| @jim:acmegating.com | nope, that's a can of worms as far as the query planner goes | 00:25 |
|---|---|---|
| @clarkb:matrix.org | Would two indexes work? Not sure how that gets layer out in the datastructure | 00:44 |
| @clarkb:matrix.org | Like if we do an index on project, and index on branch etc. Will that still allow it to be reasonably fast as it can filter sequentially? | 01:06 |
| @clarkb:matrix.org | I guess you want a unique constraint though which would require setting that across the 4 tuple? | 01:06 |
| @jim:acmegating.com | yeah, the search would be fast, but we'd lose the unique... that | 01:06 |
| @jim:acmegating.com | we could do our own hacky unique constraint with a hash or something | 01:06 |
| @jim:acmegating.com | but honestly, we're going pretty far down the rabbit hole; i'd like to take a step back and think a bit more before proceeding. | 01:08 |
| @clarkb:matrix.org | wfm | 01:14 |
| -@gerrit:opendev.org- Felix Edel proposed: [zuul/zuul] Don't use executor.builds when processing build result events https://review.opendev.org/c/zuul/zuul/+/808091 | 06:38 | |
| -@gerrit:opendev.org- Felix Edel proposed: | 08:15 | |
| - [zuul/zuul] Don't use executor.builds when processing build result events https://review.opendev.org/c/zuul/zuul/+/808091 | ||
| - [zuul/zuul] Don't use executor.builds to find out if tests are settled https://review.opendev.org/c/zuul/zuul/+/808792 | ||
| - [zuul/zuul] Remove the local builds list from the executor client https://review.opendev.org/c/zuul/zuul/+/809175 | ||
| @felixedel:matrix.org | @corvus: I've updated the changes to remove the local builds list in the executor (https://review.opendev.org/c/zuul/zuul/+/809175/6 + parent changes). They should be good to go now. | 08:22 |
| -@gerrit:opendev.org- Simon Westphahl proposed: | 08:29 | |
| - [zuul/zuul] wip: Simplified attribute API for ZKObjects https://review.opendev.org/c/zuul/zuul/+/809532 | ||
| - [zuul/zuul] wip: Make QueueItem a Zookeeper object https://review.opendev.org/c/zuul/zuul/+/809414 | ||
| -@gerrit:opendev.org- Simon Westphahl proposed on behalf of James E. Blair https://matrix.to/#/@jim:acmegating.com: [zuul/zuul] WIP: Add an API for ZK-backed objects https://review.opendev.org/c/zuul/zuul/+/809293 | 08:29 | |
| @jpew:matrix.org | Would someone mind helping me with https://review.opendev.org/c/zuul/zuul-operator/+/806452/ ? I don't *think* my change is what's causing the failure.... I'm not too familiar with all the Zuul components though | 12:42 |
| @mhuin:matrix.org | hello zuul-maint, here are two small changes that need a final +3: https://review.opendev.org/c/zuul/zuul/+/806201 and https://review.opendev.org/c/zuul/zuul/+/809423 | 13:03 |
| @westphahl:matrix.org | corvus: I found a small issue in the node request cleanup routine: https://review.opendev.org/c/zuul/zuul/+/809970 | 13:12 |
| -@gerrit:opendev.org- Albin Vass proposed: [zuul/zuul] Exit sucessfully when manipulating project keys https://review.opendev.org/c/zuul/zuul/+/810240 | 13:44 | |
| -@gerrit:opendev.org- Felix Edel proposed: [zuul/zuul] Remove job_name attribute form NodesProvisionedEvent https://review.opendev.org/c/zuul/zuul/+/810243 | 13:56 | |
| -@gerrit:opendev.org- Albin Vass proposed: [zuul/zuul] Exit sucessfully when manipulating project keys https://review.opendev.org/c/zuul/zuul/+/810240 | 14:04 | |
| @jim:acmegating.com | mhu: tiny thing noted on 201; 423+3 thx :) | 15:04 |
| @fungicide:matrix.org | anyone else have opinions on whether we should stop testing zuul-jobs changes on ubuntu xenial and with python 3.5? http://lists.zuul-ci.org/pipermail/zuul-discuss/2021-September/001728.html https://review.opendev.org/809899 | 15:18 |
| @jim:acmegating.com | fungi: that lgtm, i think the next question is when we should remove tox-py35 from zuul-jobs. maybe... a long time from now? maybe we just update its description noting it's untested/unsupported, and may be removed at some unspecified point? | 15:25 |
| @jim:acmegating.com | (but there's no compelling reason for us to remove it once we acknowledge it's untested) | 15:26 |
| @fungicide:matrix.org | corvus: that makes sense, should we also add comments around any places where playbooks/roles have conditional clauses mentioning xenial? | 15:34 |
