openstackgerrit | Matthieu Huin proposed openstack-infra/zuul master: [WIP] admin REST API: docker-compose PoC, frontend https://review.openstack.org/643536 | 11:12 |
| *** jpena is now known as jpena|lunch | 11:37 | |
| *** quiquell|rover is now known as quique|rover|eat | 12:07 | |
| *** hashar has quit IRC | 12:07 | |
| *** hashar has joined #zuul | 12:08 | |
| *** bhavikdbavishi has joined #zuul | 12:09 | |
| *** jamesmcarthur has joined #zuul | 12:17 | |
| *** hashar has quit IRC | 12:24 | |
| *** jpena|lunch is now known as jpena | 12:29 | |
| *** rlandy has joined #zuul | 12:32 | |
| *** rlandy is now known as rlandy|ruck | 12:32 | |
| *** jamesmcarthur has quit IRC | 12:34 | |
| *** openstackgerrit has quit IRC | 12:44 | |
| *** quique|rover|eat is now known as quiquell|rover | 12:44 | |
| *** bjackman_ has quit IRC | 12:48 | |
| *** jamesmcarthur has joined #zuul | 12:50 | |
| *** openstackgerrit has joined #zuul | 12:58 | |
| openstackgerrit | Merged openstack-infra/zuul-jobs master: Allow user to set ara_report_executable https://review.openstack.org/648773 | 12:58 |
| *** jamesmcarthur has quit IRC | 13:04 | |
| *** hashar has joined #zuul | 13:19 | |
| *** quiquell|rover is now known as quiquell|off | 14:38 | |
| *** jamesmcarthur has joined #zuul | 14:47 | |
| *** hashar has quit IRC | 15:09 | |
| *** bjackman has joined #zuul | 15:21 | |
| *** bjackman_ has joined #zuul | 15:22 | |
| *** gtema has quit IRC | 15:23 | |
| *** bjackman has quit IRC | 15:26 | |
| *** jamesmcarthur has quit IRC | 15:42 | |
| *** bjackman_ has quit IRC | 15:48 | |
| Shrews | SpamapS: omg https://review.openstack.org/642964 is giving me a headache trying to merge that | 16:01 |
| Shrews | all the faulty/racey tests are finding you :( | 16:03 |
| *** jamesmcarthur has joined #zuul | 16:05 | |
| clarkb | zuulians, I'd like to approve https://review.openstack.org/#/c/650968/ now. If we see that that breaks the jobs on ipv4 I am 100% ok with a revert | 16:07 |
| clarkb | mostly pointing that out here so that people know they can revert it immediately without hurting any feelings if we see that it breaks | 16:08 |
| clarkb | with that out of the way /me +A's | 16:08 |
| Shrews | clarkb: which repos could that impact? just zuul and nodepool? | 16:09 |
| clarkb | Shrews: and some opendev stuff for gitea | 16:09 |
| clarkb | I am not aware of anyone else using it yet | 16:10 |
| Shrews | ah right | 16:10 |
| clarkb | unfortunately there isn't a good way to test it and its relatively new and its broken anyway | 16:10 |
| clarkb | just not 100% broken | 16:10 |
| Shrews | yup | 16:10 |
| *** themroc has quit IRC | 16:14 | |
| *** jamesmcarthur has quit IRC | 16:23 | |
| openstackgerrit | Merged openstack-infra/zuul-jobs master: Use /etc/hosts with docker registries to handle ipv6 https://review.openstack.org/650968 | 16:23 |
| *** bjackman_ has joined #zuul | 16:26 | |
| openstackgerrit | Clark Boylan proposed openstack-infra/zuul master: Don't cache wheels in dockerfile https://review.openstack.org/651322 | 16:29 |
| clarkb | Shrews: ^ I don't know if that is something we actualyl want to merge, but it seemed like it was possibly such a thing and should cause us to run through the docker image build system | 16:30 |
| *** bjackman_ has quit IRC | 16:34 | |
| clarkb | hrm we may have to approve that change (then immediately unapprove it) so that it goes into the gate? | 16:34 |
| clarkb | I don't think the check job pushes the image | 16:34 |
| electrofelix | battling with centos 7 and proxy settings is enough to basically make me go, this is not a sensible approach to be messing around with when it requires learning how someone elses deployment approach works | 16:40 |
| electrofelix | oops, sorry wrong room | 16:40 |
| *** jamesmcarthur has joined #zuul | 16:45 | |
| mordred | electrofelix: it's ok - i'm sure we all agree with you anyway :) | 16:48 |
| electrofelix | should I mention how proxies are the bane of my life... | 16:49 |
| *** jpena is now known as jpena|away | 16:50 | |
| SpamapS | Shrews: that whole kubernetes library seems like a fault. | 16:56 |
| *** electrofelix has quit IRC | 16:58 | |
| clarkb | Shrews: mordred http://logs.openstack.org/17/650917/1/gate/zuul-upload-image/555d041/job-output.txt.gz#_2019-04-09_16_48_39_332294 | 17:03 |
| clarkb | I think that must be an order of operations thing? | 17:04 |
| clarkb | except that use-buildset-registry is the previous role | 17:04 |
| mordred | clarkb: oh. well, that's unfortunate | 17:04 |
