| openstackgerrit | Merged zuul/zuul master: Don't capitalize Token in docs https://review.opendev.org/675413 | 00:03 |
|---|---|---|
| *** fdegir has quit IRC | 00:23 | |
| *** fdegir has joined #zuul | 00:24 | |
| *** spsurya has joined #zuul | 02:12 | |
| *** rlandy|rover|bbl is now known as rlandy|rover | 02:15 | |
| *** igordc has quit IRC | 02:17 | |
| *** rlandy|rover has quit IRC | 02:18 | |
| *** igordc has joined #zuul | 02:39 | |
| *** igordc has quit IRC | 02:55 | |
| *** bhavikdbavishi has joined #zuul | 03:05 | |
| *** bhavikdbavishi1 has joined #zuul | 03:12 | |
| *** bhavikdbavishi has quit IRC | 03:14 | |
| *** bhavikdbavishi1 is now known as bhavikdbavishi | 03:14 | |
| *** rfolco has quit IRC | 03:27 | |
| *** raukadah is now known as chkumar|ruck | 04:03 | |
| *** bjackman has joined #zuul | 05:24 | |
| *** igordc has joined #zuul | 05:28 | |
| *** pcaruana has joined #zuul | 05:40 | |
| *** pcaruana has quit IRC | 05:49 | |
| *** igordc has quit IRC | 06:02 | |
| openstackgerrit | Benedikt Löffler proposed zuul/zuul master: Report retried builds in a build set via mqtt. https://review.opendev.org/632727 | 07:19 |
| openstackgerrit | Benedikt Löffler proposed zuul/zuul master: Report retried builds via sql reporter. https://review.opendev.org/633501 | 07:19 |
| *** saneax has joined #zuul | 07:27 | |
| *** jangutter has joined #zuul | 07:34 | |
| *** tosky has joined #zuul | 07:37 | |
| *** jpena|off is now known as jpena | 07:39 | |
| *** pcaruana has joined #zuul | 07:45 | |
| *** themroc has joined #zuul | 07:53 | |
| *** yolanda has quit IRC | 08:07 | |
| openstackgerrit | Benedikt Löffler proposed zuul/zuul master: Report retried builds in a build set via mqtt. https://review.opendev.org/632727 | 08:45 |
| openstackgerrit | Benedikt Löffler proposed zuul/zuul master: Report retried builds via sql reporter. https://review.opendev.org/633501 | 08:45 |
| *** bhavikdbavishi has quit IRC | 10:04 | |
| *** sshnaidm has quit IRC | 10:04 | |
| *** sshnaidm has joined #zuul | 10:08 | |
| *** hwangbo has quit IRC | 10:27 | |
| *** jpena is now known as jpena|lunch | 11:25 | |
| *** jpena|lunch is now known as jpena | 12:22 | |
| *** rfolco has joined #zuul | 12:26 | |
| *** electrofelix has joined #zuul | 12:32 | |
| *** themroc has quit IRC | 12:37 | |
| *** rlandy has joined #zuul | 12:37 | |
| *** rlandy is now known as rlandy|rover | 12:38 | |
| *** themroc has joined #zuul | 12:41 | |
| *** jangutter_ has joined #zuul | 12:49 | |
| *** jangutter has quit IRC | 12:52 | |
| *** bjackman has quit IRC | 13:01 | |
| *** bjackman has joined #zuul | 13:03 | |
| *** jeliu_ has joined #zuul | 13:17 | |
| *** bhavikdbavishi has joined #zuul | 13:19 | |
| *** pcaruana has quit IRC | 13:23 | |
| *** bhavikdbavishi has quit IRC | 13:43 | |
| *** pcaruana has joined #zuul | 14:00 | |
| *** bjackman has quit IRC | 14:26 | |
| openstackgerrit | Jeff Liu proposed zuul/zuul-jobs master: Add skip-registry-check parameter to deploy openshift role https://review.opendev.org/675688 | 14:28 |
| *** themr0c has joined #zuul | 14:28 | |
| *** themroc has quit IRC | 14:31 | |
| *** bjackman has joined #zuul | 14:32 | |
| *** chkumar|ruck is now known as raukadah | 14:57 | |
| *** bjackman has quit IRC | 14:57 | |
| *** olaph has joined #zuul | 14:58 | |
| *** bjackman has joined #zuul | 15:00 | |
| *** mattw4 has joined #zuul | 15:12 | |
| *** bhavikdbavishi has joined #zuul | 15:18 | |
| *** themr0c has quit IRC | 15:19 | |
| openstackgerrit | Jeff Liu proposed zuul/zuul-jobs master: Add skip-registry-check parameter to deploy openshift role https://review.opendev.org/675688 | 15:21 |
| openstackgerrit | Jeff Liu proposed zuul/zuul-operator master: WIP: Add zuul-operator-functional-openshift job https://review.opendev.org/674355 | 15:22 |
