| *** rlandy|ruck2 has quit IRC | 00:01 | |
| corvus | mordred: oh, we use buildx for nodepool -- looks like we have more cleanup to do: https://zuul.opendev.org/t/zuul/build/a370356fb916463ebdac72b0698a70ed | 00:05 |
|---|---|---|
| *** rfolco has quit IRC | 00:28 | |
| *** wuchunyang has joined #zuul | 01:01 | |
| *** Goneri has quit IRC | 01:07 | |
| openstackgerrit | Guillaume Chauvel proposed zuul/zuul master: [DNM] TEST zuul_stream+callback test using multinode jobs https://review.opendev.org/735483 | 01:09 |
| openstackgerrit | Guillaume Chauvel proposed zuul/zuul master: Add 'uuid' to 'src_dir' in order to allow parallel jobs for a static node https://review.opendev.org/735981 | 01:09 |
| openstackgerrit | Guillaume Chauvel proposed zuul/zuul master: [DNM] TEST 'uuid' in 'src_dir': 1 node, 4 parallel jobs https://review.opendev.org/735576 | 01:09 |
| *** wuchunyang has quit IRC | 01:20 | |
| *** wuchunyang has joined #zuul | 01:43 | |
| *** wuchunyang has quit IRC | 01:47 | |
| *** swest has quit IRC | 01:49 | |
| *** jamesmcarthur has joined #zuul | 02:02 | |
| *** swest has joined #zuul | 02:04 | |
| *** jamesmcarthur has quit IRC | 02:08 | |
| *** jamesmcarthur has joined #zuul | 02:08 | |
| *** jamesmcarthur has quit IRC | 02:15 | |
| *** jamesmcarthur has joined #zuul | 02:24 | |
| *** jamesmcarthur has quit IRC | 02:37 | |
| *** jamesmcarthur has joined #zuul | 02:38 | |
| *** wuchunyang has joined #zuul | 02:38 | |
| *** sgw has quit IRC | 02:42 | |
| *** jamesmcarthur has quit IRC | 02:43 | |
| *** bhavikdbavishi has joined #zuul | 02:44 | |
| *** sgw has joined #zuul | 03:00 | |
| *** wuchunyang has quit IRC | 03:04 | |
| *** bhavikdbavishi1 has joined #zuul | 03:05 | |
| *** bhavikdbavishi has quit IRC | 03:07 | |
| *** bhavikdbavishi1 is now known as bhavikdbavishi | 03:07 | |
| *** jamesmcarthur has joined #zuul | 03:19 | |
| *** jamesmcarthur has quit IRC | 03:22 | |
| *** jamesmcarthur has joined #zuul | 03:23 | |
| *** harrymichal has quit IRC | 03:35 | |
| *** wuchunyang has joined #zuul | 03:37 | |
| *** jamesmcarthur has quit IRC | 03:58 | |
| *** wuchunyang has quit IRC | 03:58 | |
| *** jamesmcarthur has joined #zuul | 03:58 | |
| *** wuchunyang has joined #zuul | 04:01 | |
| *** jamesmcarthur has quit IRC | 04:04 | |
| *** ysandeep|away is now known as ysandeep | 04:05 | |
| *** wuchunyang has quit IRC | 04:06 | |
| *** sgw has quit IRC | 04:08 | |
| *** sgw has joined #zuul | 04:09 | |
| *** sgw1 has quit IRC | 04:09 | |
| *** evrardjp has quit IRC | 04:33 | |
| *** evrardjp has joined #zuul | 04:33 | |
| *** wuchunyang has joined #zuul | 04:39 | |
| *** jamesmcarthur has joined #zuul | 04:40 | |
| *** jamesmcarthur has quit IRC | 04:44 | |
| *** jamesmcarthur has joined #zuul | 04:45 | |
| *** wuchunyang has quit IRC | 04:47 | |
| *** sgw has quit IRC | 05:03 | |
| *** jamesmcarthur has quit IRC | 05:09 | |
| *** jamesmcarthur has joined #zuul | 05:09 | |
| *** sgw has joined #zuul | 05:10 | |
| *** sgw1 has joined #zuul | 05:10 | |
| *** jamesmcarthur has quit IRC | 05:15 | |
| *** bhavikdbavishi has quit IRC | 05:37 | |
| *** armstrongs has joined #zuul | 05:51 | |
| *** jamesmcarthur has joined #zuul | 05:52 | |
| *** bhavikdbavishi has joined #zuul | 05:55 | |
| *** armstrongs has quit IRC | 06:01 | |
| *** bhavikdbavishi1 has joined #zuul | 06:04 | |
| *** bhavikdbavishi has quit IRC | 06:06 | |
| *** bhavikdbavishi1 is now known as bhavikdbavishi | 06:06 | |
| *** jamesmcarthur has quit IRC | 06:06 | |
| *** armstrongs has joined #zuul | 06:28 | |
| *** tosky has joined #zuul | 06:30 | |
| *** harrymichal has joined #zuul | 06:36 | |
| *** armstrongs has quit IRC | 06:36 | |
| *** rpittau|afk is now known as rpittau | 06:44 | |
| *** harrymichal has quit IRC | 07:01 | |
| *** harrymichal has joined #zuul | 07:01 | |
| *** bhavikdbavishi has quit IRC | 07:05 | |
| openstackgerrit | Matthieu Huin proposed zuul/zuul master: web UI: allow a privileged user to dequeue a change https://review.opendev.org/734850 | 07:08 |
| openstackgerrit | Matthieu Huin proposed zuul/zuul master: web UI: allow a privileged user to re-enqueue a change https://review.opendev.org/736772 | 07:09 |
| openstackgerrit | Matthieu Huin proposed zuul/zuul master: web UI: allow a privileged user to re-enqueue a change https://review.opendev.org/736772 | 07:11 |
| mhu | can someone explain to me what the promote subcommand of the CLI does exactly? My understanding is that it moves a change at the top of pipeline queue, to be built in priority. Is that correct? | 07:13 |
| mhu | Also does it apply only to dependent pipelines like gate? | 07:13 |
| openstackgerrit | Matthieu Huin proposed zuul/zuul master: Builds: add "held" attribute https://review.opendev.org/728994 | 07:14 |