| clarkb | http://logs.openstack.org/17/650917/1/gate/zuul-upload-image/555d041/job-output.txt.gz#_2019-04-09_16_48_24_974566 and the same task worked there? | 17:04 |
| * Shrews returns from lunch | 17:05 | |
| clarkb | ok the task that failed ran on localhost | 17:06 |
| clarkb | the one that succeeded ran on ubuntu-bionic | 17:06 |
| clarkb | I'm guessing that 1) we can't edit /etc/hosts on localhost at all even if that condition didn't break and 2) we don't set buildset_registry on localhost? | 17:07 |
| clarkb | ok I see the bug for why it didn't load | 17:07 |
| clarkb | meaning I know how to fix 2) but I don't have any quick ideas for 1) | 17:07 |
| clarkb | considering that I'll push a revert, then propose a change that fixes 2) but leave it WIP until we understand 1) | 17:08 |
| Shrews | localhost is going to be the executor there i guess? | 17:09 |
| Shrews | i need to get a mental map of that entire process in my head first, i think | 17:09 |
| clarkb | yup. Thinking about it more we load these roles from a trusted context so we may be able to edit /etc/hosts | 17:09 |
| openstackgerrit | Clark Boylan proposed openstack-infra/zuul-jobs master: Revert "Use /etc/hosts with docker registries to handle ipv6" https://review.openstack.org/651332 | 17:10 |
| clarkb | I'll work on a followup that deals with 2) | 17:10 |
| openstackgerrit | Clark Boylan proposed openstack-infra/zuul-jobs master: Revert "Revert "Use /etc/hosts with docker registries to handle ipv6"" https://review.openstack.org/651334 | 17:13 |
| clarkb | if we decide that 651334 will work because the roles run from a trusted context and can edit /etc/hosts I can remove the revert in the middle. But I don't know enough about zuul to be sure about that | 17:14 |
| * Shrews is a bit uneasy about mucking around with the executor's host file, especially since simultaneous jobs could attempt to edit it at the same time | 17:19 | |
| clarkb | Shrews: it'll be the hosts file in the container. | 17:20 |
| clarkb | though if that isn't mounted rw it will fail | 17:20 |
| Shrews | oh, that's different indeed | 17:20 |
| * clarkb digs up the zuul bwrap command | 17:20 | |
| clarkb | hrm we don't really have that written down in a singple place | 17:23 |
| clarkb | Shrews: '--ro-bind', '/etc/hosts', '/etc/hosts', | 17:26 |
| clarkb | we won't be able to modify that file | 17:26 |
| clarkb | so my change doesn't work as is yet. I'll WIP it with a comment about that ro-bind | 17:27 |
| clarkb | thinking about options here, its possible that we don't need to do the aliasing on the localhost tasks because the localhost tasks run skopeo | 17:37 |
| clarkb | and if skopeo accepts the ipv6 address literal those bits may just work as is. It is annoying that we'd have to do different setups in different places though | 17:38 |
| clarkb | Other options include giving each bwrap container its own writeable /etc/hosts file | 17:38 |
| clarkb | Fixing docker (whcih I've filed a bug for) | 17:38 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool master: Fix race with test_provider_removal unit test https://review.openstack.org/651339 | 17:41 |
| openstackgerrit | Merged openstack-infra/zuul-jobs master: Revert "Use /etc/hosts with docker registries to handle ipv6" https://review.openstack.org/651332 | 17:42 |
| clarkb | where the skopeo stuff gets weird is we appear to use a docker config with the skopeo command | 17:42 |
| clarkb | so to be extra careful we shouldn't write out a docker config that docker itself can't use | 17:42 |
| *** jpena|away is now known as jpena | 17:43 | |
| *** jpena is now known as jpena|off | 17:53 | |
| *** rlandy|ruck is now known as rlandy|ruck|hw | 18:19 | |
| openstackgerrit | Clark Boylan proposed openstack-infra/zuul-jobs master: Only alias registries on build nodes https://review.openstack.org/651353 | 18:51 |
| clarkb | Shrews: mordred ^ thats a cleanup that only runs the /etc/host aliasing where we run docker commands (basically on the test ndoes) and not on localhost where we skopeo | 18:51 |
| clarkb | if skopeo handles ipv6 properly this may work | 18:51 |
| * mordred worships skopeo | 18:52 | |
| clarkb | that makes the roles slightly less general purpose | 18:53 |
| clarkb | and assumes skopeo doesn't also have broken cli parsing | 18:54 |
| clarkb | also we may want to squash that into the revert revert to ensure we don't merge what we know is broken but I think the diff reads a bit better this way so you can see what I've removed | 18:54 |
