| *** mattw4 has quit IRC | 00:09 | |
| *** jamesmcarthur has quit IRC | 00:09 | |
| *** Defolos has quit IRC | 00:13 | |
| *** tosky has quit IRC | 00:18 | |
| *** igordc has quit IRC | 00:23 | |
| *** rlandy has quit IRC | 00:49 | |
| *** jcapitao_off has quit IRC | 00:50 | |
| *** armstrongs has joined #zuul | 01:18 | |
| *** sgw has quit IRC | 01:26 | |
| *** armstrongs has quit IRC | 01:28 | |
| *** saneax has quit IRC | 02:24 | |
| *** swest has quit IRC | 02:27 | |
| *** swest has joined #zuul | 02:41 | |
| *** jamesmcarthur has joined #zuul | 02:53 | |
| *** jamesmcarthur has quit IRC | 03:22 | |
| *** swest has quit IRC | 05:26 | |
| *** evrardjp has quit IRC | 05:35 | |
| *** evrardjp has joined #zuul | 05:35 | |
| *** saneax has joined #zuul | 05:37 | |
| *** bolg has joined #zuul | 05:53 | |
| *** swest has joined #zuul | 06:10 | |
| *** felixedel has joined #zuul | 06:16 | |
| *** bolg has quit IRC | 06:27 | |
| *** sgw has joined #zuul | 06:30 | |
| *** felixedel has joined #zuul | 06:30 | |
| *** smyers has quit IRC | 06:45 | |
| *** smyers has joined #zuul | 06:45 | |
| *** Defolos has joined #zuul | 06:56 | |
| *** bolg has joined #zuul | 07:02 | |
| *** dpawlik has joined #zuul | 07:09 | |
| *** harrymichal has joined #zuul | 07:21 | |
| *** bolg has quit IRC | 08:06 | |
| *** bolg has joined #zuul | 08:11 | |
| *** jcapitao_off has joined #zuul | 08:11 | |
| *** jcapitao_off is now known as jcapitao | 08:19 | |
| *** tosky has joined #zuul | 08:21 | |
| *** harrymichal has quit IRC | 08:23 | |
| openstackgerrit | Felix Edel proposed zuul/zuul master: WIP Visualize retry builds as expandable rows on buildset page https://review.opendev.org/710408 | 08:26 |
|---|---|---|
| openstackgerrit | Felix Edel proposed zuul/zuul master: WIP Visualize retry builds as expandable rows on buildset page https://review.opendev.org/710408 | 08:28 |
| *** jpena|off is now known as jpena | 08:38 | |
| openstackgerrit | Tobias Henkel proposed zuul/zuul master: Optimize canMerge using graphql https://review.opendev.org/709836 | 08:44 |
| *** saneax has quit IRC | 08:45 | |
| *** saneax has joined #zuul | 08:59 | |
| *** harrymichal has joined #zuul | 09:11 | |
| *** saneax has quit IRC | 09:11 | |
| *** saneax has joined #zuul | 09:13 | |
| *** saneax has quit IRC | 09:13 | |
| *** saneax has joined #zuul | 09:14 | |
| *** saneax has quit IRC | 09:16 | |
| *** saneax has joined #zuul | 09:23 | |
| *** avass has joined #zuul | 09:26 | |
| *** harrymichal_ has joined #zuul | 09:30 | |
| *** bertu has joined #zuul | 09:31 | |
| *** harrymichal has quit IRC | 09:33 | |
| *** harrymichal_ is now known as harrymichal | 09:33 | |
| bertu | Hi! Zuul-CI newbie here. | 09:33 |
| bertu | In a few places in the Zuul-CI documentation like https://zuul-ci.org/docs/zuul/reference/jobs.html | 09:34 |
| bertu | I read | 09:34 |
| bertu | "The change’s project will have the change’s branch checked out, as will all of the other projects, if that branch exists (otherwise, a fallback or default branch will be used)" | 09:35 |
| bertu | and the like | 09:35 |
| bertu | Does it mean that if I have branch "foo" on repo A and repo B as required projects | 09:35 |
| bertu | when I trigger a build on a change pushed to A@foo also B@foo will be checked out? | 09:36 |
| bertu | *as long as there is a "foo" branch in B* | 09:36 |
| mnaser | bertu: correct | 09:40 |
| openstackgerrit | Jan Kubovy proposed zuul/zuul master: Scheduler test app manager https://review.opendev.org/708812 | 09:45 |
| *** harrymichal_ has joined #zuul | 09:46 | |
| *** harrymichal has quit IRC | 09:47 | |
| *** harrymichal_ is now known as harrymichal | 09:47 | |
| harrymichal | Hi! I'm using Zuul in SoftwareFactory. I got a question. Would it be possible to add the commit hash + link to it to the summary page of a job?? | 10:19 |
| *** jcapitao has quit IRC | 10:19 | |
| *** jcapitao has joined #zuul | 10:19 | |
| harrymichal | It is quite inconvenient when a 'periodic' pipeline is ran on the 'master' branch. There is no way of telling if there was a change introduced between pipelines (at least I don't see one). | 10:20 |
| *** jcapitao has quit IRC | 10:20 | |
| *** jcapitao has joined #zuul | 10:20 | |
| bertu | mnaser: that never happens on a Zuul 3.9.0 version that my company uses -> I supposed that means the configuration is customized not to do so... | 10:34 |
| mnaser | bertu: interesting | 10:34 |
| mnaser | bertu: if you are pushing a patch to A@foo, and B is inside required-projects, then B@foo should be checked out in CI | 10:35 |
| mnaser | is that the scenario? | 10:35 |
| mnaser | bertu: it is possible that you have override-branch setup? https://zuul-ci.org/docs/zuul/reference/job_def.html?highlight=override%20checkout#attr-job.override-checkout | 10:36 |
| bertu | mnaser: Yes. But it does not work. I have to rely on "Depends-On" for that purpose (i.e. testing changes on different repos at the same time) | 10:36 |
