| -@gerrit:opendev.org- Zuul merged on behalf of Benjamin Schanzel: [zuul/zuul] 933085: zuul-web: Simplify tenant list https://review.opendev.org/c/zuul/zuul/+/933085 | 03:43 | |
| @anoop-jose:matrix.org | Hi guys, I am facing issue while connecting Keycloak with Zuul. The keycloak container is up and running in the zuul network itself. | 04:35 |
|---|---|---|
| I have followed: https://zuul-ci.org/docs/zuul/latest/tutorials/keycloak.html | ||
| Keycloak is integrated with LDAP and they are all working fine. | ||
| Issue facing: I could see the signun icon now in zuul interface, but on clicking it is not redirecting to the keycloak login page. Could someone direct what am missing here? | ||
| -@gerrit:opendev.org- Zuul merged on behalf of Simon Westphahl: | 06:39 | |
| - [zuul/zuul] 931780: Implement re-use of ready nodes w/o request https://review.opendev.org/c/zuul/zuul/+/931780 | ||
| - [zuul/zuul] 932170: Store state time for provider nodes https://review.opendev.org/c/zuul/zuul/+/932170 | ||
| -@gerrit:opendev.org- Zuul merged on behalf of Simon Westphahl: [zuul/zuul] 932179: Only expire nodes when no pending layout update https://review.opendev.org/c/zuul/zuul/+/932179 | 06:44 | |
| -@gerrit:opendev.org- Zuul merged on behalf of James E. Blair https://matrix.to/#/@jim:acmegating.com: [zuul/zuul] 933065: Re-add volume attributes to aws driver https://review.opendev.org/c/zuul/zuul/+/933065 | 06:44 | |
| -@gerrit:opendev.org- Zuul merged on behalf of James E. Blair https://matrix.to/#/@jim:acmegating.com: [zuul/zuul] 933374: Fix openstack region handling https://review.opendev.org/c/zuul/zuul/+/933374 | 06:44 | |
| -@gerrit:opendev.org- Felix Edel proposed: [zuul/zuul] 933081: Visualize fragmentation of QueueItem progressbar https://review.opendev.org/c/zuul/zuul/+/933081 | 08:19 | |
| -@gerrit:opendev.org- Slawek Kaplonski proposed: [zuul/zuul-jobs] 933395: Drop support for user/password authentication to the readthedocs.org https://review.opendev.org/c/zuul/zuul-jobs/+/933395 | 08:29 | |
| @jangutter:matrix.org | ahasai: I think I finally found the upstream openinfra config for zuul.conf: https://opendev.org/opendev/system-config/src/branch/master/playbooks/roles/zuul/templates/zuul.conf.j2 | 09:37 |
| @karlatec:matrix.org | How do I properly pass around artifacts between builds? I have job A which generates a tarball, and job B which uses this tarball as an input. | 11:51 |
| I've uploaded the tarball onto a file server and used [zuul_return](https://zuul-ci.org/docs/zuul/latest/job-content.html#returning-artifact-urls) to pass the URL information so that job B can download the artifact (using artifact name to specify which item to download). | ||
| Several builds later I noticed that the list in zuul.artifacts keeps growing and get_url in job B always downloading the tarball from first successful job A build (because it's first in zuul.artifacts). | ||
| Do I need to filter out zuul.artifacts not just by artifact name, but also by build ID / change ID / etc? Or is there some other (better?) way? | ||
| @karlatec:matrix.org | * How do I properly pass around artifacts between builds? I have job A which generates a tarball, and job B which uses this tarball as an input. | 12:27 |
| I've uploaded the tarball onto a file server and used [zuul\_return](https://zuul-ci.org/docs/zuul/latest/job-content.html#returning-artifact-urls) to pass the URL information so that job B can download the artifact (using artifact name to specify which item to download). | ||
| Several builds later I noticed that the list in zuul.artifacts keeps growing and get\_url in job B is always downloading the tarball from first successful job A build (because it's first in zuul.artifacts). | ||
| Do I need to filter out zuul.artifacts not just by artifact name, but also by build ID / change ID / etc? Or is there some other (better?) way? | ||
| @clarkb:matrix.org | Karol: are you using the zuul-jobs' download-artifacts role? | 13:13 |
| @karlatec:matrix.org | I am not. I already looked through zuul-jobs, so I must have missed it. Looking at it now. | 13:14 |
