| -@gerrit:opendev.org- Maksim Malchuk proposed: [openstack/diskimage-builder] 975122: Fix issue with insecure HTTPS repository https://review.opendev.org/c/openstack/diskimage-builder/+/975122 | 12:18 | |
| @harbott.osism.tech:regio.chat | it was mentioned in the cinder channel that it seems like gerrit stops accepting updates to a change once there have been 1000 of them, is that a known thing? I'm assuming the number is the sum of number of patchsets, comments and reviews? see https://review.opendev.org/c/openstack/cinder/+/926844 | 15:41 |
|---|---|---|
| @clarkb:matrix.org | Jens Harbott: yes I think this has been the case since we upgraded to Gerrit 3.2 (or maybe 2.16) as part of the notedb transition. Let me see if I can find the relevant config option | 15:43 |
| @clarkb:matrix.org | Jens Harbott: https://gerrit-review.googlesource.com/Documentation/config-gerrit.html#change `change.maxPatchSets` is `1000` by default as is `change.maxUpdates`. I suspect `maxUpdates` is the one playing a role here | 15:45 |
| @clarkb:matrix.org | The documentation indicates this is due to performance reasons with notedb and a quick grep against system-config indicates we never set a new value so we should be using that default value | 15:47 |
| @clarkb:matrix.org | This is the first time I can recall it becoming an issue (not bad for ~4 years?) | 15:48 |
| @mordred:waterwanders.com | wow. 1000 revisions on a change. that deserves a cookie | 15:59 |
| @clarkb:matrix.org | mordred: looks like it is just less than 100 patchsts but there are almost 500 comments and I wonder if votes count as a distinct thing as an update? | 16:03 |
| @clarkb:matrix.org | but ya its a long lived change with a lot of updates | 16:03 |
| @fungicide:matrix.org | cinder changes certainly do get flooded with comments from (mostly broken) third-party ci systems | 16:06 |
| @fungicide:matrix.org | maybe they'd like us to block some of those accounts? | 16:06 |
| @clarkb:matrix.org | oh that is an interesting idea. Yes, I wonder if all of those comments are relevant | 16:09 |
| @fungicide:matrix.org | i would not be surprised if half of them have been broken zombies on autopilot for a decade | 16:12 |
| @fungicide:matrix.org | and their owners forgot they exist | 16:12 |
| @clarkb:matrix.org | it does look like a good number of them simply report failures | 16:24 |
| @clarkb:matrix.org | I agree a good next step here is probably for the cinder team to determine if any of those third party ci accounts need to be disabled to avoid spamming the changes with irrelevant noise | 16:25 |
| @clarkb:matrix.org | I think we should do that before we consider bumping any limits particularly since gerrit has set these values for performance reasons | 16:25 |
| @fungicide:matrix.org | agreed | 16:27 |
| @clarkb:matrix.org | I can put a note about this on the meeting agenda for communications purposes. I'll also add a note about my update all the containers to trixie changes from Friday. Is there anything else we want to see get added/edited/removed on the agenda? | 16:47 |
| -@gerrit:opendev.org- Jeremy Stanley https://matrix.to/#/@fungicide:matrix.org marked as active: [opendev/system-config] 974942: Inject docs honeypot URL into body comment https://review.opendev.org/c/opendev/system-config/+/974942 | 16:49 | |
| @fungicide:matrix.org | possible next step in the bot wars ^ | 16:49 |
| @clarkb:matrix.org | I'm willing to try it. It is an easy revert if we need to do that | 16:51 |
| @clarkb:matrix.org | I can link to that in the agenda topic as well | 16:51 |
| @clarkb:matrix.org | for the trixie updates I think most are straightforward. I was worried about the gitea one due to a change in sshd packaging which apparently creates /run/sshd/ for us now but then I realized the push of opendev/system-config into gitea in testing is using ssh so that is covered and should be fine. The main thing is the updates to the irc bots due to that deprecated ssl wrapper function going away requiring actual updates to the bots | 16:57 |
