| *** rlandy has quit IRC | 00:19 | |
| *** jamesmcarthur has joined #zuul | 00:43 | |
| *** zxiiro has quit IRC | 00:51 | |
| mordred | corvus: I'll put up a patch - but tl;dr - we're pulling from buildset registry onto an amd64 node, then pushing from there - so we're not pushing the multipart to dockerhub. we need to add logic to upload-docker-image to run the push with buildx | 00:59 |
|---|---|---|
| *** jamesmcarthur has quit IRC | 01:10 | |
| *** jamesmcarthur has joined #zuul | 01:12 | |
| openstackgerrit | Merged zuul/zuul-jobs master: ensure-tox: use venv to install https://review.opendev.org/725737 | 01:28 |
| corvus | mordred: oh wow. that's obvious in retrospect. :/ | 01:34 |
| *** jamesmcarthur has quit IRC | 01:36 | |
| *** jamesmcarthur has joined #zuul | 01:48 | |
| *** jamesmcarthur has quit IRC | 01:48 | |
| *** jamesmcarthur has joined #zuul | 01:49 | |
| *** swest has quit IRC | 01:52 | |
| *** jamesmcarthur has quit IRC | 01:53 | |
| mordred | corvus: yeah | 01:55 |
| *** swest has joined #zuul | 02:07 | |
| *** jamesmcarthur has joined #zuul | 02:10 | |
| *** jamesmcarthur has quit IRC | 02:20 | |
| *** jamesmcarthur has joined #zuul | 02:33 | |
| openstackgerrit | Ian Wienand proposed zuul/nodepool master: Use fqdn for builder hostname info https://review.opendev.org/726032 | 02:47 |
| openstackgerrit | Monty Taylor proposed zuul/zuul-jobs master: Upload images to dockerhub with buildx when using buildx https://review.opendev.org/726033 | 02:50 |
| mordred | corvus, ianw: ^^ | 02:50 |
| *** bhavikdbavishi has joined #zuul | 02:58 | |
| *** bhavikdbavishi1 has joined #zuul | 03:01 | |
| *** bhavikdbavishi has quit IRC | 03:03 | |
| *** bhavikdbavishi1 is now known as bhavikdbavishi | 03:03 | |
| *** jamesmcarthur has quit IRC | 03:37 | |
| *** jamesmcarthur has joined #zuul | 03:40 | |
| *** evrardjp has quit IRC | 04:36 | |
| *** evrardjp has joined #zuul | 04:36 | |
| *** bhavikdbavishi has quit IRC | 04:48 | |
| *** bhavikdbavishi has joined #zuul | 05:00 | |
| *** sgw has quit IRC | 05:03 | |
| *** reiterative has quit IRC | 05:36 | |
| *** reiterative has joined #zuul | 05:36 | |
| *** jamesmcarthur has quit IRC | 05:53 | |
| *** jamesmcarthur has joined #zuul | 05:59 | |
| *** jamesmcarthur has quit IRC | 06:03 | |
| *** jamesmcarthur has joined #zuul | 06:10 | |
| *** jamesmcarthur has quit IRC | 06:27 | |
| openstackgerrit | Simon Westphahl proposed zuul/nodepool master: Expose image build requests in web UI and cli https://review.opendev.org/725810 | 06:41 |
| *** saneax has joined #zuul | 06:53 | |
| *** jcapitao has joined #zuul | 07:11 | |
| *** bhavikdbavishi has quit IRC | 07:23 | |
| *** tosky has joined #zuul | 07:31 | |
| avass | zuul-jobs-maint: should we bump minimum ansible version to 2.6? | 07:38 |
| avass | since 2.6 is the lowest version supported by zuul now | 07:38 |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: Bumb ansible to version 2.6 https://review.opendev.org/726054 | 07:40 |
| *** rpittau|afk is now known as rpittau | 07:47 | |
| avass | I'm not sure if that ^ affects anything other than nested ansible plays? | 07:47 |
| avass | linting? | 07:47 |
| *** jpena|off is now known as jpena | 07:55 | |
| *** mnasiadka has quit IRC | 07:56 | |
| *** adam_g has quit IRC | 07:56 | |
| *** panda has quit IRC | 07:59 | |
| *** mnasiadka has joined #zuul | 08:01 | |
| *** adam_g has joined #zuul | 08:01 | |
| *** panda has joined #zuul | 08:02 | |
| *** bhavikdbavishi has joined #zuul | 08:04 | |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: Add upload-artifactory role https://review.opendev.org/725678 | 08:45 |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: Add upload-artifactory role https://review.opendev.org/725678 | 08:46 |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: Add upload-artifactory role https://review.opendev.org/725678 | 08:53 |
| *** tumble has joined #zuul | 09:10 | |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: Add upload-artifactory role https://review.opendev.org/725678 | 09:10 |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: Bumb ansible to version 2.6 https://review.opendev.org/726054 | 09:16 |
| avass | AJaeger: I'm think about setting up test environments like the artifactory role ^^ for our 'untestable' roles | 09:38 |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: Add upload-artifactory role https://review.opendev.org/725678 | 09:38 |
| *** masterpe has quit IRC | 09:39 | |
| avass | AJaeger: so whenever we needatrusted context, we mock away zuul by nesting ansible, passing an extra vars file with zuul: content or anything else would have come from the job, and using the fake zuul action modules we have in the repo | 09:40 |
| *** irclogbot_0 has quit IRC | 09:41 | |
| *** irclogbot_0 has joined #zuul | 09:43 | |
| *** sshnaidm|afk is now known as sshnaidm | 09:44 | |
| paladox | corvus hi, it seems https://ci.gerritcodereview.com/t/gerrit/build/89e7c54f6f374b4b9dd26f22f4f9ad9e is saying "gerrit_root" is undefined. | 09:47 |
| AJaeger | avass: Puuh, that's an interesting idea - best share with rest of Zuul maintainers later. | 09:49 |
| avass | sure :) | 09:51 |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: Add upload-artifactory role https://review.opendev.org/725678 | 09:54 |