| mnaser | or maybe it's overridden here https://zuul-ci.org/docs/zuul/reference/job_def.html?highlight=override%20checkout#attr-job.required-projects.override-checkout ? | 10:37 |
| bertu | mnaser: interesting! I will try to look into the configuration to see what is going on | 10:37 |
| mnaser | bertu: are you trying to test changes that have not landed yet in B@foo ? | 10:37 |
| bertu | mnaser: what I meant was that we have PR#1 (A@foo) and PR#2 (B@foo) - we use github, but that should be irrelevant here - and I want to test both before they are merged on the A and B repos | 10:39 |
| bertu | mnaser: on the master | 10:39 |
| mnaser | bertu: ok so in this scenario there is 3 projects, right? | 10:40 |
| mnaser | A@foo, B@foo and <some other project that depends on A & B> | 10:40 |
| bertu | mnaser: sorry I think I did not explain correctl earlier. Project/repo A, master branch, foo branch + PR#1; project/repo B, master branch, foo branch (with changes already landed) | 10:44 |
| mnaser | bertu: can the changes on any of those repos land in a specific order or they are both depending on each other? | 10:50 |
| mnaser | bertu: can you land the A change before B or vice versa or they really both need to go in together? | 10:50 |
| mnaser | If they can be ordered, then what you should do is run the B jobs inside A all the time, and run the A jobs inside B all the time — as jobs are not tied to a specific repository. That will give you coverage to make sure projects aren’t breaking each other | 10:51 |
| mnaser | bertu: if they absolutely need to land at the same time to be functional, this is something that’s in the works on Zuul — https://zuul-ci.org/docs/zuul/reference/developer/specs/circular-dependencies.html | 10:53 |
| mnaser | which is making progress but not available yet: https://review.opendev.org/#/c/685354/ | 10:54 |
| bertu | mnaser: does it mean in the last scenario I described foo is not automatically checked out? At this point is more that I cannot match my understanding of the docs to the behaviour I see. | 10:57 |
| mnaser | If you make a change to A@foo, the branch B@foo should be checked out if it’s inside required-projects | 10:57 |
| bertu | mnaser: thanks :) I will check what is going on | 10:58 |
| bertu | mnaser: about the use case and your other good suggestions... | 10:58 |
| mnaser | bertu: cool good luck, please let me know if you have other questions | 11:01 |
| bertu | mnaser: for reasons I do not fully understand yet, people put CI code (Zuul configs, playbook, etc.) on a separate repo. Let's say CI stuff in repo A and code in repo B. There were cases where I wanted to verify a change on the code together with a change on a CI job, before any of those changes were merged on master. I managed to do that using "Depends-On" but I expected to simply happen because I used the | 11:03 |
| bertu | same branch name for both changes -> but that did not happen | 11:03 |
| mnaser | bertu: I’m that scenario, you could actually just run the repo A jobs inside repo B by appending them to the zuul yaml file | 11:05 |
| mnaser | In** sorry mobile | 11:05 |
| bertu | mnaser: thanks! | 11:08 |
| mnaser | np :) | 11:09 |
| openstackgerrit | Felix Edel proposed zuul/zuul master: Dequeue changes via github checks API https://review.opendev.org/709135 | 11:17 |
| openstackgerrit | Jan Kubovy proposed zuul/zuul master: Make test setup_config more pure https://review.opendev.org/709084 | 11:37 |
| openstackgerrit | Jan Kubovy proposed zuul/zuul master: Scheduler test app manager https://review.opendev.org/708812 | 11:37 |
| openstackgerrit | Felix Edel proposed zuul/zuul master: WIP Dequeue changes via github checks API https://review.opendev.org/709135 | 11:54 |
| openstackgerrit | Jan Kubovy proposed zuul/zuul master: Use scheduler manager consistently in tests https://review.opendev.org/709542 | 11:59 |
| *** jcapitao is now known as jcapitao_lunch | 12:03 | |
| *** saneax has quit IRC | 12:16 | |
| openstackgerrit | Jan Kubovy proposed zuul/zuul master: Scheduler test app manager https://review.opendev.org/708812 | 12:16 |
| *** saneax has joined #zuul | 12:17 | |
| *** jfoufas1 has joined #zuul | 12:29 | |
| *** felixedel has quit IRC | 12:30 | |
| *** jpena is now known as jpena|lunch | 12:33 | |
| *** armstrongs has joined #zuul | 12:33 | |
| jfoufas1 | This zuul web gui looks nice, https://ci.gerritcodereview.com/t/gerrit/build/28b385d0e989449cb2ac1fcb2cb8e1ca, any info? | 12:35 |
| tristanC | harrymichal: indeed it seems like zuul doesn't report the exact ref used in periodic pipeline. perhaps that could be added, but that won't necessarly tell you all the commit that was introduced between two run | 12:41 |
| tristanC | corvus: is a zuul release planned soon? (for the security fix) | 12:42 |
| *** armstrongs has quit IRC | 12:44 | |
| openstackgerrit | Jan Kubovy proposed zuul/zuul master: Scheduler test app manager https://review.opendev.org/708812 | 12:46 |
| fungi | jfoufas1: what info are you looking for? | 12:51 |
| fungi | if you're looking for details of what was run and what output was emitted, then the console view may be more useful than the summary view | 12:54 |
