| opendevreview | Dr. Jens Harbott proposed openstack/project-config master: Only pause update_constraints.sh when needed https://review.opendev.org/c/openstack/project-config/+/946541 | 05:36 |
|---|---|---|
| *** ykarel_ is now known as ykarel | 12:52 | |
| *** vkmc_ is now known as vkmc | 13:01 | |
| clarkb | tonyb: tkajinam exactly. The schedule is there to try and bring order to chaos and make timing predictable for people around the world | 15:02 |
| clarkb | but the service is generally available | 15:02 |
| *** diablo_rojo_phone is now known as Guest13175 | 16:32 | |
| *** Guest13175 is now known as help | 17:03 | |
| *** help is now known as Guest13179 | 17:04 | |
| *** Guest13179 is now known as diablo_rojo_phone | 17:06 | |
| JayF | https://review.opendev.org/c/openstack/ironic/+/942223 This is in a super confusing status and came up in an ironic review jam, can you help me understand what happened? cc TheJulia cid_ | 18:41 |
| fungi | it looks like zuul did not merge the commit after testing succeeded in the gate pipeline | 18:42 |
| fungi | you would normally never see the "ready to submit" status because normally changes exist in that state for only a brief moment between when zuul votes +2 verified and auto-submits the change | 18:43 |
| clarkb | its based on the old patchset of the parent so cannot merge | 18:43 |
| fungi | in this case, it looks like its parent commit is an outdated revision of another change | 18:43 |
| fungi | the parent is patchset 1 of https://review.opendev.org/c/openstack/ironic/+/942204 but patchset 2 is what merged | 18:44 |
| clarkb | if the change was in the gate when the parent was updated you'd get a nice message about that. But looks like if you approve a change in that state later this is what happens | 18:44 |
| fungi | so gerrit won't allow 942223 to merge in its present state because it has a parent commit which can never appear in the branch history now | 18:44 |
| fungi | 942223 will need to be rebased onto 942204,2 or the current state of the master branch | 18:45 |
| clarkb | it is interesting that gerrit says "ready to submit" when it will fail to do so if you click the button (that requires extra perms we don't have but zuul would have attempted it and that failed) | 18:47 |
| fungi | maybe a useful zuul feature for the gerrit driver? report an error instead of running tests, maybe consider an outdated parent the same as a missing merge requirement? | 18:49 |
| clarkb | zuul won't know until it tries to submit | 18:51 |
| clarkb | because zuul relies on the "ready to submit" criteria which we can see is buggy here | 18:52 |
| clarkb | so I think two things. First would be fixing gerrit's detection of this case not actually being submittable before you try it. And second would be a message from zuul when submit fails with a hint as to why | 18:52 |
| fungi | yeah, good point, gerrit's not supposed to consider this change ready to submit if it is going to refuse to submit it | 18:55 |
| clarkb | the way zuul checks is there is query to check what is missing from the change to be ready to submit. Zuul queries that and ignores the verified vote missing error and if everything else looks good it proceeds | 18:56 |
| clarkb | gerrit uses the same check to determine if the change is ready to submit but doesn't ignore the verified vote | 18:56 |
| JayF | thanks guys for the context, that's useful | 19:15 |
| JayF | we're going to rebase it so it can land | 19:15 |
| JayF | It's rare to see a new gerrit thing these days so I was curious how it got there :D | 19:15 |
| JayF | https://review.opendev.org/c/openstack/ironic/+/942223 rebased and workflow'd, thank you again! | 19:15 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!