| @mnasiadka:matrix.org | I’m back - a little bit confused what I’m doing in front of a keyboard :) | 07:57 |
|---|---|---|
| @gthiemonge:matrix.org | Hey, stupid question: someone has proposed a backport using the gerrit UI, the patch on master doesn't have the signed-off-by line with the name of the author (because he proposed the patch before we started using it), it only has my name because I finished the work. However the backport includes the signed-off-by line with the name of the author, which I found weird since he's never signed off the patch. Is it normal behavior in gerrit? shouldn't it only add a signed-off-by line with the name of the person who proposes the backport? https://review.opendev.org/c/openstack/octavia/+/1004617 | 09:04 |
| @clarkb:matrix.org | Gerrit doesn't add signed off by you need to do it manually. But Gerrit only checks that one of the committer or author is in the signed off by | 10:28 |
| @fungicide:matrix.org | gthiemonge: that is surprising behavior to me, makes me wonder if the cherry-pick button was really used there, warrants further testing but i agree if gerrit actually auto-added a signed-off-by trailer for someone who was not in the original commit message then that seems like a significant bug | 12:04 |
| @harbott.osism.tech:regio.chat | seems like IPv6 connectivity to gerrit is broken again. mnaser could you take another look? maybe you need to keep your ping of life running permanently? traffic from the host towards the outside seems not to be affected | 12:04 |
| @gthiemonge:matrix.org | Clark: fungi it looks like the uploader added this line, but they should have added their own name if i'm not mistaken https://review.opendev.org/c/openstack/octavia/+/1004617/1//COMMIT_MSG#15 | 12:09 |
| @fungicide:matrix.org | yes, they should have, but as Clark noted gerrit will accept it if at least one of the commit author or committer fields matches the signed-off-by address | 12:11 |
| @fungicide:matrix.org | if anyone else happens to be wondering about `retry_limit` results between 20:00 and 04:00 utc, it looks like something caused our mirror in ovh gra1 to be unable to read from some afs volumes which caused package installs to fail with `No package matching '...' is available` errors, but whatever it was subsided eventually | 13:30 |
| @fungicide:matrix.org | i also see that the configure-mirrors role's "Update apt cache" task doesn't fail when it encounters an error fetching indices, which was the earliest clear sign of a problem in those builds but wasn't highlighted because the task ended in "Changed" state and the build continued on to later playbooks | 13:33 |
| @fungicide:matrix.org | apparently `apt-get update` exits 0 in that circumstance, i guess because it falls back on older locally cached copies | 13:35 |
| @fungicide:matrix.org | according to apt-get's manpage, we may want to consider adding `-eany, --error-on=any` "Fail the update command if any error occured, even a transient one." | 13:36 |
| -@gerrit:opendev.org- Jeremy Stanley https://matrix.to/#/@fungicide:matrix.org proposed wip: [zuul/zuul-jobs] 1004809: configure-mirrors: Fail apt-get update on errors https://review.opendev.org/c/zuul/zuul-jobs/+/1004809 | 13:49 | |
| @fungicide:matrix.org | uploaded as wip ^ so we can test with base-test first, of course | 13:49 |
| @fungicide:matrix.org | stephenfin: ^ that's followup to what we were discussing in #openstack-infra irc | 13:50 |
| @fungicide:matrix.org | in theory that would have stopped the job sooner and highlighted the real problem more clearly too | 13:50 |
| -@gerrit:opendev.org- Jeremy Stanley https://matrix.to/#/@fungicide:matrix.org proposed wip: [zuul/zuul-jobs] 1004809: configure-mirrors: Fail apt update on errors https://review.opendev.org/c/zuul/zuul-jobs/+/1004809 | 15:20 | |
| -@gerrit:opendev.org- Jeremy Stanley https://matrix.to/#/@fungicide:matrix.org proposed: [zuul/zuul-jobs] 1004873: configure-mirrors: Test fail apt update on errors https://review.opendev.org/c/zuul/zuul-jobs/+/1004873 | 15:20 | |
| -@gerrit:opendev.org- Zuul merged on behalf of Stephen Finucane: [openstack/project-config] 988094: Add nova-reviewers to openstack/nova ACL https://review.opendev.org/c/openstack/project-config/+/988094 | 15:48 | |
| @harbott.osism.tech:regio.chat | fungi: I think e.g. https://zuul.opendev.org/t/zuul/build/4b7ca988d6964e579315831de2dcd14a/console already proves that the added option has no bad effect on the happy path. IMO with that you could squash the two changes back into one, since I don't think this would actually change what the base-test job does, but I also don't think that that is necessary. or would you want to verify what happens in case of an actual mirror error? IMO we can do that once one shows up again in production after merging the change | 16:40 |