| jfoufas1 | fungi if the additions to the web gui, logs and console will be added | 12:55 |
| fungi | or the logs view, though that particular build only seems to collect stdout/stderr from the commands plus some job metadata | 12:55 |
| fungi | jfoufas1: what additions? | 12:55 |
| fungi | if the job collects more log files they'll appear in the logs view, yes | 12:55 |
| fungi | and whatever ansible playbooks/tasks are added will appear in the console view | 12:56 |
| fungi | i feel like i might be missing your question | 12:56 |
| jfoufas1 | fungi thank you, that answers it. | 12:57 |
| fungi | okay, cool. i expect all of us in here are happy to answer any questions you have | 12:58 |
| *** rlandy has joined #zuul | 12:59 | |
| fungi | jfoufas1: looking at a failed build might provide a better example too: https://ci.gerritcodereview.com/t/gerrit/build/06b7799459f049b79e3d15d2e5740278 | 13:00 |
| fungi | more effort is taken to expose the likely failure point from the output stream with some context, since people tend to look at failure results a lot more often than successful build details | 13:02 |
| *** jfoufas1 has quit IRC | 13:02 | |
| fungi | oh, i guess they's already disconnected | 13:02 |
| fungi | er, they'd | 13:02 |
| Shrews | fungi: you live in a southern state. "they's" is a perfectly cromulent word | 13:04 |
| fungi | thanks, your knowledge of southern dialects embiggens us both | 13:05 |
| harrymichal | tristanC, that is true but it can tell me if the problem (if suddenly tests start failing) was introduced by the project itself or some of its dependencies | 13:07 |
| harrymichal | for my use case that is quite handy to know | 13:07 |
| *** jamesmcarthur has joined #zuul | 13:09 | |
| *** jcapitao_lunch is now known as jcapitao | 13:14 | |
| *** jamesmcarthur has quit IRC | 13:18 | |
| *** jamesmcarthur has joined #zuul | 13:21 | |
| *** jamesmcarthur has quit IRC | 13:36 | |
| *** jamesmcarthur has joined #zuul | 13:37 | |
| *** jpena|lunch is now known as jpena | 13:38 | |
| *** toabctl has quit IRC | 13:38 | |
| openstackgerrit | Merged zuul/zuul master: Support file comments via Github checks API https://review.opendev.org/708611 | 13:41 |
| *** jamesmcarthur has quit IRC | 13:53 | |
| *** jamesmcarthur has joined #zuul | 13:55 | |
| openstackgerrit | Jan Kubovy proposed zuul/zuul master: Scheduler test app manager https://review.opendev.org/708812 | 13:55 |
| *** Goneri has joined #zuul | 14:00 | |
| *** saneax has quit IRC | 14:16 | |
| *** jfoufas1 has joined #zuul | 14:30 | |
| jfoufas1 | fungi i got thrown out of IRC...network porblems. We are still quite new to zuul, and the real devops like avass still do not understand what needs to be published in order to appear in console view. | 14:39 |
| *** jamesmcarthur has quit IRC | 14:41 | |
| corvus | jfoufas1: for log files to show up in the logs tab, basically, any file that's in the logs directory of the executor workdir at the end of the job will end up there; the easiest way to do that is to put log files in ~/logs on the worker nodes, then use this role, https://zuul-ci.org/docs/zuul-jobs/log-roles.html#role-fetch-output in the post playbook of your base job to pull them back at the end of the job. | 14:43 |
| corvus | jfoufas1: the console tab is built automatically from the ansible tasks that run in the job | 14:43 |
| mordred | ~/zuul-output/logs | 14:44 |
| corvus | oh sorry, yes. that's the directory to use on the worker node. | 14:44 |
| *** avass has quit IRC | 14:46 | |
| jfoufas1 | corvus will have to look into it, for some reason we do not get the tabs, even though logs are in ~/zuul-output/logs | 14:46 |
| corvus | jfoufas1: ah -- if the tabs are missing, there might be something else missing, 1 sec | 14:47 |
| *** jamesmcarthur has joined #zuul | 14:47 | |
| jfoufas1 | avass: says we use the role too.. | 14:47 |
| corvus | jfoufas1: for the tabs to work, you'll need the https://zuul-ci.org/docs/zuul-jobs/log-roles.html#role-generate-zuul-manifest role in the base job post playbook too, right before the upload logs role. | 14:48 |
| jfoufas1 | corvus: that's it!!! | 14:48 |
| corvus | jfoufas1: and whatever log storage system you're using will need to support CORS headers. the swift and gcp log roles should automatically do that, but if you're running a static server, you'll need to configure the http server to set them | 14:49 |
| *** jfoufas1 has quit IRC | 14:59 | |
| corvus | tristanC, fungi: does this look good for a release to get the security fix: commit 0684df0dd191427d000f0cee2e18ccdc07f5f3c8 (tag: 3.17.0, refs/changes/87/710287/2) | 15:00 |
| corvus | that's one back from master; we could probably release master (it's one more github change which i assume tobiash has tested in prod) | 15:00 |
| corvus | but since it's a security fix, i was tending toward being conservative | 15:00 |
| mordred | corvus: d191427 is what we're running in opendev? | 15:01 |
| mordred | gah - 0684de0 I mean | 15:02 |
