| *** __ministry is now known as Guest5847 | 01:55 | |
| *** __ministry is now known as Guest5870 | 07:12 | |
| opendevreview | Lajos Katona proposed openstack/project-config master: Remove taas-tempest-plugin and taas-dashboard jobs https://review.opendev.org/c/openstack/project-config/+/938665 | 10:30 |
|---|---|---|
| opendevreview | Merged openstack/project-config master: Remove taas-tempest-plugin and taas-dashboard jobs https://review.opendev.org/c/openstack/project-config/+/938665 | 14:50 |
| *** __ministry is now known as Guest5909 | 16:08 | |
| JayF | I'm working on (as mentioned previously) deprecating the ironic-lib repo. Is there a suggested or example config for "enough jobs for zuul to merge the final commit"? I believe if I just pull the zuul config it won't work | 21:03 |
| JayF | https://docs.opendev.org/opendev/infra-manual/latest/drivers.html#step-1-end-project-gating 🦆 | 21:06 |
| clarkb | JayF: the minimal job is the noop job | 21:19 |
| JayF | yeah, that doc answered my question | 21:19 |
| clarkb | its a built in in zuul that immediately returns success | 21:20 |
| JayF | clarkb: hmm, it seems to still have a test-release-openstack: https://review.opendev.org/c/openstack/ironic-lib/+/939277 | 21:23 |
| JayF | did I miss something? | 21:23 |
| clarkb | JayF: https://opendev.org/openstack/project-config/src/branch/master/zuul.d/projects.yaml#L2022-L2026 I think that comes from here | 21:24 |
| clarkb | you probably do need to keep those jobs in some capacity since you are trying to keep older branches alive. | 21:24 |
| clarkb | you might need to break official-openstack-repo-jobs out of the template and make them branch specific? I don't think publish-to-pypi is a problem as that should only run post merge | 21:25 |
| opendevreview | Jay Faulkner proposed openstack/project-config master: Deprecate ironic-lib https://review.opendev.org/c/openstack/project-config/+/939282 | 21:25 |
| clarkb | JayF: ^ only makes sense if you're never releasing the library again | 21:25 |
| JayF | ooooooh | 21:26 |
| clarkb | which is what you do when retiring but I think ironic-lib is keeping old branches alive | 21:26 |
| JayF | you are 100% correct | 21:26 |
| JayF | I just -1'd my patch there | 21:26 |
| JayF | could I do something like, restore what was there, and add a separate stanza breaking out master with only noop-jobs | 21:26 |
| JayF | or are those always additive | 21:26 |
| clarkb | the official-openstack-repo-jobs are always going to add so you need to remove that template then add back in only what you want | 21:27 |
| JayF | ack | 21:27 |
| JayF | official-openstack-repo-jobs is just the sync to github, so it can stay | 21:29 |
| JayF | so I should just need to do publish-to-pypi | 21:30 |
| clarkb | oh I thought that included the release check but ya I may have mixed up the templates | 21:30 |
| opendevreview | Jay Faulkner proposed openstack/project-config master: Deprecate ironic-lib https://review.opendev.org/c/openstack/project-config/+/939282 | 21:34 |
| JayF | that should be better | 21:34 |
| clarkb | I think zuul isn't happy about the syntax. Looks liek you defined a second block for the same project. I was suggesting that you take the jobs from the template and expand them out of the template directly into the project config then you can specify the jobs run on not master | 21:36 |
| JayF | oh, that's unfortunate | 21:37 |
| clarkb | I don't know what openstack's rules are but an alternative may be to just leave master and everything else the way it is except to update the README on master saying this is deprecated | 21:38 |
| clarkb | having the job config isn't a big deal for one extra branch. Then when you retire the project because all stable branches haev completely aged out you just remove everything at once | 21:38 |
| JayF | At this point, this patch is the only thing that needs work, so I'll just work it out here | 21:39 |