| *** masterpe has joined #zuul | 09:56 | |
| *** bhavikdbavishi has quit IRC | 10:05 | |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: Add upload-artifactory role https://review.opendev.org/725678 | 10:07 |
| *** rpittau is now known as rpittau|bbl | 10:10 | |
| *** bhavikdbavishi has joined #zuul | 10:10 | |
| noonedeadpunk | o/ | 10:20 |
| noonedeadpunk | Can you give me a hint how to use base jobs for zuul deployment? https://zuul-ci.org/docs/zuul-base-jobs/ | 10:21 |
| noonedeadpunk | Like should I just add repo to untrusted projects or to project-config? | 10:21 |
| avass | noonedeadpunk: base jobs (jobs without a parent) needs to be in a config project | 10:25 |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: Add upload-artifactory role https://review.opendev.org/725678 | 10:27 |
| noonedeadpunk | avass: ok, and zuul/zuul-jobs are supposed to be under project-config as well? Or I add them as untrusted project and reference them from base jobs? | 10:29 |
| avass | noonedeadpunk: it's better to let zuul-jobs be an untrusted-project :) | 10:29 |
| avass | noonedeadpunk: But zuul-base-jobs is supposed to be more of an example of how to set things up, I don't think it's a good idea to use that directly | 10:30 |
| avass | noonedeadpunk: at least fork it :) | 10:30 |
| noonedeadpunk | ok, so let's say I have http://paste.openstack.org/show/793260/ in my /etc/zuul/main.yaml | 10:30 |
| noonedeadpunk | so I should reference it from base-jobs like "roles: [{zuul: opendev/zuul/zuul-jobs}]"? | 10:31 |
| avass | yeah, but you can refernce it wwith just 'zuul/zuul-jobs' | 10:32 |
| noonedeadpunk | oh, ok | 10:32 |
| noonedeadpunk | Just wasn't sure how to reference source and tenant name on project-config | 10:33 |
| noonedeadpunk | trying at least to run noop thing) | 10:34 |
| avass | If you have a name collision I think you can to <canonical_name>/zuul/zuul-jobs | 10:35 |
| avass | do* | 10:35 |
| avass | so opendev.org/zuul/zuul-jobs in this case | 10:35 |
| avass | and you can check how to access it in the web dashboard: https://zuul.opendev.org/t/zuul/project/opendev.org/zuul/zuul-jobs | 10:36 |
| avass | in your tenant, just using opendev as an example :) | 10:36 |
| noonedeadpunk | opendev perfectly suits at the moment. But for sure now have no conflict. | 10:37 |
| noonedeadpunk | Btw, maybe you can fastly say where to look for noop template? | 10:37 |
| noonedeadpunk | was looking for in https://opendev.org/openstack/project-config | 10:37 |
| noonedeadpunk | but either don't see it or it's not htere | 10:38 |
| avass | you mean the '- noop' job? | 10:38 |
| noonedeadpunk | yeah | 10:38 |
| avass | that's built-in :) | 10:38 |
| noonedeadpunk | so this is built in? https://opendev.org/openstack/project-config/src/branch/master/zuul.d/projects.yaml#L76-L79 | 10:39 |
| noonedeadpunk | cool | 10:39 |
| avass | okay, no idea what that is | 10:39 |
| noonedeadpunk | I mean I can just reference templates: - noop-jobs? | 10:40 |
| noonedeadpunk | or what's the way to use noop? | 10:40 |
| avass | noonedeadpunk: but it's just a noop https://opendev.org/openstack/openstack-zuul-jobs/src/branch/master/zuul.d/project-templates.yaml#L18 | 10:40 |
| noonedeadpunk | ah | 10:41 |
| noonedeadpunk | yeah | 10:41 |
| avass | noonedeadpunk: the '- noop' in check and gate in that link is built in, opendev has just set up a template to make sure something runs in zuul so things are merged | 10:41 |
| avass | if you need to search for code in opendev you can use: codesearch.openstack.org/ :) | 10:42 |
| noonedeadpunk | yeah, thanks) use it heavily but searching jsut "noop" is not so helpful :p | 10:43 |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: Add upload-artifactory role https://review.opendev.org/725678 | 10:46 |
| openstackgerrit | Tobias Henkel proposed zuul/zuul master: Revert "Revert "Tune automatic garbage collection of git repos"" https://review.opendev.org/723800 | 10:49 |
| tobiash | corvus, AJaeger: now with a test case that reproduces the issue without the fix ^ | 10:49 |
| noonedeadpunk | avass: thanks for your time and patience!) | 10:55 |
| *** jcapitao is now known as jcapitao_lunch | 10:57 | |
| avass | noonedeadpunk: no problem! | 10:59 |
| openstackgerrit | Tobias Henkel proposed zuul/zuul master: d0;95;0cRevert "Revert "Tune automatic garbage collection of git repos"" https://review.opendev.org/723800 | 11:05 |
| openstackgerrit | Tobias Henkel proposed zuul/zuul master: Revert "Revert "Tune automatic garbage collection of git repos"" https://review.opendev.org/723800 | 11:07 |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: Add upload-artifactory role https://review.opendev.org/725678 | 11:18 |
| zbr | do we have any work done to publish zuul_return inside a galaxy collection? | 11:26 |
| AJaeger | tobiash: tahnks | 11:26 |
| avass | zbr: zuul_return doesn't really have a use-case outside of zuul, and zuul doesn't support collections yet, so I don't think so | 11:28 |
| zbr | avass: i need to module in order to lint ansible code from other repositories | 11:28 |
| zbr | so basically I need to publish it inside a collection, and reference it from there. | 11:29 |
| avass | zbr: zuul-jobs uses a fake module in zuul-jobs/tests/fake-ansible | 11:29 |