| *** mattw4 has quit IRC | 15:22 | |
| *** AshBullock has joined #zuul | 15:26 | |
| *** armstrongs has joined #zuul | 15:31 | |
| AshBullock | Hey all, does anyone have any experience with the variable job.cleanup-run? https://zuul-ci.org/docs/zuul/user/config.html#attr-job.cleanup-run Am I right in thinking the playbooks under this variable should execute even when a job is interrupted by a new trigger event? | 15:31 |
| corvus | AshBullock: that's correct | 15:33 |
| corvus | AshBullock: the only time they may not run is if the executor is stopped or crashes | 15:33 |
| corvus | not much we can do about that :) | 15:33 |
| tristanC | corvus: iiuc the scheduler scale out, then it should enable enforcing the cleanup run happen even when the executor service crashes right? | 15:36 |
| corvus | tristanC: we might be able to do that. however, if the cleanup playbook depends on stuff that happens in the pre- or run- playbooks, then it won't work. so maybe we shouldn't try, and instead should just try to have nice graceful shutdowns :) | 15:40 |
| *** bhavikdbavishi has quit IRC | 15:43 | |
| *** bhavikdbavishi1 has joined #zuul | 15:43 | |
| *** bhavikdbavishi1 is now known as bhavikdbavishi | 15:46 | |
| AshBullock | Thanks, I can see the playbook being executed now, was just not looking at the log stream for it's output | 15:53 |
| openstackgerrit | Jeff Liu proposed zuul/zuul-operator master: WIP: Add zuul-operator-functional-openshift job https://review.opendev.org/674355 | 15:53 |
| *** jpena is now known as jpena|off | 15:58 | |
| *** bjackman has quit IRC | 15:58 | |
| *** spsurya has quit IRC | 16:04 | |
| *** bjackman has joined #zuul | 16:08 | |
| corvus | mordred: i was just looking at this ansible parse error: https://zuul.opendev.org/t/zuul/build/120fbbef25cd4ef9bd62f1039f8167e8/log/job-output.txt#267 | 16:14 |
| corvus | it shows up in the text log, but i don't think it shows up in the json log... so all our neat automatic find-the-error stuff isn't working... do you think it's possible for that to end up in json? | 16:15 |
| corvus | mordred: i guess the callback isn't getting run, so if we did it, we'd maybe have the executor write a "fake" playbook entry to the json log before it runs the next one? | 16:16 |
| zbr|flu | tristanC: thanks for the link to ansible-tags change. by any chance this would also apply to "roles:" in job configuration? | 16:17 |
| mordred | corvus: oh - is that because when we get a PARSE error the executor just appends the error text it got from the command line to the text log | 16:22 |
| *** bjackman has quit IRC | 16:22 | |
| mordred | corvus: I think I thought about putting that into the json log before - but as it is structured right now that would involve the executor needing to read the existing json log in to memory | 16:23 |
| mordred | corvus: maybe what we should do is have the executor append the info to the text log, and also put it into a file that the json callback can look for and then add the fake playbook structure like you were just suggesting | 16:24 |
| corvus | mordred: why is it a problem for the executor to load it into memory? | 16:25 |
| corvus | just wanting to keep the memory footprint low, because it can be big? | 16:25 |
| corvus | (and we don't mind that ansible does that because it's a separate process?) | 16:25 |
| mordred | yeah. and we wouldn't reclaim the memory directly by the process going away | 16:25 |
| mordred | yeah | 16:25 |
| corvus | oh oh! | 16:25 |
| corvus | mordred: we can do it without loading it | 16:25 |
| mordred | this is one of the reasons I was thinking we should switch to an append-only yaml format | 16:25 |
| mordred | we can? | 16:26 |
| corvus | mordred: oh wait, no we can't | 16:26 |
| corvus | i mean, we could, but it'd be yucky | 16:26 |