| JayF | IMO it's important to "break" master so we don't have any unattentive downstreams trying to run old-master ironic-lib against ironic/ipa that don't need it | 21:40 |
| opendevreview | Jay Faulkner proposed openstack/project-config master: Deprecate ironic-lib https://review.opendev.org/c/openstack/project-config/+/939282 | 21:44 |
| JayF | clarkb: I'm a little confused about what it's angry about now? | 21:46 |
| clarkb | JayF: your depends on have created a cycle which is not allowed | 21:47 |
| clarkb | ironic-lib should depends on project-config and not the other direction | 21:47 |
| JayF | OH | 21:47 |
| clarkb | but also the branch specification is only needed in check and gate | 21:47 |
| JayF | that depends-on is just a bad paste | 21:47 |
| JayF | it was supposed to eb the governance | 21:47 |
| clarkb | the other two pipelines are tag specific and on't care about branches iirc | 21:47 |
| JayF | I was looking for answers in the yaml file :) | 21:47 |
| opendevreview | Jay Faulkner proposed openstack/project-config master: Deprecate ironic-lib https://review.opendev.org/c/openstack/project-config/+/939282 | 21:47 |
| clarkb | JayF: you also need some :'s to start the yaml dict context | 21:48 |
| clarkb | - test-release-openstack:\n branches: etc | 21:48 |
| clarkb | but the shape of that is largely correct I think | 21:49 |
| opendevreview | Jay Faulkner proposed openstack/project-config master: Deprecate ironic-lib https://review.opendev.org/c/openstack/project-config/+/939282 | 21:49 |
| JayF | makes sense, I always forget it when non-voting a job, might as well forget it here lol | 21:50 |
| clarkb | JayF: the latest error is due to the unindenting the change does. YOu need to reindent a level (see the old side diff and it shows the unindent | 21:52 |
| JayF | ack | 21:52 |
| opendevreview | Jay Faulkner proposed openstack/project-config master: Deprecate ironic-lib https://review.opendev.org/c/openstack/project-config/+/939282 | 21:55 |
| clarkb | fungi can double check that | 21:58 |
| clarkb | I +2'd I think that is roughly correct | 21:58 |
| fungi | lookin' | 21:59 |
| fungi | it's funny with all the work that went into splitting out *-lib repos so that projects could depend on (previousls) service internals, but is what it is i guess | 22:01 |
| JayF | fungi: this wasn't one of those | 22:03 |
| JayF | fungi: this was just for ironic use only, we had a LOT of shared disk imaging/management code in there when the iscsi driver existed | 22:03 |
| JayF | since it's been retired, ironic-lib had maybe 4-5 modules, only one of which had any meaningful sharing between Ironic and IPA, and it's existance made fixing the recent CVE *harder*, not easier, so it was a pretty easy choice for us tbh | 22:04 |
| JayF | (since the iscsi driver has been retired) | 22:04 |
| fungi | aha, so not necessarily a sign than neutron would want to re-absorb neutron-lib, for example | 22:05 |
| JayF | yes, exactly | 22:05 |
| JayF | this was explicitly for ironic use only, the only things that *ever* depended on it were ironic / ipa | 22:05 |
| JayF | and like I said, it sorta became a vestige/obstacle as soon as the conductor got outta the business of imaging disks and all that code concentrated in IPA | 22:05 |
| fungi | makes sense. neutron-lib evolved because of all the external drivers reusing and depending on code from inside the neutron repo, for comparison | 22:07 |
| JayF | yeah, we are a primary consumer of that too, fun enough (networking-baremetal + networking-generic-switch) | 22:08 |
| JayF | I still remember suggesting ironic lib be created back at the Paris summit. So at least it's appropriate that I'm the one to pay back the technical debt that I helped create lol | 22:12 |
| JayF | fungi: re: your review comment; I'd say there's like a 1% chance we'll ever need to release from those ironic-lib stable repos, so we'll likely never know :D | 22:37 |
| fungi | fair enough | 22:38 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!