| corvus | mordred: no, i don't think we're running it | 15:02 |
| tristanC | corvus: that sounds good, fwiw we are running with a cherry-pick of the fix on 3.13.0 and 3.10.2 | 15:04 |
| tobiash | corvus: which github change? | 15:04 |
| openstackgerrit | Jan Kubovy proposed zuul/zuul master: Use scheduler manager consistently in tests https://review.opendev.org/709542 | 15:04 |
| openstackgerrit | Jan Kubovy proposed zuul/zuul master: Refactor executor_client in tests https://review.opendev.org/709672 | 15:04 |
| corvus | tobiash: file comments | 15:04 |
| tobiash | that's one we don't have yet in prod | 15:04 |
| tobiash | but the other checks api changes are in and working :) | 15:04 |
| corvus | oh good, then let's stick with what i proposed :) | 15:05 |
| corvus | mordred: i think with the feedback from tristanC i'm pretty comfortable releasing that commit before opendev restarts | 15:05 |
| mordred | corvus: ++ | 15:05 |
| corvus | okay, i'll make it so | 15:05 |
| fungi | yep, sgtm | 15:07 |
| fungi | sorry, spending my morning down an openstack vulnerability management spring cleaning hole | 15:07 |
| corvus | it's that season | 15:08 |
| corvus | gerrit folks: you may be interested to know that numeric change ids (indeed, all formats of change ids) are no longer deprecated (ie, they will continue to be supported): https://gerrit-review.googlesource.com/c/gerrit/+/256792 | 15:09 |
| fungi | yay! that's great news | 15:12 |
| mordred | that IS great news! | 15:13 |
| * mordred has been dreading that future removal | 15:13 | |
| tosky | oooh, people realized that compatibility is not a bad thing | 15:15 |
| *** felixedel has joined #zuul | 15:23 | |
| *** felixedel has quit IRC | 15:23 | |
| *** jamesmcarthur has quit IRC | 15:41 | |
| *** jcapitao is now known as jcapitao_afk | 15:47 | |
| *** jamesmcarthur has joined #zuul | 15:52 | |
| *** jamesmcarthur has quit IRC | 15:57 | |
| openstackgerrit | Jan Kubovy proposed zuul/zuul master: Refactor merge_client in tests https://review.opendev.org/709676 | 16:00 |
| *** bertu has quit IRC | 16:04 | |
| *** jcapitao_afk is now known as jcapitao | 16:05 | |
| *** y2kenny has joined #zuul | 16:09 | |
| openstackgerrit | Jan Kubovy proposed zuul/zuul master: Refactor nodepool in tests https://review.opendev.org/709703 | 16:10 |
| y2kenny | Hi, I just started looking into Zuul. If I connect Zuul to Gerrit, what happens to Gerrit changes if the connection between Zuul and Gerrit is disrupted? (for example, if stream events are dropped/lost?) How does Zuul recover? | 16:12 |
| clarkb | y2kenny: you'll have a gap in time where events are not received | 16:13 |
| clarkb | Zuul also has experimental support for Gerrit's new checks api plugin (which polls and would notice things were missed) | 16:13 |
| clarkb | in general though stream events has been pretty solid for us (opendev). We miss events due to outages in Zuul (upgrades etc) and not due to network connectivity or gerrit ssh problems | 16:14 |
| y2kenny | ok. Are there ways to tell Zuul to bulk check things manually? (or for me to "inject" potentially missed events?) I think Gerrit and Zuul are great. I am only asking because the network environment I am in can be flicky sometimes. | 16:17 |
| clarkb | y2kenny: yes, the way we do that is we add a magic comment string that triggers zuul pipelines (in our case 'recheck'). This allows anyone with comment ability in gerrit to trigger those pipelines. You can also use the zuul-enqueue utility (and maybe the api now?) to enqueue those as an admin | 16:18 |
| clarkb | we tend to use a mixture of both. Let anyone leave comments in gerrit for rerunning premerge tests, then rely on admins for post merge enqueues. The reason for that is post merge tends to have side effects you may need to think about before running them (though that is possibly specific to our use of those pipelines) | 16:20 |
| *** mattw4 has joined #zuul | 16:20 | |
| y2kenny | cool. I saw the recheck bit in the zuul quick start tutorial but did not know about the zuul-enqueue utility. I will check that out. | 16:21 |
| y2kenny | thanks for the tips! | 16:21 |
| corvus | y2kenny, clarkb: actually it's "zuul enqueue" (ie, the command is zuul, with enqueue as the subcommand) | 16:22 |
| y2kenny | corvus: ok, thanks! | 16:22 |
| corvus | y2kenny: https://zuul-ci.org/docs/zuul/reference/client.html#enqueue | 16:23 |
| *** jamesmcarthur has joined #zuul | 16:26 | |
| *** harrymichal has quit IRC | 16:34 | |
| *** y2kenny has left #zuul | 16:34 | |
| clarkb | corvus: mordred I think we can restart opendev zuul to pick up that change | 16:36 |
| corvus | yep, i'll do that | 16:37 |
| openstackgerrit | Jan Kubovy proposed zuul/zuul master: Refactor zookeeper in tests https://review.opendev.org/709709 | 16:39 |
| openstackgerrit | Jan Kubovy proposed zuul/zuul master: Consolidate scheduler pause/exit as hibernation https://review.opendev.org/709723 | 16:51 |
| *** tosky has quit IRC | 17:09 | |
| *** michael-beaver has joined #zuul | 17:11 | |