| zbr | avass: stil, I would need to publish fake module as a collection in order to consume it | 11:30 |
| zbr | i do not find it cool to copy/paste that part to lots of repositories | 11:30 |
| zbr | it does not scale well, at all | 11:30 |
| zbr | for start it can be fake, but longer term it could even be the real thing | 11:30 |
| avass | I think there was an idea to break the modules out to their own repositories whenever we do support collections though | 11:31 |
| zbr | i think it may be possible to publish the module directly w/o much fuss. | 11:32 |
| *** jpena is now known as jpena|lunch | 11:32 | |
| avass | AJaeger: so, I know we only support ansible2.6, but 2.7 adds the 'body' parameter to the uri module, and without it I have to slurp the file and then template the content into the 'body' parameter | 11:35 |
| avass | AJaeger: would it be alright in this case to only support ansible 2.7 for that role and put a notice about in the readme? | 11:36 |
| masterpe | /msg fbo | 11:39 |
| avass | AJaeger: 2.7 adds the 'src' parameter, sorry :) | 11:41 |
| AJaeger | avass: I think that would be ok for this specific role - if it comes with a test for ansible < 2.7. But that needs a bit larger review of the team. | 11:42 |
| avass | AJaeger: you mean a runtime test to detect if someone is trying to use it with an unsupported ansible version? | 11:44 |
| AJaeger | avass: ansible 2.6 is also end-of-life, so we could slowly migrate to 2.7. I think there was a discussion in the past when to do it but don't remember details anymore. Worth checking as well | 11:44 |
| AJaeger | avass: yes | 11:44 |
| avass | AJaeger: sure | 11:44 |
| AJaeger | like we add a check for Debian-only roles so that the role fails on Fedora (etc) | 11:44 |
| avass | Everything else seems to work otherwise, except for the fake ansible modules | 11:44 |
| AJaeger | best discuss this all with corvus later | 11:47 |
| zbr | avass: my goal is to assure that anyone can run "ansible-lint" on a repo like zuul-jobs w/o having to do anything to make it work | 11:57 |
| avass | zbr: yeah I got it | 11:59 |
| noonedeadpunk | btw, should I be bothered by "Unable to get change for <GerritTriggerEvent ref-replicated"? | 12:01 |
| noonedeadpunk | Like I see warning but don't see anywhere it's used | 12:01 |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: Add upload-artifactory role https://review.opendev.org/725678 | 12:01 |
| noonedeadpunk | except https://opendev.org/zuul/zuul/src/branch/master/zuul/driver/gerrit/gerritconnection.py#L213-L215 | 12:02 |
| avass | zbr: I think zuul-jobs used to install zuul while testing, but that's a bit much | 12:02 |
| *** rlandy has joined #zuul | 12:02 | |
| AJaeger | we removed zuul install from zuul-jobs | 12:06 |
| avass | yeah | 12:06 |
| Shrews | avass: fwiw, uri before 2.7 didn't support uploading any binary data so you probably are better off not supporting 2.6 or earlier. | 12:08 |
| avass | Shrews: just did that :) | 12:09 |
| avass | but thanks! | 12:09 |
| *** jcapitao_lunch is now known as jcapitao | 12:10 | |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: Add upload-artifactory role https://review.opendev.org/725678 | 12:11 |
| *** rpittau|bbl is now known as rpittau | 12:11 | |
| *** hashar has joined #zuul | 12:24 | |
| AJaeger | corvus: could you check https://review.opendev.org/#/c/726054/2 - moving zuul-jobs to 2.6, please? | 12:27 |
| *** Goneri has joined #zuul | 12:37 | |
| *** jpena|lunch is now known as jpena | 12:38 | |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: Add upload-artifactory role https://review.opendev.org/725678 | 12:52 |
| avass | it works! | 13:02 |
| AJaeger | congrats, avass ! | 13:09 |
| openstackgerrit | Sorin Sbarnea (zbr) proposed zuul/zuul-jobs master: Made sequence indent consistent https://review.opendev.org/725538 | 13:15 |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: Add upload-artifactory role https://review.opendev.org/725678 | 13:17 |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: Add upload-artifactory role https://review.opendev.org/725678 | 13:22 |
| openstackgerrit | Monty Taylor proposed zuul/zuul-jobs master: Upload images to dockerhub with buildx when using buildx https://review.opendev.org/726033 | 13:26 |
| *** bhavikdbavishi has quit IRC | 13:27 | |
| mordred | avass: ^^ | 13:29 |
| mordred | zbr: yes - I do believe we would like to put the modules we have into collections and publish them - but it's going to be a bit before we can do that because zuul-jobs still supports ansible versions that do not support collections | 13:30 |
| avass | mordred: docker buildx build, not docker buildx push? | 13:32 |
| mordred | avass: yeah - see the --push at the end | 13:32 |
| mordred | it should be a noop, because the image will have already been built | 13:32 |
| mordred | avass: and yes, this is exceedingly strange | 13:33 |
| avass | mordred: I don't see a --push at the end :) | 13:34 |
| mordred | avass: oh - then that's a problem! | 13:37 |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: Add upload-artifactory role https://review.opendev.org/725678 | 13:42 |
| *** jhesketh has quit IRC | 13:43 | |
| *** sgw has joined #zuul | 13:43 | |