| @clarkb:matrix.org | in particular it defaults to querying for the current change and patchset's artifacts which should limit the artifact space | 13:15 |
| @clarkb:matrix.org | ok I would start there and see if that provides the desired behavior. And if not debug from there | 13:15 |
| @karlatec:matrix.org | For now I've worked around this with something similar - filtering artifacts using name & metadata + event_id. | 13:18 |
| So is the list in zuul.artifacts supposed to grow like that? | ||
| @fungicide:matrix.org | you'll get artifacts from changes ahead of the current change in a dependent pipeline too, so potentially yes | 13:20 |
| @fungicide:matrix.org | builds can (and often do, depending on how you've designed them) consume artifacts from builds for other changes that a change depends on | 13:21 |
| @clarkb:matrix.org | I wonder if download-artifacts predates having that info in the zuul variable and maybe it should refer to there too? Not sure | 13:22 |
| @clarkb:matrix.org | but yes if you've got other changes providing the artifact running builds in the same pipeline then they are supposed to show up in that list | 13:22 |
| @karlatec:matrix.org | > you'll get artifacts from changes ahead of the current change in a dependent pipeline too, so potentially yes | 13:28 |
| That already happens in my pipeline, I just simplified this for asking the question :) | ||
| By asking about growing zuul.artifacts I had something else in mind. I have a series of changes on Gerrit, eg. 100, 101 and 102. I was a bit surprised too see artifacts from older changes in chain when testing change 102. But now I know it's "global" and already know what to expect. | ||
| @clarkb:matrix.org | looking at things like the pull-from-intermediate-registry it loops over the artifacts list pulling in containers. It does so in order which I think means the most recent tagged version overwrites at the end (since most recent is last) | 13:35 |
| @clarkb:matrix.org | other roles also iterate over the list in order so that may be how things are generally handled to ensure latest wins | 13:35 |
| @jangutter:matrix.org | Karol: we saw something similar in the gate (unrelated to artifacts, more to the merger). We got these weird errors about missing jobs. The cause was we had two unrelated projects on differing branching structures... if a change goes into the gate, the merger starts with the job variants from the first, if a second change queues behind it, the merger might fail because the branches don't exist on the second project. The solution was to ensure they're on separate queues. Took us a while to notice because the teams were timezone separated. | 13:38 |
| Our experience has been mostly with passing artifacts between builds in a buildset, however. | ||
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: | 15:18 | |
| - [zuul/zuul] 933377: Launcher: fix deleting servers https://review.opendev.org/c/zuul/zuul/+/933377 | ||
| - [zuul/zuul] 933436: Fix endpoint races https://review.opendev.org/c/zuul/zuul/+/933436 | ||
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: | 18:21 | |
| - [zuul/zuul] 933436: Fix endpoint races https://review.opendev.org/c/zuul/zuul/+/933436 | ||
| - [zuul/zuul] 933377: Launcher: fix deleting servers https://review.opendev.org/c/zuul/zuul/+/933377 | ||
| @blart_versenwald_3:matrix.org | Hello! via the '/api/tenant/{tenant_name}/autohold' API I can get the ID's of the nodes. Is it possible via REST API to get the IP address to these nodes ? | 18:27 |
| I see '/api/tenant/{tenant_name}/nodes', but it does not give me the IP addr, but something called **external_id** However I am not sure how (if) I can use that in some way. | ||
| (Usually I via browser go to the build -> logs -> zuul-info -> inventory.log and find the IP addr in that file (not sure it always looks like that, I am not a zuul admin just a user trying to find out if I can grab this info somewhat easier) | ||
| Thanks | ||
| @jim:acmegating.com | blart_versenwald_3: the thing you currently do is the only current way to get that info | 18:28 |