| @jim:acmegating.com | fungi: hrm... maybe... or maybe we should just make a blanket statement in the docs? | 15:36 |
| @fungicide:matrix.org | somewhere central in the z-j docs or in the individual readmes for those various roles/playbooks? | 15:37 |
| @jim:acmegating.com | z-j docs | 15:39 |
| @fungicide:matrix.org | thanks, i'll come up with something appropriate | 15:40 |
| -@gerrit:opendev.org- Matthieu Huin https://matrix.to/#/@mhuin:matrix.org proposed: [zuul/zuul] web UI: add "show retries" toggles on buildset page https://review.opendev.org/c/zuul/zuul/+/806201 | 15:53 | |
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: [zuul/zuul] Fix race in job request queue cleanup https://review.opendev.org/c/zuul/zuul/+/810290 | 16:36 | |
| @jim:acmegating.com | looking at some of the recent test race/failures, i think that ^ should improve things | 16:36 |
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: [zuul/zuul] Add an API for ZK-backed objects https://review.opendev.org/c/zuul/zuul/+/809293 | 17:09 | |
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed on behalf of Simon Westphahl: | 17:09 | |
| - [zuul/zuul] Simplified attribute API for ZKObjects https://review.opendev.org/c/zuul/zuul/+/809532 | ||
| - [zuul/zuul] wip: Make QueueItem a Zookeeper object https://review.opendev.org/c/zuul/zuul/+/809414 | ||
| @jim:acmegating.com | swest: ^ i removed wip from the first 2 patches there and updated them to fix the copy() rollback issue. i think they're probably basically done, but i'll go ahead and set the actual gerrit WIP flag on the first change though in order to hold it until the series is ready. | 17:11 |
| -@gerrit:opendev.org- Jeremy Stanley proposed: [zuul/zuul-jobs] Deprecate EOL Python releases and OS versions https://review.opendev.org/c/zuul/zuul-jobs/+/810299 | 17:33 | |
| @fungicide:matrix.org | corvus: does that ^ cover what you had in mind? | 17:33 |
| @jim:acmegating.com | fungi: that looks great; i guess we are/should be keeping py27 around still because of centos? | 17:37 |
| @fungicide:matrix.org | corvus: sure, i'll buy that argument, we're effectively dropping 3.5 because we were relying on xenial to test it | 17:38 |
| @clarkb:matrix.org | CentOS 7 is still out there if the DOS opendev just dealt with is any indication | 17:41 |
| @fungicide:matrix.org | that's a lot of rechecks for unit test failures on a javascript-only change: https://review.opendev.org/803866 | 18:04 |
| @jim:acmegating.com | https://review.opendev.org/810290 should help with that | 18:04 |
| @fungicide:matrix.org | seems like test_zuul_refs is failing on py38? is that a known... | 18:04 |
| @fungicide:matrix.org | aha, thanks. reviewing | 18:05 |
| @jim:acmegating.com | though it failed on an OOM; i don't think that's our fault :) | 18:05 |
| @fungicide:matrix.org | probably inmotion-iad3 | 18:05 |
| @fungicide:matrix.org | we lowered the max_servers even further there earlier today | 18:05 |
| @jim:acmegating.com | airship-kna1 actually | 18:06 |
| @fungicide:matrix.org | oh, okay so different issue i guess :/ | 18:06 |
| @clarkb:matrix.org | All the things are keeping us on our toes today | 18:07 |
| @clarkb:matrix.org | https://zuul.opendev.org/t/zuul/build/db03012ff939456aa00ccce2b6001d11 seems registry.yarnpkg.com is giving us trouble | 18:09 |
| @clarkb:matrix.org | that job did run in the inmotion cloud. I can wget the file it failed to connect to fomr our mirror node in that cloud | 18:10 |
| @fungicide:matrix.org | Clark: there was a point in time where the mirror host was in a shutoff state and i had to start it | 18:11 |
| @clarkb:matrix.org | That is apparently just a cname for npm's registry and that has a status tracker https://status.npmjs.org/ | 18:11 |
| @clarkb:matrix.org | fungi: this is direct access to the url and not through the mirror | 18:11 |
| @clarkb:matrix.org | and it happened less than an hour ago. I was just using the mirror as a standin for host from that cloud | 18:12 |
| @fungicide:matrix.org | oh, yep sorry, i misread and thought you said it failed to connect to the mirror node | 18:12 |
| @clarkb:matrix.org | There is a host with that IP currently up and running and I can wget the url from that host now so unlikely to be an issue with that specific IP either | 18:13 |