| openstackgerrit | Matthieu Huin proposed zuul/zuul master: Web UI: add checkbox filters to TableFilters component, filter builds by held status https://review.opendev.org/729265 | 07:15 |
| *** bhavikdbavishi has joined #zuul | 07:19 | |
| *** jcapitao has joined #zuul | 07:22 | |
| openstackgerrit | Felix Edel proposed zuul/zuul master: WIP: Introduce Patternfly 4 https://review.opendev.org/736225 | 07:23 |
| *** hashar has joined #zuul | 07:30 | |
| *** bhagyashris|away is now known as bhagyashris | 07:39 | |
| *** felixedel has joined #zuul | 07:43 | |
| frickler | the stage-output role claims to "Copies contents into {{ zuul_output_dir }}" but actually uses "stage_dir" as variable. Is just the comment in the readme wrong or should the variable be changed? | 07:45 |
| frickler | also maybe someone wants to build a test for https://review.opendev.org/736714 | 07:47 |
| *** michael-beaver has quit IRC | 07:52 | |
| sshnaidm|off | please review/merge https://review.opendev.org/#/c/730360/ - zuul roles for working with ansible collections | 07:55 |
| *** jpena|off is now known as jpena | 07:57 | |
| *** nils has joined #zuul | 08:01 | |
| *** wuchunyang has joined #zuul | 08:04 | |
| *** felixedel has quit IRC | 08:23 | |
| openstackgerrit | Matthieu Huin proposed zuul/zuul master: REST API: improve tenant scoping of autohold, authorizations https://review.opendev.org/728118 | 08:30 |
| avass | Today's the day for eating herring, drinking schnapps and dancing around a maypole. Happy midsummer everybody! | 08:40 |
| mhu | skol ! | 08:41 |
| mhu | (altough it might be a bit early) | 08:41 |
| avass | skål! | 08:46 |
| *** mugsie has quit IRC | 08:48 | |
| *** mugsie has joined #zuul | 08:48 | |
| *** jamesmcarthur has joined #zuul | 09:21 | |
| *** clarkb has quit IRC | 09:23 | |
| *** jamesmcarthur has quit IRC | 09:27 | |
| *** clarkb has joined #zuul | 09:28 | |
| *** ysandeep is now known as ysandeep|brb | 09:29 | |
| openstackgerrit | Matthieu Huin proposed zuul/zuul master: REST API, Web UI: add pipelines' manager, triggers data in status https://review.opendev.org/736968 | 10:07 |
| *** bhavikdbavishi has quit IRC | 10:08 | |
| *** rpittau is now known as rpittau|bbl | 10:22 | |
| openstackgerrit | Matthieu Huin proposed zuul/zuul master: REST API, Web UI: add pipelines' manager, triggers data in status https://review.opendev.org/736968 | 10:23 |
| *** wuchunyang has quit IRC | 10:29 | |
| *** vishalmanchanda has joined #zuul | 10:34 | |
| *** bhavikdbavishi has joined #zuul | 10:35 | |
| *** jcapitao is now known as jcapitao_lunch | 10:47 | |
| *** harrymichal has quit IRC | 10:51 | |
| *** harrymichal has joined #zuul | 10:52 | |
| *** jamesmcarthur has joined #zuul | 11:24 | |
| *** wuchunyang has joined #zuul | 11:28 | |
| *** jamesmcarthur has quit IRC | 11:31 | |
| *** harrymichal has quit IRC | 11:32 | |
| *** harrymichal has joined #zuul | 11:32 | |
| *** wuchunyang has quit IRC | 11:33 | |
| vishalmanchanda | Hi all | 11:40 |
| vishalmanchanda | Due to recent changes in zuul horizon node gate job is failing. | 11:41 |
| *** jpena is now known as jpena|lunch | 11:41 | |
| vishalmanchanda | I am not sure but it happened after this patch https://review.opendev.org/#/c/736330 is merged. | 11:41 |
| vishalmanchanda | Can Someone take a look... | 11:41 |
| vishalmanchanda | thanks | 11:41 |
| *** bhavikdbavishi has quit IRC | 11:53 | |
| *** bhavikdbavishi has joined #zuul | 11:53 | |
| *** rlandy has joined #zuul | 11:56 | |
| *** rlandy is now known as rlandy|ruck | 11:57 | |
| *** bhavikdbavishi has quit IRC | 11:57 | |
| openstackgerrit | Felix Edel proposed zuul/zuul master: WIP: Introduce Patternfly 4 https://review.opendev.org/736225 | 12:08 |
| *** ysandeep|brb is now known as ysandeep | 12:11 | |
| *** rfolco has joined #zuul | 12:12 | |
| openstackgerrit | Benjamin Schanzel proposed zuul/nodepool master: Kubernetes Driver: Allow cpu/mem resource limits https://review.opendev.org/736984 | 12:17 |
| *** rpittau|bbl is now known as rpittau | 12:19 | |
| *** jcapitao_lunch is now known as jcapitao | 12:20 | |
| openstackgerrit | Felix Edel proposed zuul/zuul master: Introduce Patternfly 4 https://review.opendev.org/736225 | 12:23 |
| AJaeger | vishalmanchanda: do you have a link to a failed job? | 12:27 |
| vishalmanchanda | AJaeger: https://zuul.opendev.org/t/openstack/build/00049c2aff4742ce8a2b96dbc0be4708/log/job-output.txt#3652-3693 | 12:28 |
| AJaeger | tox not found, I see... | 12:30 |
| AJaeger | vishalmanchanda: in horizon's package.json this fails: "postinstall": "if [ ! -d .tox/npm ]; then tox -e npm --notest; fi", | 12:33 |
| AJaeger | vishalmanchanda: the job is not installing tox, let's add that to a pre-run for your job, you need ensure-tox role there | 12:34 |
| AJaeger | vishalmanchanda: can you do that or shall I send a patch and let you take over? | 12:34 |