| @clarkb:matrix.org | We probably want to land https://review.opendev.org/c/opendev/system-config/+/975319 first to ensure we've updated all of the base container images before we start updating the consumers. | 17:03 |
| @clarkb:matrix.org | ok meeting agenda has its first update | 18:04 |
| -@gerrit:opendev.org- Clark Boylan proposed: [opendev/base-jobs] 975444: Add an early base job check for CPU counts https://review.opendev.org/c/opendev/base-jobs/+/975444 | 18:24 | |
| @clarkb:matrix.org | fungi: ^ that is a quick chagne for the cpu count thing that was discussed in #openstack-infra | 18:24 |
| @clarkb:matrix.org | corvus: do you know of anywhere where we are using the new smaller labels? I think ensuring we cover 975444 with those using base-test would be good | 19:10 |
| -@gerrit:opendev.org- Zuul merged on behalf of Clark Boylan: [opendev/base-jobs] 975444: Add an early base job check for CPU counts https://review.opendev.org/c/opendev/base-jobs/+/975444 | 19:11 | |
| @jim:acmegating.com | Clark: i can't think of any. maybe check codesearch | 19:14 |
| @jim:acmegating.com | may not be used yet | 19:14 |
| @clarkb:matrix.org | https://opendev.org/openstack/puppet-openstack-integration/src/branch/master/zuul.d/libraries.yaml#L4-L12 this is the only use I find in codesearch | 19:15 |
| @clarkb:matrix.org | so not unused but barely used | 19:15 |
| @clarkb:matrix.org | I'll work on proposing a WIP change to that repo now to test it | 19:15 |
| @clarkb:matrix.org | and while I figure that out I have rechecekd https://review.opendev.org/c/zuul/zuul-jobs/+/680178 which is already set up to do base-test testing against zuul-jobs own unittests | 19:16 |
| @clarkb:matrix.org | seems to work generally | 19:23 |
| @clarkb:matrix.org | `remote: https://review.opendev.org/c/openstack/puppet-openstack-integration/+/975455 DNM Testing some zuul base job stuff` | 19:25 |
| @clarkb:matrix.org | * seems to work generally: https://zuul.opendev.org/t/zuul/build/cef87de483934ac98d00449e49fa9eae/console#0/3/7/ubuntu-jammy | 19:26 |
| @fungicide:matrix.org | agreed, looks safe | 19:29 |
| @clarkb:matrix.org | The test jobs against puppet Openstack failed but not in the pre playbook so I think it looks good there too. Do we want a change to apply this to the other base jobs? | 19:46 |
| @fungicide:matrix.org | i think so, and then we can talk about it in the meeting i guess | 19:55 |
| @clarkb:matrix.org | I'll get that up after lunch and update the meeting agenda | 20:01 |
| -@gerrit:opendev.org- Clark Boylan proposed: [opendev/base-jobs] 975474: Apply the cpu count hardware check in our base job https://review.opendev.org/c/opendev/base-jobs/+/975474 | 21:00 | |
| @clarkb:matrix.org | looking at these gitea backup errors I think the filesystem backup (so not the db stream backup) is exiting 1 due to files (in the most recent case a journal file) disappearing during the backup. It looks like older borg (like on gitea) may also exit 1 when this happens and treats it as a warning based on the output `terminating with warning status, rc 1` but our script only treats it as a warning if we're running newer borg | 21:15 |
| @clarkb:matrix.org | any objection to me dropping the borg version check treating an rc 1 as a warning rather than an error? Or would we prefer to get these emails as htey can be an indicator of problems (thus the warning) | 21:16 |
| @fungicide:matrix.org | if we were ignoring them before, i'm not too worried about getting them now | 21:23 |
| @clarkb:matrix.org | we are only ignoring them on noble and newer with the newer borg version | 21:26 |
| @clarkb:matrix.org | I guess the version on jammy is also new enough to have this behavior | 21:27 |
Generated by irclog2html.py 4.0.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!