| openstackgerrit | Monty Taylor proposed zuul/zuul-jobs master: Upload images to dockerhub with buildx when using buildx https://review.opendev.org/726033 | 13:46 |
| mordred | avass: you were very right | 13:46 |
| *** jhesketh has joined #zuul | 13:47 | |
| avass | mordred: ++ | 13:48 |
| corvus | avass, AJaeger: z-j version bump seems legit; i don't think we need to wait any longer? | 14:02 |
| corvus | mordred: ^? | 14:02 |
| *** mhayden has joined #zuul | 14:03 | |
| mordred | corvus: I agree | 14:03 |
| mordred | corvus: I mildly worry that we | 14:03 |
| mordred | nope - nevermind | 14:03 |
| corvus | now i'm worried ;) | 14:04 |
| corvus | but i'll +3 anyway :) | 14:04 |
| avass | I'm just not sure what the ansible cap in test-requirements.txt does | 14:04 |
| avass | except for linting if there's a new module | 14:05 |
| avass | (I guess) | 14:05 |
| avass | mordred, corvus: what do you think about doing something like this for testing our roles that requires a trusted context: https://review.opendev.org/#/c/725678/ | 14:06 |
| avass | in short: run ansible nested on the remote with extra vars to mock zuul variables, and use the fakse ansible modules in zuul-jobs | 14:07 |
| avass | it's not perfect, but it tests more than we've got today :) | 14:08 |
| * mhayden waves at mordred | 14:08 | |
| *** y2kenny has joined #zuul | 14:09 | |
| corvus | avass: and you're using a nested ansible because of delegate_to: localhost? | 14:10 |
| avass | corvus: yeah | 14:10 |
| avass | corvus: it's either toggling that with a variable, but that wouldn't work for other roles that uses lookups | 14:11 |
| corvus | avass: another approach would be to do what we did in roles like pull-from-intermediate-registry; that also runs entirely on localhost, but it doesn't use any delegate_to tasks. we just put it an a play that applies to localhost. that lets us run the role on a remote node in test without using a nested ansible. | 14:14 |
| avass | corvus: you dynamically add a host that could either be the remote or localhost I guess? | 14:15 |
| y2kenny | Does merger keep its own copy of git repositories or does it only do git operations? (My reading of the docs and the helm chart use of k8s Deployment suggest that it only do git ops but I just want to double confirm.) | 14:15 |
| y2kenny | but then I am not sure what the merger.git_dir does... | 14:16 |
| avass | corvus: oh, or the playbook is just remote vs localhost I guess that could work | 14:16 |
| corvus | avass: no, it's just a normal host for the job. that particular job is complicated and has several hosts, one of them is called 'executor', and we run the pull-from role on that host. a simpler test like this which doesn't need multiple hosts could just re-use a single remote host probably. | 14:16 |
| corvus | avass: yeah that | 14:17 |
| avass | corvus: I guess that could work, you would always need a separate play just for that role though | 14:17 |
| corvus | avass: so in a "real" job, a playbook looks like "hosts: localhost, roles: foobar"; in the test job, the playbook looks like: "hosts: controller, roles: foobar" | 14:18 |
| corvus | avass: yes | 14:18 |
| avass | corvus: I'll see if I can update it to do that instead, shouldn't be hard | 14:18 |
| corvus | avass: it's a trade-off | 14:18 |
| corvus | avass: the current approach may be simpler to use, and maybe it's a better fit since this is more like the "fetch-javascript" "fetch-tox" etc roles... | 14:19 |
| corvus | avass: so folks don't have to think about adding a new play to a playbook | 14:19 |
| avass | corvus: yeah | 14:19 |
| corvus | avass: but it's harder to test, and all the interesting bits are going to be in the nested ansible | 14:20 |
| corvus | avass: maybe your idea of switching the delegate_to could be a good compromise | 14:20 |
| * mordred waves at mhayden ! | 14:21 | |
| mordred | corvus: https://review.opendev.org/#/c/726033/ when you get a sec | 14:21 |
| corvus | avass: maybe just change those to "delegate_to: {{ _undocumented_test_variable | default('localhost') }}" | 14:21 |
| avass | corvus: for this case at least, other roles that relies on lookups can't do that | 14:21 |
| openstackgerrit | Merged zuul/zuul-jobs master: Bumb ansible to version 2.6 https://review.opendev.org/726054 | 14:21 |
| mhayden | mordred: thanks for the email reply -- i think i mostly understand nodepool but i'm still working through some of the zuul docs | 14:22 |
| corvus | avass: true. but maybe one size doesn't fit all :) | 14:22 |
| avass | corvus: true :) | 14:22 |
| tobiash | mordred: commented on https://review.opendev.org/726033 | 14:23 |
| mordred | mhayden: the zuul quickstart might be a good place for you to get your feet wet: https://zuul-ci.org/docs/zuul/tutorials/quick-start.html | 14:23 |
| mordred | tobiash: thank you | 14:23 |
| openstackgerrit | Monty Taylor proposed zuul/zuul-jobs master: Upload images to dockerhub with buildx when using buildx https://review.opendev.org/726033 | 14:24 |
| corvus | tristanC, mhu: do you happen to know when fbo|off will be back around? | 14:25 |
| tristanC | corvus: i'm not sure, perhaps i can help? | 14:28 |
| *** tosky_ has joined #zuul | 14:29 | |
| mordred | tristanC: also - you should meet mhayden - you might want to have some internal-RH-type discussions | 14:30 |