| @clarkb:matrix.org | Three different jobs all failed to get the same resource at roughly the same time from inmotion though | 18:14 |
| @clarkb:matrix.org | Interesting two of the jobs had the asme IP (it was recycled when the first job failed) | 18:14 |
| @jpew:matrix.org | How do I set the user that zuul uses when running on my nodes? It's currently using root and I don't want it too | 19:54 |
| @clarkb:matrix.org | jpew: it is an attribute of the nodepool settings iirc | 19:55 |
| @clarkb:matrix.org | https://zuul-ci.org/docs/nodepool/openstack.html#attr-providers.[openstack].cloud-images.username for example | 19:56 |
| @jpew:matrix.org | Hmm, I'm using kubernetes pods; doesn't seem to be an option to set a username there | 20:02 |
| -@gerrit:opendev.org- Zuul merged on behalf of James E. Blair https://matrix.to/#/@jim:acmegating.com: [zuul/zuul] Fix race in job request queue cleanup https://review.opendev.org/c/zuul/zuul/+/810290 | 20:36 | |
| -@gerrit:opendev.org- Zuul merged on behalf of Andrii Ostapenko: [zuul/zuul] Treat 'waiting' jobs same as 'queued' in buildset progress bar https://review.opendev.org/c/zuul/zuul/+/803866 | 20:46 | |
| -@gerrit:opendev.org- Zuul merged on behalf of Simon Westphahl: [zuul/zuul] Fix node request cleanup routine https://review.opendev.org/c/zuul/zuul/+/809970 | 20:46 | |
| @jpew:matrix.org | Ah, it appears to respect the `USER` directive in my image Dockerfile if I use that.... that seems to do what I need | 20:47 |
| -@gerrit:opendev.org- Joshua Watt proposed: [zuul/nodepool] kubernetes, openshift: Add volume mount support https://review.opendev.org/c/zuul/nodepool/+/810322 | 21:53 | |
| -@gerrit:opendev.org- Zuul merged on behalf of Felix Edel: [zuul/zuul] Fix wrong if condition in job request state watch https://review.opendev.org/c/zuul/zuul/+/809629 | 21:55 | |
| @iwienand:matrix.org | when i "yarn install" it does a 5000 line diff to yarn.lock. am i missing something on how to just install what the gate does? | 22:04 |
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: [zuul/zuul] Temporarily enqueue cycle changes for reporting https://review.opendev.org/c/zuul/zuul/+/810328 | 22:22 | |
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed on behalf of Simon Westphahl: [zuul/zuul] wip: Make QueueItem a Zookeeper object https://review.opendev.org/c/zuul/zuul/+/809414 | 22:22 | |
| @clarkb:matrix.org | corvus: felixedel left a question on https://review.opendev.org/c/zuul/zuul/+/809410 if you have a chance for that | 22:24 |
| @jim:acmegating.com | swest: ^ i looked into what's breaking on that change, and found 2 things -- one is a quick fix with 810328, the other i left in a note on 809414. | 22:24 |
| @clarkb:matrix.org | corvus: in 810328 is it possible for the enqueued change to get processed out of band by another scheduler and merge (assuming the current scheduler context switches for a very long time maybe?) | 22:27 |
| @jim:acmegating.com | Clark, felixedel oh hrm, i missed that change in the PS revisions... yeah, i can't really see why zuul-web and fingergw are different in this case? | 22:28 |
| @jim:acmegating.com | i'd probably be inclined to let the cache happen on both | 22:29 |
| @clarkb:matrix.org | that was my thinking. It seems they both want to avoid watches | 22:29 |
| @jim:acmegating.com | or both have watches :) | 22:29 |
| @clarkb:matrix.org | so if we avoid on one we should avoid on the other. Ya I suppose if we don't care about the extra watches then that should be fine too | 22:29 |
| @clarkb:matrix.org | consistency seems like a good idea though | 22:29 |
| @jim:acmegating.com | just restarted opendev's zuul on master | 23:01 |
| @jim:acmegating.com | that's got a bunch of fixes, plus the change cache in zk | 23:02 |
| @jim:acmegating.com | Clark: all happens with the scheduler holding a lock on that pipeline; if it crashes between enqueue/dequeue the next scheduler should figure out it doesn't belong (missing dependencies) and dequeue it. | 23:17 |
| @jim:acmegating.com | sorry that was re the 810328 question which i missed earlier | 23:18 |
| @clarkb:matrix.org | corvus: thanks | 23:31 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!