| mordred | yeah | 16:26 |
| corvus | like, we'd have to drop the final ']' | 16:26 |
| corvus | and also change }, to } | 16:27 |
| corvus | er other way around | 16:27 |
| corvus | dangerous to do without parsing | 16:27 |
| mordred | yeah. this is why I started working on this: https://review.opendev.org/#/c/623256/ a while back | 16:27 |
| corvus | mordred: we could fork/exec to do the json append | 16:28 |
| mordred | we could | 16:28 |
| corvus | meh. we can probably live with the status quo and focus our effort on the yaml thing. better long-term payoff | 16:29 |
| corvus | mordred: thanks for the brainstorming :) | 16:29 |
| mordred | corvus: yeah. as a stepping stone, what if we had the callbacks write append-only yaml - but then we have the upload task transform it into json? is that too rube-goldberg? | 16:29 |
| corvus | mordred: i think it is. my bet is that's more effort than fixing whatever tripped up the functional stream test :) | 16:30 |
| mordred | :) | 16:30 |
| corvus | mordred: i rechceked your yaml change; it still applies cleanly | 16:30 |
| corvus | okay, back to swift | 16:31 |
| mordred | cool. I think we could slightly rework it to not need multi-document support on the reading side pretty easily | 16:31 |
| mordred | (turns out yaml lists directly append pretty nicely too) | 16:31 |
| openstackgerrit | Jeff Liu proposed zuul/zuul-operator master: WIP: Add zuul-operator-functional-openshift job https://review.opendev.org/674355 | 16:34 |
| corvus | mordred: i'm assuming there's some simple yaml->object stuff for js | 16:35 |
| corvus | (cause the current json -> object is more or less automatic in axios) | 16:35 |
| mordred | yeah. http://nodeca.github.io/js-yaml/ | 16:35 |
| mordred | would have to figure out how to wire that up | 16:36 |
| *** AshBullock has quit IRC | 16:37 | |
| *** jangutter_ has quit IRC | 16:39 | |
| mordred | corvus: I've got a change to just do it as a list in a single doc - ok if I push that up? | 16:40 |
| corvus | mordred: ++ | 16:41 |
| openstackgerrit | Monty Taylor proposed zuul/zuul master: Add appending yaml log plugin https://review.opendev.org/623256 | 16:41 |
| mordred | I think that'll be a cleaner format to work with | 16:41 |
| corvus | yeah, i assume this will slot into existing code better | 16:42 |
| *** mattw4 has joined #zuul | 16:42 | |
| mordred | corvus: I'm going to rebase on master so that the javascript stuff is there to look at | 16:43 |
| corvus | mordred: ack | 16:46 |
| openstackgerrit | Monty Taylor proposed zuul/zuul master: Add appending yaml log plugin https://review.opendev.org/623256 | 16:57 |
| openstackgerrit | Monty Taylor proposed zuul/zuul master: Process yaml log files if they exist https://review.opendev.org/676246 | 16:57 |
| mordred | corvus: there's an initial stab at parsing the yaml from axios in the dashboard | 16:58 |
| *** jangutter has joined #zuul | 16:59 | |
| openstackgerrit | Sorin Sbarnea proposed zuul/zuul master: job: add ansible-tags and ansible-skip-tags attribute https://review.opendev.org/575672 | 17:06 |
| *** jangutter has quit IRC | 17:10 | |
| zbr|flu | tristanC: i rebased your tags change ^ -- but it would be useful to get some feedback from corvus mordred regarding other aspects. | 17:19 |
| zbr|flu | should we reuse already implemented 'tags" and only add "skip-tags" to ? i would vote for it, seems less confusing than ansible-tags and ansible-skip-tags. | 17:20 |
| openstackgerrit | Monty Taylor proposed zuul/zuul master: Write errors from ansible execution into yaml log https://review.opendev.org/676250 | 17:28 |
| corvus | zbr|flu, tristanC: my guess is that the tags feature is little-enough used that a behavior change would be okay, but we should ask on the mailing list to get more input. | 17:36 |