| openstackgerrit | Jan Kubovy proposed zuul/zuul master: Refactor `self.event_queues` in tests https://review.opendev.org/709990 | 17:12 |
| *** jcapitao has quit IRC | 17:14 | |
| openstackgerrit | Jan Kubovy proposed zuul/zuul master: Scheduler's pause/resume functionality https://review.opendev.org/709735 | 17:34 |
| *** evrardjp has quit IRC | 17:35 | |
| *** evrardjp has joined #zuul | 17:35 | |
| zenkuro | hi, is there anything necessary to make a commit on zuul repo besedes ubuntu one account? | 17:41 |
| clarkb | zenkuro: you need to create a gerrit account (via ubuntu on login) and add an ssh key to push the code (or use http auth, but ssh will be simpler if it works for you) | 17:42 |
| clarkb | that should be it | 17:43 |
| *** jpena is now known as jpena|off | 17:50 | |
| openstackgerrit | Jan Kubovy proposed zuul/zuul master: WIP: Store unparsed branch config in Zookeeper https://review.opendev.org/705716 | 17:50 |
| fungi | was there any story on storyboard related to https://review.opendev.org/710287 ? | 17:52 |
| fungi | just want to make sure we switch it to public if so | 17:52 |
| tristanC | fungi: i just made it public https://storyboard.openstack.org/#!/story/2007327 | 17:53 |
| fungi | thanks! | 17:54 |
| fungi | i've added the zuul-security story tag, since that's what we usually rely on to query for them | 17:55 |
| fungi | probably easiest workflow is to ass the zuul-security when triaging a private security story, make it public shortly before pushing a patch to gerrit, and reference the task id in the commit message | 17:56 |
| fungi | that way we get the relevant updates | 17:56 |
| fungi | s/ass/add/ obviously. it's friday | 17:57 |
| fungi | i'll eventually get something written up about a vulnerability workflow for our maintainer docs | 17:57 |
| fungi | now that we've mostly got it figured out | 17:58 |
| *** jamesmcarthur has quit IRC | 18:06 | |
| paladox | corvus https://gerrit-review.googlesource.com/c/gerrit/+/257077 awesome! | 18:08 |
| zenkuro | clarkb: git clone ssh://zenkuro@review.opendev.org/zuul/zuul is it correct? It says I have no access writes or repo do not exists | 18:08 |
| fungi | zenkuro: git clone https://opendev.org/zuul/zuul | 18:10 |
| clarkb | zenkuro: gerrit uses port 29418 (not default of 22) for ssh. Typically though you'll clone from https://opendev.org/zuul/zuul then run `git review -s` to configure the gerrit remote | 18:10 |
| clarkb | then you can use gerrit just for pushing proposed changes using the git review tool | 18:10 |
| fungi | use the https://pypi.org/project/git-review tool to submit changes | 18:10 |
| fungi | it will interpret the correct gerrit remote from the .gitreview file stored in the repository | 18:11 |
| corvus | paladox: thanks! https://gerrit-review.googlesource.com/c/zuul/jobs/+/257078 found a bug, so i'm fixing that now; i think i should be able to remove WIP on those and send an update email soon | 18:20 |
| paladox | oh, heh | 18:21 |
| clarkb | it would be really neat if the "relationship chain" was a graph. | 18:23 |
| clarkb | (I filed a bug about that like 4 yaers ago and am sad it never happened with all of the other UI work) | 18:23 |
| clarkb | also I guess the checks api doesn't incorporate voting and that still happens as regular comments so you get them in the comments list? | 18:25 |
| paladox | nope | 18:25 |
| paladox | checkers is not a comment :) | 18:25 |
| paladox | you click on the checks tab | 18:25 |
| clarkb | paladox: ya I'm more concerned that we've gone out of our way to remove the vote comments from the ui and sumarrize them in the table instead | 18:26 |
| clarkb | (for some reason I thought the checks api sort of approximated that with its table and voting, but it doesn't) | 18:26 |
| clarkb | mordred: ^ something to think about when upgrading. We may still need js to clean that up | 18:26 |
| fungi | or does the checks api allow you to set a status which can be used to signal/block? | 18:27 |
| mordred | what fungi said | 18:27 |
| mordred | the subchecks spec that corvus has been working on is the key part there | 18:28 |
| fungi | yeah, so basically "votes" is oldthink anyway | 18:28 |
| clarkb | mordred: subchecks in order to represent check and gate? | 18:28 |
| fungi | so people "vote" on changes, bots set a status of some sort | 18:28 |
| mordred | clarkb, fungi: https://gerrit-review.googlesource.com/c/homepage/+/235692 and https://gerrit-review.googlesource.com/c/homepage/+/235693 | 18:28 |
| mordred | it won't be usable by opendev until subchecks exists | 18:29 |
| mordred | (I'd try to summarize in irc real quick - but honestly it's one of those topics where the details are the important part) | 18:30 |
| clarkb | mordred: that splits jobs out so you can see them seprate from the top level yay or nay right? but how do you express that there are two levels of pre merge testing? | 18:30 |
| clarkb | mordred: maybe thats just two different top level checkers and you need both to merge? | 18:30 |