| vishalmanchanda | AJaeger: I think we have to add a role here https://github.com/openstack/horizon/blob/master/playbooks/npm-test-xstatic-master/pre.yaml | 12:35 |
| AJaeger | vishalmanchanda: not for horizon-nodejs10-run-test | 12:36 |
| openstackgerrit | Benjamin Schanzel proposed zuul/nodepool master: Kubernetes Driver: Allow cpu/mem resource limits https://review.opendev.org/736984 | 12:36 |
| AJaeger | vishalmanchanda: let me send a first patch | 12:37 |
| vishalmanchanda | AJaeger: I am not so familiar with this part. | 12:37 |
| vishalmanchanda | AJaeger: Could you push a patch for adding the 'ensure-tox role' and let' see.. | 12:38 |
| openstackgerrit | Guillaume Chauvel proposed zuul/nodepool master: Initial change to rename zuul-quick-start to zuul-tutorial-quick-start https://review.opendev.org/728262 | 12:39 |
| AJaeger | vishalmanchanda: on my way... | 12:39 |
| vishalmanchanda | AJaeger: thanks for looking into it. | 12:39 |
| AJaeger | vishalmanchanda: let's see whether https://review.opendev.org/736992 fixes it | 12:41 |
| vishalmanchanda | AJaeger: +1. | 12:41 |
| *** jpena|lunch is now known as jpena | 12:43 | |
| openstackgerrit | Guillaume Chauvel proposed zuul/zuul master: Fix zuul_run_command ret when command is not found https://review.opendev.org/735475 | 12:45 |
| openstackgerrit | Guillaume Chauvel proposed zuul/zuul master: zuul_stream: close streamer when task is skipped https://review.opendev.org/735476 | 12:45 |
| openstackgerrit | Guillaume Chauvel proposed zuul/zuul master: Add 'uuid' to 'src_dir' in order to allow parallel jobs for a static node https://review.opendev.org/735981 | 12:45 |
| *** harrymichal has quit IRC | 12:52 | |
| *** harrymichal has joined #zuul | 12:53 | |
| ttx | Hi all, was wondering if semaphores could be used to avoid the errors when parallel mirroring operations occur in https://zuul.opendev.org/t/openstack/builds?job_name=openstack-upload-github-mirror# | 12:53 |
| ttx | ideally it would be a per-repo semaphore, but since semaphores need to be pre-declared we could settle with a global github-mirroring semaphore | 12:54 |
| ttx | also what's a "resource" in the context of job.semaphore.resources-first ? | 12:56 |
| ttx | The glossary does not tell | 12:56 |
| AJaeger | ttx, a semaphore could declare that only 1 (or whatever you configure) jobs run in parallel, so it would serialize everything if you set it to 1 | 12:57 |
| ttx | right | 12:57 |
| ttx | that would allow to serialize that job but not anything else in the post pipeline | 12:57 |
| *** harrymichal has quit IRC | 12:57 | |
| AJaeger | correct | 12:57 |
| AJaeger | WE use a semaphore for translations - but limit to 10; our translation server cannot handle a higher load... | 12:58 |
| *** harrymichal has joined #zuul | 12:58 | |
| ttx | here it's more to avoid two updates on the same repo at the same time, It would avoid two updates at the same time, period. Which is a bit too much, but acceptable if there is no backlog forming. Which is why I was curious about resources-first | 12:59 |
| AJaeger | ttx, this runs in post - corvus, don't we "merge" multiple jobs in post anymore? Or is that only in promote? | 12:59 |
| openstackgerrit | Benjamin Schanzel proposed zuul/nodepool master: Kubernetes Driver: Allow cpu/mem resource limits https://review.opendev.org/736984 | 13:00 |
| ttx | AJaeger: post is supercedent, which does merge but only for same project-ref | 13:01 |
| ttx | here it's two project-refs (one updating stable/train and the other on master for example) | 13:01 |
| AJaeger | I see | 13:02 |
| ttx | (or pushing two different commits in master) | 13:03 |
| *** jamesmcarthur has joined #zuul | 13:26 | |
| *** Goneri has joined #zuul | 13:36 | |
| AJaeger | ttx, sorry, coming back now to your question: AFAIK: with resources-first: We ask for a node, acquire the semaphore; with resources-first: false, first acqure the semaphore, then request the node. So, if you have 50 post jobs: With resources-first, you would start 50 nodes, and then one runs after the other - with false, it's get semaphore, get node, get semaphore, get node; Since post pipeline has | 13:44 |
| AJaeger | highest priority, I would go with "false" and see how that works | 13:44 |
| openstackgerrit | Guillaume Chauvel proposed zuul/zuul master: [DNM] TEST run zuul tutorials to test 'uuid' in 'src_dir' https://review.opendev.org/735576 | 13:50 |
| openstackgerrit | Guillaume Chauvel proposed zuul/zuul master: [DNM] TEST run zuul tutorials to test stream+callback (+ zuul-jobs change) https://review.opendev.org/735477 | 13:50 |
| openstackgerrit | Guillaume Chauvel proposed zuul/zuul master: [DNM] TEST run zuul tutorials to test stream+callback and multiple nodes https://review.opendev.org/735483 | 13:50 |
| *** rpittau is now known as rpittau|brb | 13:54 | |