| *** jamesmcarthur has quit IRC | 18:55 | |
| *** hashar has joined #zuul | 19:15 | |
| *** jamesmcarthur has joined #zuul | 19:21 | |
| *** pwhalen has quit IRC | 19:21 | |
| fungi | anybody had any interaction with the easystack folks scheduled to give this presentation in denver? https://www.openstack.org/summit/denver-2019/summit-schedule/events/23249/best-practice-to-run-zuul-on-kubernetes | 20:13 |
| fungi | seems like they may have been working on a solution in a vacuum, but it's hard to tell just from an abstract | 20:13 |
| Shrews | clarkb: changes seem sensible, though maybe we should squash after it's been reviewed | 20:14 |
| clarkb | Shrews: ya I think it should be squashed | 20:14 |
| clarkb | ok really getting lunch now | 20:14 |
| Shrews | fungi: i seem to recall a similar presentation by them (I at least remember them dev'ing in a vacuum part) at the last summit. I thought corvus was going to speak with them at that time, but maybe that didn't happen... or I'm misremebering | 20:16 |
| *** jamesmcarthur has quit IRC | 20:17 | |
| mordred | Shrews, fungi: yeah - corvus and I chatted with them a bit at the end of their talk | 20:17 |
| mordred | in vancouver | 20:17 |
| *** jamesmcarthur has joined #zuul | 20:18 | |
| fungi | neat. looks like track chairs enjoyed it so much they've approved a rerun | 20:18 |
| Shrews | mordred: and did that chat go something like "hey, we seem to be working on the same things. let's cooperate kplzthx?" :) | 20:19 |
| mordred | Shrews: dude. that was a year ago | 20:21 |
| mordred | Shrews: you have so much more faith in my powers of memory | 20:22 |
| Shrews | mordred: apologies. i'll do my best to exhibit less faith in you going forward :) | 20:25 |
| *** rlandy|ruck|hw is now known as rlandy|ruck | 20:27 | |
| *** pcaruana has quit IRC | 20:31 | |
| *** pcaruana has joined #zuul | 20:33 | |
| *** pcaruana has quit IRC | 20:36 | |
| *** pcaruana has joined #zuul | 20:39 | |
| *** jamesmcarthur has quit IRC | 20:44 | |
| *** jamesmcarthur has joined #zuul | 20:45 | |
| *** pcaruana has quit IRC | 20:47 | |
| *** SpamapS has quit IRC | 21:10 | |
| *** SpamapS has joined #zuul | 21:11 | |
| *** jamesmcarthur has quit IRC | 21:11 | |
| SpamapS | I kind of have a crazy idea for a zuul feature to help with github. People like it when zuul comments on PR's because it provides them with an email notification. But the comments are super noisy. I wonder if we could make Zuul hide its own comments (there's a way to do it in the GH UI, I assume it has an API) | 21:17 |
| fungi | not that crazy i suppose. we do something similar on our gerrit to hide comments from our zuul (and other ci systems) | 21:20 |
| fungi | though not implemented from the zuul end | 21:20 |
| clarkb | I think as long as you can toggle them open like our gerrit shim does | 21:20 |
| clarkb | would be a great feature | 21:20 |
| clarkb | Shrews: mordred should I got ahead and squash the two changes together or are you still needing to review them in the slightly nicer diff setup? | 21:21 |
| fungi | yeah, i like being able to view those comments under certain circumstances, but hiding them by default helps to pick out discussion with actual humans | 21:21 |
| SpamapS | Basically I think we could just add an action to take on success/fail/start that is "hide my comments" or "show my comments" | 21:22 |
| SpamapS | Because ideally what would happen is when checks pass, it hides all the old fails. | 21:23 |
| clarkb | I know what I can do, push a third change that is the squash of the other two then WIP those other two for reviewing | 21:44 |
| * clarkb does this | 21:44 | |
| openstackgerrit | Clark Boylan proposed openstack-infra/zuul-jobs master: Only alias registries on build nodes https://review.openstack.org/651393 | 21:45 |
| clarkb | Shrews: mordred ^ there is the squash. I'll link the other two changes there and wip them | 21:45 |
| *** jamesmcarthur has joined #zuul | 22:03 | |
| *** hashar has quit IRC | 22:08 | |
| *** jamesmcarthur has quit IRC | 22:42 | |
| *** jamesmcarthur has joined #zuul | 22:52 | |
| *** jamesmcarthur has quit IRC | 23:03 | |
| *** jamesmcarthur has joined #zuul | 23:06 | |
| *** rlandy|ruck is now known as rlandy | 23:06 | |
| *** rlandy has quit IRC | 23:09 | |
| *** jamesmcarthur has quit IRC | 23:14 | |
| *** jamesmcarthur has joined #zuul | 23:30 | |
| *** jamesmcarthur has quit IRC | 23:30 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!