| openstackgerrit | Zen Kurosaky proposed zuul/zuul master: This patch extends zuul deployment from scratch documentation. it adds guide on MySQL setup it extends zuul-from-scratch setup by adding mysql connection to zuul.conf file it extends gerrit setup by adding Verified label to gerrit it exten https://review.opendev.org/710504 | 18:31 |
| mordred | clarkb: yeah - for opendev we'd have a "Zuul Check" checker and a "Zuul Gate" checker - and those would be required. then Zuul would define one or more subchecks corresponding to each job, and should mark them as either required or not required based on non-voting flag | 18:31 |
| mordred | we would also be able to define Checkers for 3pci systems (obviously marked as not required) | 18:32 |
| clarkb | ok so you'll have a "Gate missing" or whatever message until it runs and reports back. Thats actually nice if it renders that way as it will help people understand what they need next in order to merge | 18:32 |
| mordred | yeah - I thnk it's going to be VERY nice once it's all in place | 18:32 |
| zenkuro | many thanks fugi, clarkb, and corvus! I havent specified account on gerrit >_< | 18:33 |
| clarkb | mordred: what that doesn't do though is force third party ci to use the checker api (and i'm not sure we can turn off stream events because we need the post merge stuff) | 18:33 |
| corvus | mordred, clarkb: re gate -- i think we can actually add a gate to the zuul/* repos on gerrit's gerrit and see that in action. | 18:33 |
| clarkb | corvus: ooh that would be cool | 18:33 |
| corvus | i don't think we've sold the gerrit community on that in general, but i doubt they would object to seeing that on those repos, and i think it would help exercise checks | 18:34 |
| clarkb | ++ | 18:34 |
| mordred | ++ | 18:34 |
| mordred | clarkb: luca tells me the webhooks plugin should be sufficient for post-merge things ... but I agree, I think we'll have stream-events for the forseeable future | 18:35 |
| corvus | paladox, clarkb, mordred: and re voting -- i think there's some rough consensus that at the moment, it's helpful to have checkers vote as well. so i think i will propose that for gerrit's zuul. (note that the existing gerritforge ci system does use checkers and votes) | 18:35 |
| corvus | eg https://gerrit-review.googlesource.com/c/gerrit/+/257073 | 18:36 |
| clarkb | corvus: ya gerritforge is how I noticed that. I my concern is that I had thought we were going to get awy from that noise and could delete some js | 18:36 |
| paladox | yeh, i think someone on the gerrit groups, there's something they wrote that needs to be addressed before they switch off using v+1 for gerrit-forge | 18:36 |
| clarkb | but I guess we don't have to run our CI systems against our gerrit that way | 18:36 |
| corvus | yes i think that is the goal | 18:36 |
| mordred | clarkb: that said - once it's there and in place, we can incentivize 3pci to use checks instead of comments - since we won't have a hideci.js to create a table of results for 3pci leaving results in comments | 18:36 |
| corvus | i think it's just "not quite there yet" | 18:36 |
| corvus | clarkb, paladox, mordred: see this message i wrote (which started being about gertty, but is relevant in general): https://groups.google.com/forum/#!topic/repo-discuss/mubpTaKHUZ4 | 18:37 |
| mordred | clarkb: it's not even so much that we *can* delete some js - it's more that we're going to delete that js when we upgrade to polygerrit and checks is the path to things being ina . table of results :) | 18:37 |
| paladox | mordred seems that i think Thomas has added a plugin entry point to make hideci possible in PG *i think*. | 18:37 |
| mordred | paladox: right - but we're looking towards checks and don't want hideci :) | 18:38 |
| paladox | corvus ohh, reading! | 18:38 |
| paladox | mordred :D | 18:38 |
| paladox | We made progress on our upgrade: https://gerrit.wikimedia.org/r/#/c/operations/software/gerrit/+/574092/ :) | 18:39 |
| paladox | Just need to roll that out to a test instance, run the schema upgrade and if it passes we can roll it out to prod at a specified date. | 18:40 |
| mordred | paladox: nice! | 18:40 |
| corvus | paladox: exciting! mordred has been busy working on opendev's as well | 18:40 |
| mordred | yah. we're getting close as well | 18:41 |
| corvus | the work to set up zuul jobs in gerrit is feeding back into opendev, and vice-versa. so our build jobs have been changing and are starting to converge | 18:41 |
| paladox | yup! I would like for us to upgrade this month... | 18:42 |
| corvus | https://gerrit-review.googlesource.com/c/zuul/jobs/+/257080 is the 4th interation of the code to prepare gerrit submodule in zuul jobs -- the first was in opendev, the second in gerrit, the third in opendev again, and that's the 4th back in gerrit... when we're done, we'll only have one copy and use it in both places | 18:43 |
| paladox | only because i want to give our users time to get use to the new ui. | 18:43 |
| corvus | https://review.opendev.org/710489 is prep work to start using that directly | 18:43 |
| clarkb | zenkuro: left a few notes on that change. | 18:44 |