| tristanC | mhayden: hello o/ | 14:31 |
| mhayden | howdy 🤠| 14:31 |
| *** tosky has quit IRC | 14:32 | |
| corvus | tristanC: ah, i was wondering if there's any more work planned on the gitlab driver.... masterpe was asking about it on the mailing list. it's unfinished, but it's been several months since any new work happened. i worry that we're giving people the wrong idea by having it in the tree. :( | 14:33 |
| corvus | tristanC: we didn't add any docs, but people seem to have found it anyway. maybe we should add some docs that say "this isn't finished; here's what works and what still needs to be done"? | 14:35 |
| masterpe | I think it is fine when it is not complete yet but at least be clear in the documentation what needs to be done. Also I'm happy to help. | 14:36 |
| tristanC | it seems like the next step for gitlab is documented in https://storyboard.openstack.org/#!/story/2006632 | 14:36 |
| corvus | tristanC: ah thanks :) | 14:38 |
| tristanC | well, that not much, but it seems like the linked etherpad contains documentation about the missing approval report | 14:39 |
| corvus | masterpe, tristanC: i just replied to the list with a message that hopefully makes it clear to anyone else that gitlab work is still in progress, but also has some suggested next steps for masterpe | 14:41 |
| openstackgerrit | Merged zuul/zuul-jobs master: Made sequence indent consistent https://review.opendev.org/725538 | 14:42 |
| corvus | masterpe, tristanC: it looks like we're really close :) | 14:42 |
| masterpe | I saw thanks for the e-mail. | 14:42 |
| *** panda is now known as panda|pto | 14:42 | |
| mordred | avass: feel like re-reviewing https://review.opendev.org/#/c/726033/ - tobiash found a thing | 14:44 |
| avass | mordred: done | 14:46 |
| mordred | avass: thank you! | 14:46 |
| *** bhavikdbavishi has joined #zuul | 14:48 | |
| fungi | mhayden: wow, welcome!!! long time no see | 14:49 |
| mhayden | haha indeed! | 14:49 |
| corvus | mordred: do we need any changes to promote? | 14:49 |
| corvus | mordred: i guess we're about to find out? :) | 14:49 |
| mordred | corvus: it's a great question! | 14:50 |
| mordred | corvus: I really hope that skopeo copies the whole thing and not just the platform-specific sub-image | 14:50 |
| mordred | corvus: or - it just retags remotely doens't it? | 14:51 |
| corvus | mordred: yeah, we fetch something and push it back | 14:51 |
| corvus | so hopefully we're just fetching the top-level manifest and pushing it back and everything works | 14:51 |
| *** bhavikdbavishi1 has joined #zuul | 14:52 | |
| mordred | corvus: yeah - I'd imagine the rest api isn't going to do that dereferencing for us | 14:52 |
| *** bhavikdbavishi has quit IRC | 14:52 | |
| *** bhavikdbavishi1 is now known as bhavikdbavishi | 14:52 | |
| mordred | corvus: we are, in fact, just fetching the manifest and not pulling the image | 14:53 |
| corvus | mordred: fingers crossed then :) | 14:56 |
| mordred | corvus: once the job lands, there's a dockefile improvement from ianw that's a good candidate for landing and triggering new image builds | 15:01 |
| openstackgerrit | Merged zuul/zuul-jobs master: Upload images to dockerhub with buildx when using buildx https://review.opendev.org/726033 | 15:01 |
| mordred | speak of the devil | 15:01 |
| *** tosky_ is now known as tosky | 15:04 | |
| tobiash | corvus: I'm +2 on your pipeline manager work | 15:04 |
| corvus | tobiash: thanks! mordred ^ if you have a sec you might want to look at that -- we should be able to use it in opendev soon | 15:05 |
| corvus | 3 changes starting at https://review.opendev.org/722981 | 15:06 |
| mordred | corvus: looking | 15:06 |
| openstackgerrit | Jan Kubovy proposed zuul/zuul master: Driver event ingestion https://review.opendev.org/717299 | 15:07 |
| openstackgerrit | Pierre-Louis Bonicoli proposed zuul/nodepool master: openshiftpods: define ca_crt parameter if available https://review.opendev.org/726171 | 15:10 |
| *** bhavikdbavishi has quit IRC | 15:19 | |
| openstackgerrit | masterpe proposed zuul/zuul master: add installation manual for Ubuntu https://review.opendev.org/726179 | 15:29 |
| *** jamesmcarthur has joined #zuul | 15:32 | |
| AJaeger | anybody around for the final round of zuul-job loop vars fixes? please review https://review.opendev.org/#/c/724967/ | 15:46 |
| avass | promize this is the last one ;) | 15:47 |
| AJaeger | avass: let me have some hope ;) | 15:47 |
| mordred | AJaeger, avass this one seems to mostly be updates to the tests right? | 15:48 |
| AJaeger | we can stop the opensuse plain image testing on zuul-jobs, please review https://review.opendev.org/#/c/725750/ | 15:49 |
| AJaeger | mordred: yes! | 15:49 |
| avass | mordred: we already merged the fixes those updates detect | 15:49 |
| mordred | awesome | 15:49 |
| avass | mordred: if we don't merge that last one we don't catch anything inside a block | 15:50 |
| mordred | avass: I hear blocks are important | 15:50 |
| mordred | tristanC, tobiash: I made a silly little patch that I think you might like, since you have multi-tenant zuuls: https://review.opendev.org/#/c/725850/ | 15:52 |
| tobiash | mordred: cool :) | 15:53 |