| corvus | zbr|flu, tristanC: i wrote it originally to help with job classification in openstack's logstash system. to be honest, i'm not even sure if we're using it there. | 17:36 |
| corvus | i'm going to start making releases now | 17:37 |
| zbr|flu | corvus: i guess one use case does not rule out the other, right? | 17:37 |
| corvus | zbr|flu: right. i think the use cases are compatible, so we don't necessarily need another option. | 17:37 |
| zbr|flu | the reason i asked yesterday was because I needed ansible tag in order to call some roles with special testing tags. glad that I found tristanC old change. | 17:39 |
| corvus | Shrews, tobiash: i don't see any release-critical nodepool patches, and i think we should be good to release current master 7bbbe73c0daddce68fd3901d322e777aa37520e0 as 3.8.0. does that sound right? | 17:40 |
| Shrews | corvus: lemme review real quick | 17:40 |
| Shrews | corvus: i think you restarted all of nodepool with 7bb the other day, right? | 17:41 |
| corvus | Shrews: no i haven't, i only did a zuul restart | 17:42 |
| corvus | oh, i did do some restarts due to the swift errors we were getting | 17:43 |
| corvus | that would have been some launchers, but probably not all of them | 17:43 |
| Shrews | i see builders restarted on Aug 5th. Wasn't that for the pwd issue? | 17:43 |
| corvus | (sorry, not really swift errors, more like, errors i caused with cloud accounts during swift testing) | 17:43 |
| corvus | yeah | 17:43 |
| corvus | Shrews: so i guess the answer is yes! :) | 17:43 |
| Shrews | lol | 17:43 |
| Shrews | corvus: then we are gtg | 17:43 |
| corvus | Shrews: i get the two of us confused sometimes | 17:44 |
| tobiash | So I guess a release is fine :) | 17:44 |
| Shrews | corvus: practically twins | 17:44 |
| corvus | on the zuul side, i think master, 618be8a06380c23744224267c81137bca81e3986, should be good as 3.10.0 | 17:44 |
| corvus | pushed nodepool 3.8.0 | 17:45 |
| Shrews | w00t | 17:46 |
| *** saneax has quit IRC | 17:54 | |
| *** armstrongs has quit IRC | 18:02 | |
| mordred | \o/ | 18:03 |
| *** hwangbo has joined #zuul | 18:05 | |
| corvus | looks like nodepool was published okay. pushing zuul now. | 18:06 |
| openstackgerrit | Monty Taylor proposed zuul/zuul master: Add appending yaml log plugin https://review.opendev.org/623256 | 18:13 |
| openstackgerrit | Monty Taylor proposed zuul/zuul master: Process yaml log files if they exist https://review.opendev.org/676246 | 18:14 |
| openstackgerrit | Monty Taylor proposed zuul/zuul master: Write errors from ansible execution into yaml log https://review.opendev.org/676250 | 18:14 |
| corvus | mordred got nerdsniped | 18:14 |
| mordred | corvus: great example there of a patch which merges fine and worked (mostly) ok in its original form, but which fails when merged with current master ... because the directory where those things went changed | 18:15 |
| openstackgerrit | Monty Taylor proposed zuul/zuul master: Add appending yaml log plugin https://review.opendev.org/623256 | 18:17 |
| openstackgerrit | Monty Taylor proposed zuul/zuul master: Process yaml log files if they exist https://review.opendev.org/676246 | 18:17 |
| openstackgerrit | Monty Taylor proposed zuul/zuul master: Write errors from ansible execution into yaml log https://review.opendev.org/676250 | 18:17 |
| corvus | mordred: yeah, and our testing of the js is weak (as tristanC pointed out the other day) | 18:18 |
| mordred | yeah. we'll get there | 18:18 |
| corvus | zuul and nodepool are up on pypi, i will unleash the release announcements | 19:02 |
| corvus | hrm | 19:03 |
| corvus | i do note that we don't have updated documentation | 19:04 |
| tobiash | \o/ | 19:04 |
| corvus | (i mean, all the release notes are there, but the "In Development" header hasn't been updated to the new release number) | 19:04 |