| corvus | when we have that working, i'll write that up for the repo-discuss list too, so folks can see it. cross-community sharing of ci implementation is pretty novel. :) | 18:44 |
| zenkuro | clarkb: write it here or add to gerrit? | 18:45 |
| clarkb | zenkuro: not sure if you are familiar with updating gerrit changes but typically you will udpate the existing change (either git commit --amend or git rebase -i and squash back) then rerun git review and it will update the existing change as long as you keep the same change-id in the commit message | 18:45 |
| clarkb | zenkuro: in gerrit on the change (you might have to refresh the page in your brwoser to see them) | 18:46 |
| paladox | corvus should "- jgit" that be "- modules/jgit"? | 18:46 |
| clarkb | zenkuro: overall it looks good. Suggestions are mostly minor edits | 18:47 |
| paladox | oh, i see i've misread it, it's cloning that. Sorry :) | 18:47 |
| corvus | paladox: that's the name of the repo in gerrit (it's just "jgit"); the role has a mapping table that tells where to put projects that go into places other than their name: https://gerrit.googlesource.com/zuul/jobs/+/refs/heads/master/roles/prepare-gerrit-repos/defaults/main.yaml | 18:48 |
| corvus | paladox: (and zuul would have failed the change with an error if the project name were wrong) | 18:48 |
| paladox | ah | 18:48 |
| corvus | paladox: you can see what a zuul syntax error looks like on PS6 of this change: https://gerrit-review.googlesource.com/c/zuul/jobs/+/255652 | 18:49 |
| *** jamesmcarthur has joined #zuul | 18:49 | |
| *** jamesmcarthur_ has joined #zuul | 18:50 | |
| paladox | Hmm, seems to https://ci.gerritcodereview.com/t/gerrit/buildset/f852104e0abc4a7cba30e84a412c2aee stay stuck loading. | 18:50 |
| corvus | neat | 18:50 |
| paladox | in the console i see "Unhandled Promise Rejection: TypeError: undefined is not an object (evaluating 'e.request.responseURL')" | 18:51 |
| corvus | i see an "e.request is undefined" | 18:52 |
| *** jamesmcarthur has quit IRC | 18:54 | |
| *** jamesmcarthur_ has quit IRC | 18:54 | |
| *** mattw4 has quit IRC | 18:54 | |
| *** jamesmcarthur has joined #zuul | 18:56 | |
| *** jamesmcarthur_ has joined #zuul | 18:56 | |
| zenkuro | clarkb: gerrit do only supports rsa keys? | 18:57 |
| paladox | corvus https://opendev.org/zuul/zuul/src/branch/master/web/src/actions/errors.js#L27 | 18:59 |
| paladox | i think that can be "url: error && error.request && error.request.responseURL || error.url" | 18:59 |
| *** jamesmcarthur has quit IRC | 19:00 | |
| clarkb | zenkuro: older gerrit does, newer gerrit supports more. But your examples use rsa filenames. I think you should set -t foo and the filenames to match just to avoid any changing of keygen defaults | 19:00 |
| corvus | paladox: ah thanks, yep! and that gets me to the real error, which is that the builds value is not returned in the json response if there are no builds | 19:01 |
| paladox | :) | 19:01 |
| corvus | so we either need to handle that being absent, or update the server to return the empty list | 19:01 |
| corvus | tristanC: ^ do you have a preference? | 19:02 |
| mordred | corvus, paladox: https://review.opendev.org/709602 Use prepare-gerrit-repo role for gerrit submodules ... there's the stab at using prepare-gerrit-repos from opendev ... although now that I'm looking at it I realize I removed one too many things so that's not going to do what we want | 19:06 |
| openstackgerrit | James E. Blair proposed zuul/zuul master: web ui: fix buildset display when no builds https://review.opendev.org/710508 | 19:07 |
| corvus | paladox: ^ that fixes it for me locally; thanks for catching that! | 19:08 |
| mordred | corvus: the current thign in tree has a "gerrit_additional_plugins" which tells the job which repos should also be copied into the gerrit tree ... we still need that. BUT - I think I could rewrite it to look for things in the projects list that don't already exist in the plugins dir | 19:08 |
| corvus | mordred: ack; i'll take a look after lunch; gotta run, biab | 19:08 |
| mordred | corvus: have good lunch | 19:08 |
| paladox | corvus thanks! | 19:09 |
| paladox | mordred nice! | 19:10 |
| openstackgerrit | Zen Kurosaky proposed zuul/zuul master: This patch extends zuul deployment from scratch documentation. https://review.opendev.org/710504 | 19:17 |
| *** jamesmcarthur_ has quit IRC | 19:19 | |
| tristanC | corvus: i have no preference, it seems like there are a few edge case the webui doesn't handle well | 19:20 |
| mordred | tristanC: corvus pushed up a "fix it in the ui" patch :) | 19:21 |
| *** mattw4 has joined #zuul | 19:22 | |
| openstackgerrit | Zen Kurosaky proposed zuul/zuul master: This patch extends zuul deployment from scratch documentation. https://review.opendev.org/710504 | 19:22 |
| *** sgw has quit IRC | 19:22 | |
| zenkuro | clarkb: I will think on how to get rid of rsa... thanks for advises. I think I fixed most of them | 19:23 |
| *** felixedel has joined #zuul | 19:25 | |