| fungi | mhu: doesn't look like anyone else has answered you here yet, your impressions are correct. promote is only really relevant for reordering change queues in dependent pipelines (say you have 10 items queued and one of the most recently enqueued items is urgent but you're concerned there's a risk one of the changes ahead of it may fail, perhaps it's a fix for a nondeterministic failure case which you're afraid a | 13:57 |
| fungi | change ahead of it may hit), then you can promote that change to the head of the queue and restart testing of all the other items behind it | 13:57 |
| mhu | fungi, thanks! Do you mind if I take this and add it to the doc? So next time I forget about it, I can just refer to the manual | 13:58 |
| ttx | AJaeger: since most of that job runtime is around setting up resources, I fear that would create a backlog. I'd rather try resources-first:true | 13:59 |
| *** rpittau|brb is now known as rpittau|afk | 14:00 | |
| fungi | mhu: please do | 14:02 |
| fungi | ttx: this is specifically for the release notes jobs which are essentially branchless? if so, maybe the ideal is for zuul to get a new pipeline manager which is like supercedent but also ignores branch differences when deciding what to dequeue | 14:03 |
| ttx | fungi: no I ws trying to serialize the github mirroring | 14:05 |
| ttx | but yes that would also solve it | 14:05 |
| fungi | oh, yep | 14:05 |
| fungi | alternatively, if the job only pushed the branch or tag which triggered it, that would avoid collissions | 14:05 |
| ttx | But I was thinking it would still fail | 14:05 |
| ttx | like the branch-agnostic-supercedent pipeline manager would kill previously-enqueued jobs and only run the latest right | 14:06 |
| ttx | but I could see how that killing could intervene after the start of the mirror for job 2 | 14:08 |
| fungi | right now supercedent enqueues up to one item behind the running item for the same project+branch, and if there is a change already queued when another is eneuqued, the newly-enqueued one replaced the one waiting in the queue | 14:08 |
| *** tosky has quit IRC | 14:08 | |
| fungi | but it only runs at most one at a time | 14:08 |
| ttx | what happens if the first job is already running? | 14:09 |
| ttx | Like you have job 1 started, and job 2 arrives | 14:09 |
| fungi | the queued item waits for all builds for the running item to finish | 14:09 |
| fungi | so it should not run two concurrently (for the same project+branch at the moment) | 14:09 |
| fungi | if we had a version of that algorithm which only took the project into account, then say you have a build running for master and one queued for a stable/stein commit which merged, then a stable/train commit merges for the same project, the stable/stein one would be dequeued without ever running | 14:10 |
| ttx | OK so a supercedent-but-at-repo-level would work... just a bit costly as you;d have to define a separate post pipeline for that | 14:11 |
| fungi | right. well, i mean, pipelines aren't *really* that costly, they're a few lines in a file | 14:11 |
| fungi | but the semaphore solution would likely also work since it's just one job you're concerned about | 14:12 |
| fungi | though i can't recall if you'd need a separate semaphore per project for that solution | 14:12 |
| ttx | fungi: but a semaphore would also work (locking a bit too much), as would a change that would limit what is actually mirrored | 14:12 |
| ttx | would be a github-semaphore, which would lock more than necessary | 14:13 |
| ttx | (there is no way apparently to create one semaphore per project) | 14:13 |
| fungi | yeah, if there was one semaphore for all runs of the same job, and the job is used by multiple repos, you'd never push more than one repo at a time | 14:13 |
| fungi | which could get fairly backlogged | 14:13 |
| *** ysandeep is now known as ysandeep|away | 14:14 | |
| ttx | maybe adding an implicit semaphore for each project would be a better approach | 14:14 |
| fungi | would also like to pick others' brains once they wake up... there are likely other solutions i'm just not thinking of | 14:15 |
| ttx | (limiting the github_mirror script to the affected branch would not work for things like tags, which would need to be pushed from all branches | 14:15 |
| ttx | so we'd still get conflicts | 14:15 |
| fungi | wouldn't you mirror tags to github only when tags get pushed to gerrit? | 14:16 |
| ttx | you want the job to be idempotent | 14:16 |
| fungi | like, if you limited to just replicating the ref which triggered the build, then you could use push rather than mirror | 14:17 |
| ttx | like if that tag does not get pushed for whatever reason, you want another future run to make it eventually consistent | 14:17 |
| fungi | though maybe you could special-case tags and push all tags to github each time a tag ref triggers the job | 14:17 |
| AJaeger | and if it fails, we would need to re-enqueue. Right now, it's self-healing with the next merge... | 14:18 |