| corvus | it'll get fixed on the next patch landing | 19:04 |
| corvus | but it's supposed to happen with releases too | 19:04 |
| *** electrofelix has quit IRC | 19:06 | |
| corvus | ohhhh | 19:07 |
| corvus | it published the 3.10.0 directory, but didn't publish master -- https://zuul-ci.org/docs/zuul/3.10.0/releasenotes.html | 19:08 |
| corvus | that's technically the most correct thing to do | 19:08 |
| corvus | (since a release might trail master) | 19:08 |
| corvus | just slightly annoying in this case | 19:08 |
| mordred | corvus: yeah. that is the most correct thing. and also weird | 19:09 |
| *** jeliu_ has quit IRC | 19:09 | |
| corvus | i guess in the future i should link to the version-specific release notes | 19:09 |
| corvus | in the emails | 19:09 |
| mordred | corvus: oh - that's not a terrible idea | 19:10 |
| openstackgerrit | Tobias Henkel proposed zuul/nodepool master: Support optional post upload hooks https://review.opendev.org/676266 | 19:17 |
| tristanC | corvus: shouldn't the nodepool source be published here https://tarballs.opendev.org/zuul/nodepool/ or there https://tarballs.openstack.org/nodepool/ ? | 19:31 |
| corvus | tristanC: perhaps? :) | 19:32 |
| corvus | it certainly looks like we were publishing release tarballs and wheels to https://tarballs.openstack.org/nodepool/ | 19:33 |
| tristanC | yes, that's where we used to get zuul and nodepool sources... should we use the gitea archive endpoint instead? | 19:33 |
| corvus | i know we're publishing the branch-tip tarballs and wheels to https://tarballs.opendev.org/zuul/nodepool/ | 19:33 |
| corvus | tristanC: no no never use the gitea archive endpoint | 19:33 |
| corvus | there's a reason we don't link to it anywhere, and we have no idea how people are still finding it | 19:34 |
| corvus | i guess we missed the release tarball publication jobs... | 19:34 |
| corvus | tristanC: you can get the tarballs from https://pypi.org/project/nodepool/#files | 19:34 |
| tristanC | corvus: well, for zuul, that's the only place to get the source only, what got published on tarballs.openstack.org contains compiled js that we have to remove and it makes our src.rpm very heavy | 19:34 |
| corvus | tristanC: well, it's definitely not a zuul release. the packaging is all wrong | 19:35 |
| corvus | tristanC: will https://pypi.org/project/zuul/#files work for zuul? | 19:35 |
| mordred | should we maybe publish two source tarballs - one with compiled javascript and one without? | 19:36 |
| mordred | we publish with compiled js so that people can pip install without needing the javascript stack - but I can also understand someone wanting only a source tarball with no built things in it | 19:37 |
| mordred | but definitely the source exports from gitea should be avoided since they are not source releases | 19:38 |
| *** bhavikdbavishi has quit IRC | 19:38 | |
| tristanC | corvus: that 3.10.0 release from pypi would works, but is it normal it doesn't contains the compiled js? | 19:38 |
| tristanC | ( https://pypi.org/project/zuul/3.9.0/#files , tar.gz is 10.9 MB ) | 19:38 |
| mordred | that seems like an issue | 19:39 |
| tristanC | mordred: that seems like a fix for our src.rpm ;) | 19:39 |
| corvus | wow i really don't like having the pre-releases there | 19:39 |
| corvus | 3.10.0 is 2.5M | 19:39 |
| corvus | i don't know why it would be smaller | 19:40 |
| tristanC | corvus: because it doesn't have the web ui compiled files and fonts | 19:40 |
| mordred | corvus: somethign else is off - the wheel is marked as python2 | 19:40 |
| corvus | tristanC: did we make a change to stop doing that? | 19:40 |
| tristanC | corvus: not that i'm aware of | 19:41 |
| mordred | corvus: I'm looking through the jobs now - it's possible when we shifted tenants we missed the bits that add the javascript tools to the python build jobs | 19:41 |