| *** jamesmcarthur has joined #zuul | 19:26 | |
| *** jamesmcarthur_ has joined #zuul | 19:26 | |
| *** jamesmcarthur has quit IRC | 19:30 | |
| *** EmilienM is now known as EvilienM | 19:30 | |
| *** jamesmcarthur_ has quit IRC | 19:33 | |
| *** sgw has joined #zuul | 19:41 | |
| mordred | corvus: when you get back ... I've produced an Unknown configuation error! | 19:41 |
| fungi | that sounds glorious | 19:41 |
| mordred | yeah | 19:42 |
| mordred | I think it's related to system-config being in opendev and openstack again ... but still, it's pretty glorious | 19:42 |
| *** dpawlik has quit IRC | 19:43 | |
| *** dpawlik has joined #zuul | 19:43 | |
| mordred | http://paste.openstack.org/show/790132/ | 19:43 |
| *** mattw4 has quit IRC | 19:44 | |
| *** mattw4 has joined #zuul | 19:46 | |
| fungi | so the parent_layout came through as None? | 19:52 |
| mordred | yeah | 19:54 |
| pabelanger | We've upgraded zuul to 3.17.0, thanks! About to try new check API integration for github | 19:55 |
| mordred | pabelanger: woot! | 19:55 |
| mordred | pabelanger: oh - wait | 19:55 |
| mordred | pabelanger: I don't think check api is in 3.17 | 19:55 |
| pabelanger | oh, relnotes says it is | 19:56 |
| mordred | oh - yes | 19:56 |
| mordred | it's the inline comments that just missed 3.17 | 19:56 |
| mordred | you're probably fine without that :) | 19:56 |
| pabelanger | ack | 19:56 |
| *** jamesmcarthur has joined #zuul | 19:58 | |
| *** jamesmcarthur_ has joined #zuul | 19:59 | |
| mordred | corvus, fungi: yup - the issue is just on the opendev tenant - the openstack tenant job is running fine | 20:03 |
| mordred | although there is an error I'd liek to ponder what the best way to deal with is | 20:03 |
| *** jamesmcarthur has quit IRC | 20:03 | |
| mordred | oh - I think I've got it | 20:03 |
| *** jamesmcarthur_ has quit IRC | 20:18 | |
| pabelanger | yay | 20:19 |
| pabelanger | https://github.com/ansible-collections/openvswitch/pull/4/checks | 20:19 |
| *** jamesmcarthur has joined #zuul | 20:19 | |
| *** jamesmcarthur_ has joined #zuul | 20:20 | |
| pabelanger | with checks now, i guess that could be broken out into jobs vs buildset | 20:22 |
| *** jamesmcarthur has quit IRC | 20:24 | |
| mordred | pabelanger: woot! | 20:47 |
| pabelanger | neat, summary of results looks good too: https://github.com/ansible-collections/openvswitch/pull/4/checks?check_run_id=475685548 | 20:56 |
| pabelanger | waiting to get failure results, but if they also work, I think we could stop commenting directly on the PR | 20:56 |
| pabelanger | which some folks do not like and consider spam | 20:56 |
| *** jamesmcarthur_ has quit IRC | 21:03 | |
| mordred | pabelanger: yay improvements! | 21:03 |
| pabelanger | and failure | 21:04 |
| pabelanger | https://github.com/ansible/project-config/pull/371/checks?check_run_id=475745916 | 21:04 |
| pabelanger | I wonder if that 'recheck' button can be disabled | 21:04 |
| pabelanger | or if that is an event that zuul sees | 21:04 |
| pabelanger | oh, maybe that is check_run | 21:04 |
| mordred | pabelanger: I don't see recheck - so prolly only a committer visible thing? | 21:05 |
| pabelanger | oh, maybe | 21:05 |
| pabelanger | need to read up | 21:05 |
| mordred | corvus: interesting. I just got "Change in gerrit/jobs[master]: Copy release.war into zuul artifact location" in email for the change I pushed up to zuul/jobs ... I'm guessing there's something that rename-project may not have updated | 21:29 |
| corvus | was that project renamed? i thought it was created that way | 21:32 |
| *** armstrongs has joined #zuul | 21:35 | |
| mordred | corvus: dunno - not sure why it's saying gerrit/jobs in the email then | 21:36 |
| *** mattw4 has quit IRC | 21:46 | |
| pabelanger | regarding https://zuul-ci.org/docs/zuul/reference/drivers/github.html#attr-pipeline.require.%3Cgithub%20source%3E.status that should also work for checks API too? | 21:47 |
| pabelanger | that might be a tobiash question | 21:48 |
| clarkb | pabelanger: is the checks api used to trigger or just to report? | 22:00 |
| pabelanger | we use it to trigger, so we know the check jobs first all passed | 22:02 |
| pabelanger | from what I see, the checks api does not show up as a statuses now | 22:02 |
| pabelanger | https://api.github.com/repos/ansible/project-config/statuses/9d6cec08710162593ddf3e63731168c5f3377518 | 22:07 |
| pabelanger | those are the results for status, we set | 22:07 |
| pabelanger | but you can see https://github.com/ansible/project-config/pull/375 there are 2 ansible/check now | 22:07 |
| *** bolg has quit IRC | 22:07 | |
| pabelanger | sorry, internet died | 22:07 |
| *** rlandy has quit IRC | 22:21 | |
| *** tosky has joined #zuul | 22:24 | |
| *** mattw4 has joined #zuul | 22:54 | |
| *** EvilienM is now known as EmilienM | 22:56 | |
| openstackgerrit | Merged zuul/zuul master: Stream output from kubectl pods https://review.opendev.org/709261 | 23:07 |
| *** dpawlik has quit IRC | 23:13 | |
| *** mattw4 has quit IRC | 23:29 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!