| @blart_versenwald_3:matrix.org | > <@jim:acmegating.com> blart_versenwald_3: the thing you currently do is the only current way to get that info | 18:50 |
| Thanks, then I know :) | ||
| @jim:acmegating.com | i think we may be able to improve this later after the nodepool in zuul work; but probably not until after all of that is done. | 18:54 |
| @sean-k-mooney:matrix.org | nodepool in zuul? are you looking at absorbing that functionatly into zuul itself instead of keepign it as a seperate process iteracted with vai zookeeper? | 18:55 |
| @jim:acmegating.com | sean-k-mooney: yes, implementation in progress: https://zuul-ci.org/docs/zuul/latest/developer/specs/nodepool-in-zuul.html | 18:56 |
| @sean-k-mooney:matrix.org | ah interesting | 18:57 |
| @sean-k-mooney:matrix.org | moving the actual image building into a zuul job has some advantages for sure. i.e. you can use resouces form the nodepool provider rather then the nodepool host ot do that building which is proably more secure the the current approch | 18:59 |
| @jim:acmegating.com | yeah, i think it's going to be nice and a lot more user-serviceable | 18:59 |
| @sean-k-mooney:matrix.org | the only downside that comes to mind diretly is perhasp a more complex requirement for strogage of the image | 19:00 |
| @sean-k-mooney:matrix.org | depending on how you do the output | 19:01 |
| @sean-k-mooney:matrix.org | i.e. copying the executore vs as an artifact stored on an object store ectra | 19:01 |
| @jim:acmegating.com | yep... that one is a mixed bag (some things a little better, some things a little worse). | 19:03 |
| @sean-k-mooney:matrix.org | i might need to add https://en.wikipedia.org/wiki/Rendezvous_hashing to my reading list | 19:04 |
| @jim:acmegating.com | it's pretty nifty; we avoid a lot of zk thundering herd behavior for locks that way. that's been a problem with nodepool today. | 19:05 |
| @sean-k-mooney:matrix.org | im wondering if it has any applciation for nova's schduler and how we select hsot fpr wokloads | 19:06 |
| @sean-k-mooney:matrix.org | placement mostly allow use to avoid racing for the last resouce when you have multiple schdluer but not always | 19:07 |
| @jim:acmegating.com | > <@sean-k-mooney:matrix.org> im wondering if it has any applciation for nova's schduler and how we select hsot fpr wokloads | 19:08 |
| after you read up on it; let me know what conclusion you come to :) | ||
| @blart_versenwald_3:matrix.org | corvus: Found out that I could use information in autohold to get the build id, then use that to find out the path to the inventory file and find the IP in there. So I will get the ip addr, but with some digging ;) | 19:09 |
| @clarkb:matrix.org | We aren't using it outside of the niz work right? Or did it get used elsewhere? | 19:09 |
| @jim:acmegating.com | blart_versenwald_3: ++ yep you can probably script that :) | 19:09 |
| @jim:acmegating.com | Clark: just niz | 19:09 |
| @jim:acmegating.com | it may have some applicability to the merger/executor job queue; but let's see what we think of it for niz first :) | 19:11 |
| @sean-k-mooney:matrix.org | this all look very interesting but probaly too much for a firday after the ptg :) but ya ill let you know if that is somethign we can learn form. | 19:11 |
| @clarkb:matrix.org | corvus: question(s) on https://review.opendev.org/c/zuul/zuul/+/931704 around snychronization for file writes to the same file in different threasds | 20:19 |
| @jim:acmegating.com | Clark: excellent question. i have replied and welcome your further thoughts. :) | 21:02 |
| @clarkb:matrix.org | and posted | 21:17 |
| @clarkb:matrix.org | I'm good with it as is but didn't approve in case we want more feedback on that point | 21:17 |
| @mordred:waterwanders.com | corvus: ^^ I don't even want to know how that was diagnosed :) | 22:45 |
| @mordred:waterwanders.com | corvus: also - in the conflict list is https://review.opendev.org/c/zuul/zuul/+/541434 - which reminded me I never finished that :( | 22:48 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!