| *** mazzy50988129295808 is now known as mazzy509881292958 | 01:23 | |
| opendevreview | OpenStack Proposal Bot proposed openstack/project-config master: Normalize projects.yaml https://review.opendev.org/c/openstack/project-config/+/827938 | 02:52 |
|---|---|---|
| opendevreview | Merged openstack/project-config master: Normalize projects.yaml https://review.opendev.org/c/openstack/project-config/+/827938 | 04:38 |
| *** chandankumar is now known as raukadah | 10:38 | |
| clarkb | fungi: should we propose a pbr release now? | 15:30 |
| fungi | clarkb: yep, it was on my agenda for the morning, i slept in a bit but coffee has been imbibed so i can work on that now | 15:30 |
| clarkb | no rush :) it is the weekend afterall. | 15:31 |
| fungi | there are now four commits since 5.8.0, none add features or change dependencies, so 5.8.1 is probably in order. do we want to add any release notes about that last change? | 15:33 |
| fungi | the other changes are basically just testing cleanups or fixing warnings | 15:40 |
| fungi | this is the only regression fix | 15:40 |
| fungi | might not be a bad idea to let users know the packages made with 5.8.0 were missing pbr's metadata | 15:43 |
| Clark[m] | Does PBR do Reno? | 16:05 |
| Clark[m] | I suppose we could add a note to explain the bug and rebuilding packages if necessary | 16:05 |
| fungi | it does, and yes i'm about to push it now | 16:07 |
| fungi | Clark[m]: pushed now as 827977 | 16:09 |
| Clark[m] | Sorry morning distracted me. Looking soon | 17:56 |
| clarkb | approved | 17:59 |
| fungi | thanks! once it merges i'll push up a release request | 18:07 |
| frickler | does the dco-license job need any special parameters or is it just broken? https://zuul.opendev.org/t/openstack/build/b33dc881e16148dca061bacf5545a3a1 | 19:00 |
| fungi | good question... i doubt it's been used since the kata tenant was torn down | 19:21 |
| fungi | i'm not sure we have examples left of how it was being employed | 19:22 |
| fungi | frickler: note that job was, as far as i know, only every used to check dco footers in github pull request commits, because doing so required a paid feature (at least at the time that was written). for a project in gerrit, you don't need a zuul job. there's an actual gerrit feature to require signed-off-by footers, just set that in the acl | 19:42 |
| fungi | frickler: https://review.opendev.org/Documentation/config-project-config.html#receive-section | 19:51 |
| fungi | more specifically, https://review.opendev.org/Documentation/config-project-config.html#receive.requireSignedOffBy | 19:51 |
| corvus | i'd like to restart zuul, any objections? | 20:23 |
| corvus | a rolling restart -- though there will be a database schema upgrade which could make things exciting | 20:24 |
| corvus | (i mean sql db) | 20:24 |
| fungi | corvus: sounds good to me, thanks! | 20:26 |
| fungi | i'm around if database schema upgrades get more fun than anticipated (which is to say, fun at all) | 20:27 |
| corvus | okay, i'm going to start with a stop/start of the mergers and graceful restarts of executors 6 at a time. i'm trying out a new playbook that i'll upload if this works. | 20:28 |
| fungi | ooh | 20:28 |
| corvus | okay, zm* are hard-stopped (because we haven't implemented graceful for them yet) and ze01-ze06 are gracefully stopping | 20:29 |
| fungi | hard-stopping mergers is relatively low-impact because the merge request would simply get unlocked in zk and become available for another one to process, right? | 20:33 |
| corvus | fungi: i think if we hard-stop them, they can report failure | 20:47 |
| fungi | oh, got it | 21:00 |
| fungi | as in "this change or one of its dependencies cannot be merged" | 21:00 |
| corvus | yep our old friend | 21:03 |
| corvus | ze05 has stopped, so it looks like my scheme to wait for containers stopping in the playbook is working | 21:05 |
| fungi | oh awesome | 21:07 |
| fungi | what's the improvement in the new playbook? | 21:07 |
| fungi | it actually lets you know when things have stopped? | 21:07 |
| corvus | well, i'm starting with a playbook that does graceful restarts of executors since we don't have that anywhere yet | 21:17 |
| corvus | we're also going to need a way to deal with schedulers and their initialization period -- i think the best way to do that is to start up the prometheus server to access the ready probe | 21:18 |
| corvus | but i don't think we have that configured, so i'll probably punt that for later | 21:18 |
| Clark[m] | For the mergers it should be straightforward to add a graceful stop to them too. I may look at that next week | 21:21 |
| Clark[m] | Set a flag to stop. Check flag before next request. | 21:22 |
| opendevreview | Gage Hugo proposed openstack/project-config master: Retire security-specs repo - Step 3 https://review.opendev.org/c/openstack/project-config/+/827178 | 21:36 |
| corvus | Clark: yeah, but most of that's already done because they do support pause | 22:06 |
| corvus | you can use the executor for reference, it's basically "go into pause mode, then exit when jobs complete" | 22:06 |
| corvus | woo, the playbook has moved on to the 2nd batch of 6 executors | 22:14 |
| fungi | righteous | 22:37 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!