| @fungicide:matrix.org | i'm not sure waiting for a reproducible condition in the wild makes sense, no | 16:41 |
| @fungicide:matrix.org | my main concern is that past changes to the configure-mirrors role have broken most/all jobs if we don't take the extra step of exercising them directly through base-test | 16:42 |
| @fungicide:matrix.org | but yes https://zuul.opendev.org/t/zuul/build/4b7ca988d6964e579315831de2dcd14a/console#2/6/7/ubuntu-noble seems to indicate it's normally fine at least | 16:43 |
| @fungicide:matrix.org | so i'd be fine recombining them | 16:44 |
| @fungicide:matrix.org | https://zuul.opendev.org/t/zuul/build/4b7ca988d6964e579315831de2dcd14a/console#2/7/7/ubuntu-noble too, so it's exercised a couple of times in the job | 16:45 |
| -@gerrit:opendev.org- Jeremy Stanley https://matrix.to/#/@fungicide:matrix.org proposed: [zuul/zuul-jobs] 1004809: configure-mirrors: Fail apt-get update on errors https://review.opendev.org/c/zuul/zuul-jobs/+/1004809 | 16:48 | |
| @fungicide:matrix.org | welcome, Dan With! long time no see | 17:11 |
| @dan_with:matrix.org | Hey @fungi. I didn’t realize the Matrix switchover. I have sent a few messages into IRC blackhole. | 17:11 |
| @dan_with:matrix.org | How do I add other channels? openstack-cinder, etc? | 17:12 |
| @fungicide:matrix.org | Dan With: oh, if you're asking about connecting to irc channels with a matrix client, there is a bridge that presents them as matrix rooms... the matrix room name to join would be like `#_oftc_#openstack-cinder:matrix.org` | 17:15 |
| @harbott.osism.tech:regio.chat | just to be clear: other channels (like openstack ones) are still running in IRC. accessing them via a matrix bridge is possible, but does have some issues, I'd prefer to stick to native connections if possible | 18:22 |
| @fungicide:matrix.org | yes, i do it in reverse personally, and use a matrix plugin in my irc client to join matrix rooms and treat them like they're irc channels | 18:25 |
| @dan_with:matrix.org | best IRC client for you? I am trying to get wired in. I have had problems with getting people to respond in IRC before, that's why I thought that everyone had switched to Matrix. | 18:26 |
| @fungicide:matrix.org | so for me this is essentially the `matrix::server:opendev.org:#opendev` room | 18:26 |
| @fungicide:matrix.org | Dan With: my use is somewhat esoteric, but i run a debian vm (in my personal account on rackspace flex at the moment, as it happens) with the weechat-curses client package installed, and then hand-add the weechat-matrix-rs plugin (built from source) | 18:28 |
| @fungicide:matrix.org | and i run it under a tmux session which i attach to from a local terminal over mosh | 18:29 |
| @dan_with:matrix.org | I have had my own bouncer for many years, but I no longer have the patience for this :) | 18:29 |
| @fungicide:matrix.org | i also use a weechat plugin called weeslack to make slack channels work like they're irc channels | 18:29 |
| @fungicide:matrix.org | Dan With: yes, and that's why connecting to irc channels with a matrix client has become fairly popular | 18:30 |
| @fungicide:matrix.org | no need for a persistent bouncer, among other conveniences | 18:30 |
| @fungicide:matrix.org | but as Jens Harbott says, it's not without some downsides too | 18:30 |
| @dan_with:matrix.org | I can't figure out how to get the bridge working in Element. It may because I created a Matrix account called dan_with when I really needed to create oftc_dan_with? | 18:32 |
| @fungicide:matrix.org | folks in the zuul community wrote up something about it back when they moved from irc to matrix years ago: https://zuul-ci.org/docs/zuul/latest/howtos/matrix-irc.html | 18:33 |
| @fungicide:matrix.org | note the example in there really is just an example, since it's the #opendev channel on the oftc irc network, which we're obviously no longer using | 18:34 |
| @fungicide:matrix.org | also the screenshots are probably out of date since they were made with an earlier version of element | 18:35 |
| @fungicide:matrix.org | i need to go run an errand, probably 30 minutes or so, then i'll be back | 18:44 |
| @fungicide:matrix.org | okay, back now | 19:27 |
| @m00sie:matrix.org | I use https://thelounge.chat for a self hosted irc client | 19:36 |
| @fungicide:matrix.org | yeah, i've heard both it and rocket.chat are popular options in that vein | 19:43 |
| @dan_with:matrix.org | Got it working, thanks all | 22:19 |
| @fungicide:matrix.org | Dan With: cool | 23:32 |
| @fungicide:matrix.org | hopefully i can pick your brain tomorrow about a federated api key problem with one of our flex accounts, but i'll save that for tomorrow | 23:33 |
| @fungicide:matrix.org | it's getting dark over here so i'm shifting to night mode | 23:33 |
Generated by irclog2html.py 4.1.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!