| ttx | I'll send an email to discuss the various options | 14:18 |
| fungi | since git knows about all the local tags, the job could just `git push --tags` and then any which do not exist in github will be pushed to it | 14:18 |
| ttx | right. tags being infrequent that would mean slow reconciliation | 14:19 |
| fungi | that's true | 14:19 |
| ttx | each solution has drawbacks :) | 14:19 |
| fungi | add a daily periodic to fill in the gaps? ;) | 14:19 |
| fungi | of course your supercedent pipeline could also race/collide with periodic builds, so that's also not ideal | 14:20 |
| ttx | I think implicit repo-specific semaphores is the most elegant solution personally | 14:20 |
| ttx | but I'll describe them all in an email, as this discussion will be hard to follow | 14:21 |
| fungi | sounds good | 14:22 |
| *** hashar has quit IRC | 14:27 | |
| mnaser | has zuul by any chance changed behaviour around 'dynamically' including a role from the repo it lives in? | 14:28 |
| mnaser | https://opendev.org/vexxhost/ansible-role-base-server/src/branch/master/.zuul.yaml shows a simple job with https://opendev.org/vexxhost/ansible-role-base-server/src/branch/master/examples/playbook.yml -- no weird includes or symlinks | 14:28 |
| mnaser | whereas noonedeadpunk has been trying to do the same here https://review.opendev.org/#/c/736331/11 but without a symlink it doesn't end up working? | 14:29 |
| mnaser | aahhhhhh wait i notice it | 14:30 |
| mnaser | noonedeadpunk: we had to use 'frrouting' only and not 'ansible-role-frrouting' woops | 14:30 |
| * mnaser tries to push edit and see | 14:30 | |
| noonedeadpunk | hm | 14:36 |
| noonedeadpunk | I eventually also tried https://review.opendev.org/#/c/736331/10/meta/main.yml but it didn't work | 14:37 |
| mnaser | noonedeadpunk: i think what was missing is using 'ansible-role-frrouting', all good thoguh | 14:37 |
| ttx | fungi. corvus, AJaeger: posted http://lists.zuul-ci.org/pipermail/zuul-discuss/2020-June/001265.html | 14:38 |
| noonedeadpunk | oh, so just zuul just drops ansible-role with itself? | 14:38 |
| noonedeadpunk | that's interesting | 14:39 |
| mnaser | noonedeadpunk: yeah because, the standard naming procedure for ansible roles is 'ansible-role-XXX' where 'XXX' is the role name | 14:40 |
| mnaser | so i'm not sure if zuul or ansible drops it, but something does :) | 14:40 |
| *** chandankumar is now known as raukadah | 14:45 | |
| mordred | mnaser: I think this is an area where adding colllection support to zuul may be helpful | 14:52 |
| mnaser | ++ | 14:53 |
| *** sgw has quit IRC | 15:06 | |
| *** sgw has joined #zuul | 15:08 | |
| clarkb | mordred: corvus: I think the foundation is hoping for slide content by the end of today for the community update. Are we happy with the etherpad's contents at this point? I think allison said she could transcode to slide format when ready | 15:10 |
| openstackgerrit | Matthieu Huin proposed zuul/zuul master: CLI: add documentation on promote https://review.opendev.org/737027 | 15:13 |
| mordred | clarkb: I lost my tab - link me to it again? | 15:14 |
| clarkb | mordred https://etherpad.opendev.org/p/zuul-update-2020-06 | 15:15 |
| corvus | clarkb: i think so | 15:16 |
| mordred | corvus: what's the talking point on Improved support for managing release jobs in-tree. ? | 15:18 |
| clarkb | mordred: I think thats the thing where zuul tries to match tags to branches and lookup jobs from that branch config | 15:19 |
| fungi | yes, that ^ | 15:19 |
| clarkb | the old behavior was tags existed completely separate of branches so it didn't know what version of jobs to use when running under a tag | 15:19 |
| fungi | tag refs are now backtracked to the best match branch for identifying configuration | 15:20 |
| corvus | yep that | 15:20 |
| corvus | mordred: https://review.opendev.org/578557 this change | 15:20 |
| corvus | it's not really groundbreaking info for that audience, it's sort of filler. if we wanted to bump something in favor of something else, that would be the one to change :) | 15:21 |
| corvus | did we have a gce driver for nodepool the last time one of these happened? | 15:22 |
| *** bhavikdbavishi has joined #zuul | 15:22 | |
| corvus | (even if we did, do we want to mention it again? | 15:22 |
| corvus | we could also say something about ongoing progress with gitlab | 15:22 |
| corvus | i put those on the etherpad, feel free to drop anything | 15:23 |
| clarkb | I think we talked about gce last time. | 15:23 |
| clarkb | gitlab progress would be good | 15:23 |
| mordred | corvus: ah - awesome, thatnks | 15:26 |
| mordred | corvus: and yes, I thnk those are good additions | 15:26 |
| mordred | corvus: oh - also - have we landed avass' nodepool-builder ec2 support? | 15:27 |
| *** bhavikdbavishi has quit IRC | 15:28 | |