| mordred | tobiash: (I've been switching between watching zuul-jobs patches in zuul and watching system-config patches in openstack and it's driving me crazy) | 15:54 |
| mordred | I like the suggestion of using url state actually - but I think that's slighty more work to think about | 15:55 |
| openstackgerrit | Merged zuul/zuul-jobs master: Check blocks recursively for loops https://review.opendev.org/724967 | 16:02 |
| openstackgerrit | Merged zuul/zuul-jobs master: Remove opensuse-15-plain testing https://review.opendev.org/725750 | 16:06 |
| mordred | Shrews: https://goo.gl/maps/TmWmeXJa3nGc8cHP6 <-- didn't know if you knew that was a thing | 16:14 |
| Shrews | mordred: knew of the town in MA, but not LA | 16:15 |
| *** zxiiro has joined #zuul | 16:19 | |
| openstackgerrit | Merged zuul/zuul master: Add serial pipeline manager https://review.opendev.org/722981 | 16:22 |
| openstackgerrit | Merged zuul/zuul master: Make fake test Gerrit merger more realistic https://review.opendev.org/722982 | 16:23 |
| openstackgerrit | Merged zuul/zuul master: Add more docs about pipelines https://review.opendev.org/724185 | 16:23 |
| *** jcapitao has quit IRC | 16:33 | |
| *** rpittau is now known as rpittau|afk | 16:34 | |
| *** evrardjp has quit IRC | 16:36 | |
| *** evrardjp has joined #zuul | 16:36 | |
| *** hashar has quit IRC | 16:39 | |
| tobiash | corvus: I'm currently analysing ways to improve our suffering from tenant reconfigurations. I just discovered that pushed tags lead to tenant reconfiguration events | 16:43 |
| tobiash | corvus: I'm thinking if we should filter them since we don't load config from tags anyway | 16:43 |
| avass | tobiash: yes please :) | 16:44 |
| clarkb | tags do sort of live in that weird place where they aren't really a branch so zuul gets confused applying branch configs to it too | 16:44 |
| clarkb | iirc corvus has a change up to tie tags back to branches and use the configs from the branch if found? which is in line with ignoring the tag ref specific configs | 16:45 |
| tobiash | clarkb: yeah, we do branch guessing, but zuul doesn't load config from the tags | 16:45 |
| avass | clarkb: I think that's lready merged isn't it? | 16:45 |
| tobiash | yes | 16:46 |
| clarkb | ah cool | 16:46 |
| *** sshnaidm is now known as sshnaidm|afk | 16:46 | |
| clarkb | in which case ++ to ignoring the tag events for erconfiguration | 16:46 |
| avass | clarkb: yeah: https://review.opendev.org/#/c/578557/ | 16:46 |
| mordred | tobiash: ++ to ignoring tag evens for reconfig | 16:47 |
| mordred | I bet that's impacting us too we just hadn't quite noticed it -especially around release time | 16:47 |
| fungi | clarkb: avass: https://review.opendev.org/578557 was the implementation for the tag-to-branch mapping, and it merged two months ago today. https://review.opendev.org/711741 adds some more testing for it around job configuration changes to the branch | 17:00 |
| *** toabctl has quit IRC | 17:04 | |
| openstackgerrit | Monty Taylor proposed zuul/zuul-jobs master: Add docker_dockerfile to upload-docker-image defaults https://review.opendev.org/726210 | 17:08 |
| mordred | clarkb, fungi, avass, tristanC: ^^ please | 17:08 |
| mordred | clarkb, fungi, avass, tristanC: https://zuul.opendev.org/t/zuul/build/dd142d271ca544af9f99f4a034c60884 is the error we hit in upload | 17:09 |
| openstackgerrit | Merged zuul/zuul master: Tie status filter text to pathname https://review.opendev.org/725850 | 17:11 |
| mordred | clarkb, avass: tyty | 17:12 |
| *** jpena is now known as jpena|off | 17:15 | |
| openstackgerrit | Pierre-Louis Bonicoli proposed zuul/nodepool master: openshiftpods: define ca_crt parameter if available https://review.opendev.org/726171 | 17:17 |
| *** gmann is now known as gmann_afk | 17:25 | |
| openstackgerrit | Merged zuul/zuul-jobs master: Add docker_dockerfile to upload-docker-image defaults https://review.opendev.org/726210 | 17:25 |
| openstackgerrit | Tobias Henkel proposed zuul/zuul master: Don't reconfigure the tenant on tag creation https://review.opendev.org/726213 | 17:29 |
| tobiash | mordred, avass: that should fix that ^ | 17:30 |
| tobiash | tiny change, huge impact, I roughly calculated our zuul stalls today due to tag pushes in one tenant to ~80min | 17:32 |
| clarkb | tobiash: is that a lot of tags or expensive reconfigures (maybe both) | 17:33 |
| tobiash | clarkb: 42 reconfigures with an average of 2-3 minutes each | 17:33 |
| tobiash | reconfigure time grows linearly with the queue and that tenant always has a large queue ;) | 17:34 |
| tobiash | in total there were 57 reconfigs, so only 15 legit reconfigs | 17:35 |
| avass | wow | 17:35 |
| clarkb | I wonder if we can also short circuit earlier if the zuul configs aren't updated? | 17:37 |
| tobiash | and we noticed that we didn't have the c extension of pyyaml in our container | 17:37 |
| clarkb | tobiash: we should check that on the upstream container too | 17:37 |
| tobiash | clarkb: I didn't find libyaml-dev somewhere, so I guess upstream container doesn't have this as well | 17:38 |
| clarkb | ya I think we want toadd libyaml-dev to bindep under the compile profile and libyaml always? | 17:38 |
| tobiash | yes | 17:38 |
| * clarkb searches distros for package names | 17:39 | |
| tobiash | ubuntu and debian seem to use libyaml-0-2 | 17:39 |