| tristanC | corvus: mordred: assuming this release is going to be fixed to bundle the compiled js, would it possible to get a source only* release somewhere? | 19:42 |
| corvus | sounds like maybe we want to see if we can make a third kind of release artifact (source-only, source+jscontent, jscontent) | 19:42 |
| corvus | and also fix the wheel markings | 19:42 |
| mordred | yeah | 19:43 |
| tristanC | that would be ideal please | 19:43 |
| mordred | I think the wheel markings would be fixed by setting release_python: to python3 for the job | 19:43 |
| mordred | then maybe we make a child job that also installs yarn | 19:43 |
| mordred | and produces the additional tarball | 19:44 |
| corvus | mordred: would it make any sense to stick with 2 deliverables and put the js content in the wheel, but not the tarball? | 19:45 |
| corvus | mordred: or is that just crazypants? | 19:45 |
| mordred | hrm | 19:46 |
| mordred | corvus: actually - yeah - I kind of like that | 19:47 |
| mordred | let the source tarball be a source tarball | 19:47 |
| mordred | pip is going to prefer wheels anyway | 19:47 |
| corvus | yeah, i got to thinking "wheels are binary artifacts, right?" but i wasn't sure if there's like a "the wheel should be the same as the tarball" rule or something | 19:48 |
| corvus | i'm finishing up my 'how to use new log stuff' email, should be able to send it out in a little bit | 19:53 |
| *** jeliu_ has joined #zuul | 20:02 | |
| *** jeliu_ has quit IRC | 20:05 | |
| *** jeliu_ has joined #zuul | 20:05 | |
| corvus | sent! http://lists.zuul-ci.org/pipermail/zuul-discuss/2019-August/000986.html | 20:10 |
| clarkb | to clarify you can get the source only from the source repository right? | 20:12 |
| mordred | clarkb: aroo? | 20:15 |
| clarkb | tristanC: said "for zuul, that's the only place to get the source only" referring to the gitea archive | 20:15 |
| clarkb | but git works too right? | 20:15 |
| corvus | clarkb: i believe tristanC meant "a release artifact containly only uncompiled source code" | 20:15 |
| corvus | since the rpm building process usually starts with "download a tarball and then compile it...." they would prefer not to carry around wasteful extra compiled stuff | 20:16 |
| clarkb | and rpm building can't start from a git repo like deb building? maybe that is what I am missing | 20:17 |
| mordred | clarkb: deb building also wants a tarball | 20:17 |
| mordred | clarkb: git-buildpackage puts source tarball contents into the git repo in the orig branch | 20:18 |
| clarkb | hrm I thought all of the openstack deb packaging started with full forks of the openstack code | 20:19 |
| tristanC | clarkb: the package maintainer may convert that to a proper source release artifact for the build system (so that distrib can rebuild later, even when vcs becomes unavailable) | 20:25 |
| clarkb | got it | 20:27 |
| tristanC | at least for rpm, we use a source url, and it's better if the url points at something produced by the upstream project | 20:28 |
| mordred | ++ | 20:35 |
| *** sshnaidm is now known as sshnaidm|afk | 20:46 | |
| *** jeliu_ has quit IRC | 21:00 | |
| *** armstrongs has joined #zuul | 21:33 | |
| *** armstrongs has quit IRC | 21:43 | |
| *** igordc has joined #zuul | 21:52 | |
| *** mattw4 has quit IRC | 21:53 | |
| *** mattw4 has joined #zuul | 21:56 | |
| *** tosky has quit IRC | 22:27 | |
| *** igordc has quit IRC | 22:28 | |
| *** igordc has joined #zuul | 22:28 | |
| *** igordc has quit IRC | 22:33 | |
| *** saneax has joined #zuul | 22:54 | |
| *** rlandy|rover has quit IRC | 23:46 | |
| pabelanger | no issue upgrading nodepool to 3.8.0! | 23:54 |
| pabelanger | aiming for tomorrow to upgrade to zuul 3.10.0 | 23:55 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!