| *** harrymichal has quit IRC | 15:29 | |
| corvus | mordred: no that's wip https://review.opendev.org/735217 | 15:29 |
| mordred | corvus: k. that'll be good fodder for next time then | 15:30 |
| corvus | ya | 15:31 |
| *** sgw has quit IRC | 15:32 | |
| *** Goneri has quit IRC | 15:32 | |
| *** reiterative has quit IRC | 15:32 | |
| *** sgw has joined #zuul | 15:32 | |
| *** Goneri has joined #zuul | 15:32 | |
| *** reiterative has joined #zuul | 15:32 | |
| *** masterpe has quit IRC | 15:35 | |
| *** tosky has joined #zuul | 15:50 | |
| *** jamesmcarthur has quit IRC | 15:53 | |
| *** jamesmcarthur has joined #zuul | 15:55 | |
| *** jcapitao has quit IRC | 16:01 | |
| *** rlandy|ruck is now known as rlandy|ruck|brb | 16:09 | |
| dmsimard | Wow zuul v4 and v5 :) | 16:12 |
| dmsimard | v3 took so long haha | 16:12 |
| fungi | yeah, but now that we have v3 to gate zuul changes, v4 and v5 can be quick! ;) | 16:12 |
| dmsimard | \o/ | 16:12 |
| fungi | really though, zuul 3.0.0 is over two years old now | 16:13 |
| fungi | so v4 hasn't really happened terribly quickly (v5 likely will though) | 16:14 |
| *** jamesmcarthur has quit IRC | 16:20 | |
| *** bhavikdbavishi has joined #zuul | 16:32 | |
| *** rlandy|ruck|brb is now known as rlandy|ruck | 16:37 | |
| *** nils has quit IRC | 16:40 | |
| *** jamesmcarthur has joined #zuul | 16:41 | |
| *** Goneri has quit IRC | 16:43 | |
| *** hamalq has joined #zuul | 16:47 | |
| hamalq | hi am facing a problem trying to get zuul to work in https://zuul.opendev.org/t/openstack/build/8da2f74869744629a223ae03b698bb1e i get the error designate-tempest-plugin requires Python '>=3.6' but the running Python is 2.7.17 for stein designate | 16:49 |
| *** ChanServ has quit IRC | 16:59 | |
| fungi | hamalq: this is probably a topic for the #openstack-infra channel, but i can tell you that's almost certainly due to running with a version of pip which is too old | 17:02 |
| hamalq | fungi: thanks i will check on that | 17:03 |
| *** jamesmcarthur has quit IRC | 17:04 | |
| *** jpena is now known as jpena|off | 17:04 | |
| fungi | hamalq: oh, or possibly using the wrong constraints list | 17:04 |
| hamalq | fungi: because it should use https://github.com/openstack/designate-tempest-plugin/blob/0.7.0/setup.cfg 0.7.0 is there a way to specify that on zuul.yaml | 17:06 |
| hamalq | in here https://github.com/openstack/designate/blob/stable/stein/.zuul.yaml | 17:06 |
| fungi | hamalq: let's discuss in #openstack-infra, this isn't a zuul problem, it's an openstack job problem | 17:07 |
| hamalq | ok will do | 17:07 |
| *** gmann is now known as gmann_afk | 17:13 | |
| *** bhavikdbavishi has quit IRC | 17:31 | |
| *** ChanServ has joined #zuul | 17:37 | |
| *** tepper.freenode.net sets mode: +o ChanServ | 17:37 | |
| *** jamesmcarthur has joined #zuul | 17:37 | |
| *** Goneri has joined #zuul | 17:37 | |
| *** harrymichal has joined #zuul | 17:52 | |
| *** jamesmcarthur has quit IRC | 18:09 | |
| *** jamesmcarthur has joined #zuul | 18:10 | |
| openstackgerrit | James E. Blair proposed zuul/zuul-jobs master: Fix and test multiarch docker builds in a release pipeline https://review.opendev.org/737059 | 18:27 |
| corvus | avass: ^ using your new test :) | 18:27 |
| openstackgerrit | Sergii Golovatiuk proposed zuul/zuul-jobs master: Install source Python for supported OSes https://review.opendev.org/737060 | 18:27 |
| *** holser has joined #zuul | 18:27 | |
| AJaeger | corvus: that one breaks VEXXHOST CI - mnaser could you update your CI for 737059, please? | 18:30 |
| *** jamesmcarthur has quit IRC | 18:40 | |
| *** jamesmcarthur has joined #zuul | 18:41 | |
| *** armstrongs has joined #zuul | 18:43 | |
| corvus | AJaeger, mnaser: i'm a little confused; vexxhost reported okay on the parent | 18:44 |
| corvus | AJaeger, mnaser: i don't think vexxhost should be reading the zuul-tests.d -- that's intended for opendev | 18:45 |
| holser | AJaeger let me fix python-devel patch really quick | 18:47 |
| armstrongs | Quick question I have a common config project on multiple tenants. This however has a zuul file with dependencies on other projects that aren't included in those projects so it gives config issues. Is there any way to avoid these config issues without adding all projects to every tenant? | 18:48 |
| clarkb | armstrongs: what opendev has done is try and get our base config to avoid external deps as much as possible. Because ya, zuul wants to resolve those deps if they exist | 18:49 |
| *** armstrongs has quit IRC | 18:53 | |
| corvus | did armstrongs leave? | 18:54 |
| clarkb | looks lke it | 18:54 |
| corvus | if they come back; i would add to that that the extra-config-paths option may be useful in that case | 18:55 |
| corvus | https://zuul-ci.org/docs/zuul/reference/tenants.html#attr-tenant.untrusted-projects.%3Cproject%3E.extra-config-paths | 18:56 |