| *** rfolco|rover has quit IRC | 17:40 | |
| clarkb | yup | 17:41 |
| clarkb | fedora is libyaml and libyaml-devel | 17:41 |
| *** rfolco has joined #zuul | 17:42 | |
| clarkb | I've got a change ready just need to confirm the profiles for redhat vs suse as tehy use different names for the main lib package | 17:45 |
| *** rfolco is now known as rfolco|rover | 17:46 | |
| tobiash | cool | 17:46 |
| openstackgerrit | Clark Boylan proposed zuul/zuul master: Update bindep to include libyaml https://review.opendev.org/726216 | 17:48 |
| clarkb | oh hrm may need to split fedora and centos too | 17:51 |
| clarkb | ya there is no libyaml-devel on centos/rhel? but there is a libyaml? | 17:52 |
| clarkb | how does that work | 17:52 |
| clarkb | does the libyaml package include headers maybe? or are we not expected to link against that system lib? | 17:52 |
| clarkb | tristanC: ^ do you know? | 17:52 |
| *** jamesmcarthur has quit IRC | 17:53 | |
| tristanC | clarkb: `podman run -it --rm registry.centos.org/centos:7 yum info libyaml-devel` shows : libyaml-devel 0.1.4 | 17:55 |
| fungi | https://centos.pkgs.org/8/centos-powertools-x86_64/libyaml-devel-0.1.7-5.el8.x86_64.rpm.html | 17:55 |
| fungi | so 0.1.7 on centos8 | 17:56 |
| fungi | maybe you have to do something special to enable "powertools"? | 17:56 |
| clarkb | tristanC: I was looking at 8 there is no libyaml-devel in the base os but libyaml is | 17:56 |
| clarkb | and there is no libyaml-devel in the devel repo either | 17:56 |
| clarkb | fungi: oh weird a powertools repo is where its gone? | 17:57 |
| clarkb | (I mean there is a separate Devel repo with -devel header packages in it...) | 17:57 |
| clarkb | 7 had both in the base OS | 17:57 |
| clarkb | I guess my change is fine then and if you are on rhel8 you enable the extra repo? | 17:57 |
| clarkb | http://mirror.centos.org/centos/8/BaseOS/x86_64/os/Packages/ or http://mirror.centos.org/centos/8/Devel/x86_64/os/Packages/ is where I woudl expect ti given what 7 did and how 8 seems to do devel packages | 17:58 |
| *** jamesmcarthur has joined #zuul | 17:59 | |
| clarkb | but if it is in http://mirror.centos.org/centos/8/PowerTools/x86_64/os/Packages/ thats good enough for me I suppose | 17:59 |
| fungi | yeah, we do seem to mirror that in opendev | 18:00 |
| fungi | though jobs might need to explicitly enable that set? i guess we'll find out | 18:01 |
| fungi | https://static.opendev.org/mirror/centos/8/PowerTools/x86_64/os/Packages/libyaml-devel-0.1.7-5.el8.i686.rpm | 18:01 |
| fungi | definitely there | 18:01 |
| *** jamesmcarthur has quit IRC | 18:03 | |
| *** rlandy is now known as rlandy|biab | 18:07 | |
| openstackgerrit | Albin Vass proposed zuul/zuul master: Validate ansible extra packages https://review.opendev.org/724110 | 18:30 |
| *** rlandy|biab is now known as rlandy | 18:39 | |
| *** jamesmcarthur has joined #zuul | 18:39 | |
| *** saneax has quit IRC | 18:44 | |
| *** saneax has joined #zuul | 18:44 | |
| openstackgerrit | Albin Vass proposed zuul/zuul master: Validate ansible extra packages https://review.opendev.org/724110 | 18:55 |
| mordred | tristanC: https://zuul.opendev.org/t/zuul/build/40d4102be8ea4e308ab5ea71e2255871 <-- that looks like a temp error, yes? | 18:58 |
| mordred | tristanC: or do you think we need to investigate more deeply | 18:59 |
| tristanC | that seems odd, how can the zk class doesn't have a client set at this time... | 19:00 |
| corvus | mordred, tristanC: the context above is important -- that's after a test abort due to timeout | 19:02 |
| corvus | so the nonetype error is just unclean shutdown from a failed test | 19:02 |
| corvus | why is there no test-results log? | 19:03 |
| openstackgerrit | Merged zuul/zuul master: Update bindep to include libyaml https://review.opendev.org/726216 | 19:03 |
| mordred | clarkb: the missing libyaml could maybe also explain some of the memory pressure on zuul-scheduler for us | 19:03 |
| *** gmann_afk is now known as gmann | 19:06 | |
| corvus | mordred, tristanC: "state":{"waiting":{"reason":"ContainerCreating"} --- that's the last status before the timeout, so yeah, i'm guessing k8s was just slow there? | 19:06 |
| mordred | it was on rax-ord - so not a usual slow cloud - but could be | 19:07 |
| mordred | corvus: I think it's reasonable to do a recheck then | 19:07 |
| mordred | corvus: like - it doesn't seem like a thing that's likely related to changing how we install the gpg key of the openstack-ci-core ppa | 19:07 |
| corvus | mordred: yep | 19:20 |
| *** dpawlik has quit IRC | 19:23 | |
| openstackgerrit | Albin Vass proposed zuul/zuul master: Validate ansible extra packages https://review.opendev.org/724110 | 19:27 |
| openstackgerrit | Albin Vass proposed zuul/zuul master: Validate ansible extra packages https://review.opendev.org/724110 | 19:31 |
| avass | corvus, mordred: could one of you take a look at that ^ whenever you have time | 19:32 |
| corvus | avass: lgtm | 19:38 |
| *** y2kenny has quit IRC | 19:49 | |
| *** jamesmcarthur_ has joined #zuul | 19:58 | |
| *** jamesmcarthur_ has quit IRC | 19:59 | |
| *** jamesmcarthur_ has joined #zuul | 19:59 | |
| *** jamesmcarthur has quit IRC | 20:00 | |