| corvus | you might be able to separate config into different files/dirs for use in different tenants. so if you need some extra stuff in, say, a main tenant that manages that config-project, and then only need a subset in the rest of the repos that use it, you could put the extra stuff in its own dir | 18:57 |
| *** jamesmcarthur_ has joined #zuul | 18:57 | |
| corvus | may or may not be appropriate depending on circumstances | 18:57 |
| *** jamesmca_ has joined #zuul | 18:58 | |
| *** jamesmc__ has joined #zuul | 18:59 | |
| *** jamesmcarthur has quit IRC | 19:00 | |
| *** jamesmc__ has quit IRC | 19:00 | |
| *** jamesmcarthur_ has quit IRC | 19:02 | |
| *** jamesmca_ has quit IRC | 19:03 | |
| *** armstrongs has joined #zuul | 19:04 | |
| armstrongs | Sorry got disconnected there | 19:05 |
| clarkb | armstrongs: let me get you a link to the logs of what corvus said after you left | 19:06 |
| clarkb | armstrongs: http://eavesdrop.openstack.org/irclogs/%23zuul/%23zuul.2020-06-19.log.html#t2020-06-19T18:55:56 | 19:06 |
| armstrongs | Yeah so I could use the extra config option on the base common tenant and then it wont be read on the other tenants. That would work thank you. | 19:07 |
| avass | corvus: cool, if there's no rush I'll take a look at it tomorrow. :) | 19:07 |
| avass | otherwise you can just bypass that for now | 19:07 |
| *** armstrongs has quit IRC | 19:13 | |
| *** jamesmcarthur has joined #zuul | 19:14 | |
| *** jamesmcarthur has quit IRC | 19:20 | |
| *** jamesmcarthur has joined #zuul | 19:22 | |
| mnaser | corvus, AJaeger: ok, i think we may have made the mistake of actually loading zuul-tests.d | 19:28 |
| mnaser | i think the idea was 'we can run more downstream tests' initially | 19:29 |
| *** jamesmcarthur has quit IRC | 19:32 | |
| openstackgerrit | Andreas Jaeger proposed zuul/zuul-jobs master: Fix zuul-jobs-test-ensure-python-pyenv files https://review.opendev.org/737068 | 19:32 |
| *** jamesmcarthur has joined #zuul | 19:32 | |
| AJaeger | mnaser: that would explain it. I agree with corvus, please update your CI to ignore zuul-tests.d | 19:34 |
| mnaser | yep, already done, it's shipping :) | 19:34 |
| AJaeger | zuul-jobs maintainers, please review 737068, I noticed that we did not run tests when we should | 19:34 |
| AJaeger | mnaser: Great! | 19:34 |
| AJaeger | mnaser: not working yet, see https://review.opendev.org/#/c/737059/ - I just rechecked and get the same failure | 19:36 |
| mnaser | AJaeger: yes, sorry, should have clarified, it's in flight | 19:36 |
| mnaser | our cd takes 5-10 minutes to pick it up and reload | 19:36 |
| mnaser | we have to restart all of our zuul for this, i never got around finishing the inotify patch :( | 19:37 |
| mnaser | so no smart-reconfigure | 19:37 |
| AJaeger | mnaser: ah, please recheck when done to double check that it works | 19:38 |
| AJaeger | mnaser: also, why are there no logfiles from your CI in https://review.opendev.org/#/c/737060/1 - https://zuul.vexxhost.dev/t/opendev/build/1bbfebba5fb4498aa5cdfb9116260d8e shows the Zuul UI and a "404" ;( | 19:39 |
| mnaser | AJaeger: because the tenant is mid reload right now | 19:39 |
| mnaser | :P | 19:39 |
| mnaser | zuul is starting up | 19:39 |
| mnaser | so everything 404s https://zuul.vexxhost.dev/t/opendev | 19:39 |
| AJaeger | mnaser: Got it, thanks | 19:39 |
| * AJaeger wishes everybody a great weekend and waves good night | 19:40 | |
| mnaser | AJaeger: good night and have a good weekend | 19:40 |
| corvus | huh, that failed the original version of avass's test... i wonder if something external changed | 19:40 |
| corvus | AJaeger: good night! | 19:41 |
| AJaeger | thanks | 19:41 |
| *** yolanda has quit IRC | 19:41 | |
| mnaser | corvus: 737059 is fixed and running at https://zuul.vexxhost.dev/t/opendev/status -- thank you for pinging me about that | 19:43 |
| corvus | mnaser: woot, thanks :) | 19:43 |
| *** sanjayu_ has quit IRC | 19:53 | |
| *** saneax has joined #zuul | 19:54 | |
| *** jamesmcarthur_ has joined #zuul | 20:12 | |
| *** jamesmcarthur has quit IRC | 20:15 | |
| *** gmann_afk is now known as gmann | 20:15 | |
| *** jamesmcarthur_ has quit IRC | 20:26 | |
| *** jamesmcarthur has joined #zuul | 20:27 | |
| *** vishalmanchanda has quit IRC | 20:40 | |
| *** jamesmcarthur has quit IRC | 20:48 | |
| *** hashar has joined #zuul | 20:54 | |
| *** rfolco has quit IRC | 20:54 | |
| *** y2kenny has joined #zuul | 20:56 | |
| y2kenny | does the console stream to the web-ui come from the executor? | 21:06 |
| fungi | y2kenny: indirectly, yes | 21:08 |
| clarkb | y2kenny: it comes from the zuul-fingergw which then talks to the executors iirc | 21:08 |
| y2kenny | ok. The stream stopped working all of a sudden and I am wondering where to start doing the debugging | 21:08 |