| corvus | 09:47 < paladox> corvus hi, it seems https://ci.gerritcodereview.com/t/gerrit/build/89e7c54f6f374b4b9dd26f22f4f9ad9e is saying "gerrit_root" is undefined. | 20:15 |
| corvus | paladox: ^ sorry i missed | 20:15 |
| corvus | paladox: and also sorry i got swamped with other stuff and let the gerrit work fall behind a bit | 20:15 |
| corvus | i'll check that now | 20:15 |
| paladox | ah, it's ok :) | 20:16 |
| *** ericsysmin has joined #zuul | 20:20 | |
| corvus | paladox: remote: https://gerrit-review.googlesource.com/c/zuul/jobs/+/266472 Define gerrit_root in copy-plugin-deps [NEW] | 20:21 |
| paladox | corvus thanks! | 20:22 |
| paladox | 1+d | 20:22 |
| corvus | paladox: looks like further work is waiting on someone to express an opinion on my screed in https://gerrit-review.googlesource.com/265212 | 20:22 |
| paladox | oh | 20:23 |
| mordred | corvus: wow - we are hitting every possible flaky bug with the nodepool patch :( | 20:47 |
| mordred | corvus: https://zuul.opendev.org/t/zuul/build/cee058a867914cc795c256e2111e271e | 20:47 |
| corvus | mordred: :( | 20:48 |
| fungi | should we keep that patch around and use it to find all our bugs? ;) | 20:57 |
| *** calebb has quit IRC | 20:59 | |
| mordred | fungi: that's a great idea | 21:01 |
| openstackgerrit | James E. Blair proposed zuul/zuul master: Ignore source_context and description in job changes https://review.opendev.org/726247 | 21:02 |
| corvus | clarkb, mordred: ^ | 21:02 |
| clarkb | ah ok so the source_context updated due to the moves and that cauesd it to diff? | 21:03 |
| corvus | clarkb: yeah, that's a reasonable theory (hard to know for sure) | 21:03 |
| *** rfolco|rover is now known as rfolco|rover|off | 21:11 | |
| corvus | mordred: zuul-scheduler doesn't work on current master even | 21:13 |
| corvus | even if we haven't figured out a way to validate, we should probably at least check the service logs for errors | 21:14 |
| corvus | mordred: sorry, wrong channel | 21:14 |
| openstackgerrit | Merged zuul/nodepool master: Dockerfile: use local openstack-ci-core PPA key https://review.opendev.org/724462 | 21:24 |
| *** saneax has quit IRC | 21:44 | |
| openstackgerrit | Monty Taylor proposed zuul/zuul-jobs master: Actually include platform in the upload build https://review.opendev.org/726251 | 21:58 |
| mordred | corvus, clarkb, avass: ^^ please be prepared to laugh | 21:58 |
| clarkb | mordred: does that need an if block to check if arch is set? | 21:59 |
| clarkb | mordred: otherwise we'll do 'platform=' ? | 22:00 |
| mordred | clarkb: it's already in one | 22:00 |
| mordred | clarkb: that file only gets included if arch is set | 22:00 |
| mordred | oh - one more thing | 22:00 |
| openstackgerrit | Merged zuul/zuul master: Ignore source_context and description in job changes https://review.opendev.org/726247 | 22:01 |
| openstackgerrit | Monty Taylor proposed zuul/zuul-jobs master: Process siblings in upload-image push https://review.opendev.org/726253 | 22:02 |
| mordred | clarkb, corvus : that's for completeness ^^ - I don't think we really plan to publish images built with siblings at the moment - but if we did, it would not work right without that | 22:04 |
| clarkb | mordred: is that push step actually rebuilding images though? or is it just going to notice that the image its already got built is pushable? | 22:04 |
| mordred | clarkb: it's supposed to be the second - but my concern was that if the build context wasnt' the same that it's caching algorithm would decide to re-build the image | 22:06 |
| clarkb | ah | 22:06 |
| corvus | mordred: looks good all around. the --arch argument being necessary for push seems weird to me but, similar to what you said, everything about this is weird. | 22:12 |
| *** jamesmcarthur_ has quit IRC | 22:13 | |
| *** jamesmcarthur has joined #zuul | 22:13 | |
| mordred | corvus: yeah - I want to completely rethink it all now that we know things | 22:15 |
| *** jamesmcarthur has quit IRC | 22:19 | |
| openstackgerrit | Merged zuul/zuul-jobs master: Actually include platform in the upload build https://review.opendev.org/726251 | 22:21 |
| openstackgerrit | Merged zuul/zuul-jobs master: Process siblings in upload-image push https://review.opendev.org/726253 | 22:21 |
| *** tumble has quit IRC | 22:28 | |
| *** jamesmcarthur has joined #zuul | 22:33 | |
| *** jamesmcarthur has quit IRC | 22:37 | |
| *** jamesmcarthur has joined #zuul | 22:38 | |
| *** jamesmcarthur has quit IRC | 22:42 | |
| *** jamesmcarthur has joined #zuul | 23:01 | |
| *** jamesmcarthur has quit IRC | 23:04 | |
| *** jamesmcarthur has joined #zuul | 23:05 | |
| *** jamesmcarthur has quit IRC | 23:11 | |
| *** jamesmcarthur has joined #zuul | 23:11 | |
| openstackgerrit | Monty Taylor proposed zuul/zuul-jobs master: Don't upload to the tag with buildx - only to the change tag https://review.opendev.org/726261 | 23:18 |
| openstackgerrit | Merged zuul/zuul-jobs master: Don't upload to the tag with buildx - only to the change tag https://review.opendev.org/726261 | 23:34 |
| *** tosky has quit IRC | 23:36 | |
| *** jamesmcarthur has quit IRC | 23:38 | |
| *** jamesmcarthur has joined #zuul | 23:38 | |
| openstackgerrit | Merged zuul/nodepool master: Logs stats for nodepool automated cleanup https://review.opendev.org/614074 | 23:52 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!