| fungi | y2kenny: we've seen that if an out-of-memory event on the executor causes the kernel to sacrifice the log streaming process | 21:09 |
| fungi | restarting the executor daemon generally fixes that | 21:11 |
| *** sgw1 has quit IRC | 21:11 | |
| y2kenny | fungi: I tried that but didn't seem to work. I just restarted the web UI server and see if that fixes it... | 21:11 |
| corvus | er actually the fingergw is not involved in the web console stream; zuul-web talks to the executor directly | 21:12 |
| corvus | so it's: browser -> zuul-web -> zuul-executor | 21:13 |
| clarkb | corvus: TIL | 21:13 |
| clarkb | for some reason I thought it went throught the fingergw too | 21:13 |
| clarkb | since fingergw knows which executor to talk to but I guess web knows too? | 21:13 |
| corvus | the fingergw is just for "finger @" | 21:13 |
| corvus | yep | 21:13 |
| fungi | yeah, zuul-fingergw is in case someone wants to use finger protocol to stream the log without access to the finger port on the executor (or having to know which executor) right? | 21:13 |
| corvus | they're basically doing the same thing for both protocols | 21:13 |
| clarkb | got it | 21:13 |
| corvus | y2kenny: as an admin, you can also telnet or finger to the zuul-executor finger port to debug this | 21:15 |
| corvus | i think something like this should work: echo SHA | nc executor.example.com -p 7900 | 21:16 |
| corvus | that would get you a console stream talking directly to the executor. whether that works or fails should help narrow down what's broken. | 21:17 |
| y2kenny | corvus: I see... are there more documentation around this? is this along the line of what's describe here: https://zuul-ci.org/docs/zuul/howtos/troubleshooting.html | 21:17 |
| corvus | echo 21bee08ebf1f4c90881040f4a2ee91d7 | nc localhost 7900 | 21:18 |
| corvus | sorry i got the cmdline wrong; that ^ is a working example | 21:18 |
| y2kenny | SHA being the job id? | 21:18 |
| corvus | y2kenny: yes the build id | 21:18 |
| y2kenny | ok | 21:18 |
| corvus | y2kenny: and yeah, i don't think we documented this troubleshooting technique | 21:18 |
| corvus | avass: wow, docker really did just update the registry:2 image and 'htpasswd' is no longer in it | 21:21 |
| corvus | avass: however, we have a bunch of code to set up the registry for the intermediate registry tests and it doesn't rely on that, so i'm going to modify your change to use that approach | 21:22 |
| openstackgerrit | James E. Blair proposed zuul/zuul-jobs master: Add tests for upload-docker-image https://review.opendev.org/735402 | 21:29 |
| openstackgerrit | James E. Blair proposed zuul/zuul-jobs master: Fix and test multiarch docker builds in a release pipeline https://review.opendev.org/737059 | 21:29 |
| *** sgw1 has joined #zuul | 21:31 | |
| *** saneax has quit IRC | 21:47 | |
| *** hashar has quit IRC | 21:54 | |
| *** harrymichal has quit IRC | 21:56 | |
| *** weshay|ruck is now known as weshay_pto | 22:00 | |
| *** jamesmcarthur has joined #zuul | 22:48 | |
| *** rlandy|ruck has quit IRC | 22:59 | |
| *** jamesmcarthur has quit IRC | 23:07 | |
| *** jamesmcarthur has joined #zuul | 23:09 | |
| *** hamalq has quit IRC | 23:11 | |
| *** tosky has quit IRC | 23:45 | |
| mnaser | corvus: is there someone playing around with adding patternfly 4 to the zuul ui? | 23:49 |
| mnaser | i was looking to see if we can get it into horizon but figured the zuul web ui might be a good hit to get to know what im dealing with first | 23:49 |
| corvus | mnaser: i think so, 1 sec | 23:53 |
| corvus | mnaser: https://review.opendev.org/736225 | 23:53 |
| mnaser | oh awesome | 23:54 |
| corvus | that's a good chunk of work done | 23:54 |
| corvus | i'm having some problems navigating around a bit | 23:56 |
| corvus | in the preview site | 23:56 |
| corvus | i'm not sure if that's something that would happen for a real deployment, or something specific to the preview | 23:56 |
| corvus | all the liks go to the 'undefined' tenant | 23:56 |
| corvus | links | 23:57 |
| corvus | mnaser: anyway, there's enough there to see what it looks like, and enough left to do to warrant playing around with it :) | 23:57 |
| mnaser | corvus: yeah i noticed those too, but looks like some awesome progress, i'll dive int if there's any work left later as i see some good effort by felix starting it :) | 23:58 |
| corvus | mnaser: i really like PF's tables, but some of the more complex structures like trees, etc, are lacking | 23:59 |
| corvus | mnaser: but i think horizon is pretty table heavy and not so much into trees, so probably a good fit :) | 23:59 |
| mnaser | corvus: yeah i agree, i dont think there is a good replacement to our need for the trees, but i think we should certainly look at moving at using the tables for buildsets/nodes/etc | 23:59 |
| mnaser | but yes, agreed on the horizon thing too | 23:59 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!