Thursday, 2025-10-23

*** dmellado9 is now known as dmellado04:03
*** dmellado0 is now known as dmellado04:11
*** dmellado8 is now known as dmellado04:46
*** dmellado3 is now known as dmellado05:46
*** jroll01 is now known as jroll008:55
opendevreviewMerged opendev/irc-meetings master: Update Cinder weekly team meeting location  https://review.opendev.org/c/opendev/irc-meetings/+/96417909:00
stephenfinclarkb: I usually get a load of coding done in downtime during summit, but it was too short this go round 😅 I think we need to add proper PEP-517 support so that we can drop our setup.py files. Before that though, we need to remove our last use of easy_install and pkg_resources. That's happening here https://review.opendev.org/c/openstack/pbr/+/96013309:34
stephenfinIn fact, we might get away without the pyproject.toml stuff for now, since it's explicitly the use of easy_install that setuptools is warning about09:34
mattcreesHi everyone, I'm trying to submit a patch to ansible-collections-openstack but I'm getting denied when running trying to set up `git review -s`: "mattcrees@review.opendev.org: Permission denied (publickey)." This is only happening on this project in particular, has anyone got any idea why this could be the case?10:31
fricklermattcrees: it is very unlikely for this to happen only for a specific repo. can you retry with a fresh clone and verify you are using your correct gerrit username?10:37
stephenfinfungi: I'm guessing the py27 jobs need to go in pbr? https://review.opendev.org/c/openstack/pbr/+/960133/ If so, is it time to formally drop py27 support?10:39
mattcreesStrangely is does still happen with a fresh clone. I'm definitely using the right username. My colleague just confirmed it works fine for him, probably something else wrong on my end... 10:40
fricklermattcrees: maybe run the command with "-v" added and show the (possibly redacted) output on paste.o.o?10:41
mattcreeshttps://paste.opendev.org/show/bWK50k0joHL6OW3ix4JK/10:45
tonybmattcrees: and your certain that ~/.ssh/opendev-ssh-key matches the key you set at https://review.opendev.org/settings/#SSHKeys10:47
fricklermattcrees: that looks like the issue is with the commit hook. I think I've seen that before, but don't remember the solution. what host OS are you using?10:48
frickleryou can also try to simple copy the file from some other repo10:48
mattcreesYes the key definitely matches. I'm using Ubuntu 24.04.310:49
mattcreesI tried a fresh checkout of sandbox, it now fails there too, so you're right it's not actually project specific 10:49
tonybwhat version of git review ? `git review --version`10:50
mattcrees2.3.110:50
tonybAh.  that's it10:51
tonybI'll look for the details10:51
fricklermattcrees: ah, right, you need a newer version10:52
fricklerhttps://review.opendev.org/c/opendev/git-review/+/910275 is the relevant change10:52
mattcreesInteresting, looks like I'm on the newest Ubuntu 24.04 has. I see I can install with pip instead, will give it a try10:54
tonybmattcrees: 2.4.0 or newer10:54
tonyb`pip install --user git-review` is probably safe?10:55
mattcreesThat's worked now, cheers guys :) 10:56
tonybmattcrees: See https://opendev.org/opendev/git-review/commit/d633541ecc5cafbf5660e0bf57ecbd96bfb77da0 for a few more details.10:56
tonybbasically OpenSSH deprecated scp10:56
mattcreesFunky10:57
fricklermattcrees: trixie has 2.4.0-1, maybe you want to open a bug against ubuntu to make them sync it10:57
fricklerah, updated in plucky https://launchpad.net/ubuntu/+source/git-review. also interesting list of bugs https://bugs.launchpad.net/ubuntu/+source/git-review10:59
tonybI bet they can all be closed :)11:02
stephenfinfungi: nvm, I figured it out11:50
*** amoralej__ is now known as amoralej12:01
mnasiadkafrickler: what is funny it seems Debian OpenStack (cc: zigo) is maintaining that, maybe there’s an option to ,,backport’’ that to Noble?12:41
fungimnasiadka: ubuntu has their "sru" (stable release update, i think) process for backporting fixes to existing ubuntu releases12:56
mnasiadkaRight, maybe they just don’t know it’s broken on Noble12:58
Clark[m]stephenfin yes I don't think dropping setup.py is necessary for this particular setup tools deprecation pass. We just need to get out of the things that are going away13:04
Clark[m]We should probably focus on the specific things needed to keep working in a week and not worry about setup.py until afterwards13:05
stephenfinright, then what I have up _should_ do the trick13:05
stephenfin(I basically just vendored everything /o\)13:05
Clark[m]Re python 2.7 we can still test it on newer platforms and the swift release which last supported python2.7 is like a year old at this point. I think we're getting close to being able to drop it but if we can keep 2.7 for now that is likely safest13:06
Clark[m]Ok so we should review the stack at 960133 and then get a release out soon?13:07
stephenfinI think so13:07
Clark[m]I can review that in a few hours. Have school related stuff this morning first13:07
*** dhill is now known as Guest2973114:19
fserucashi there, is there any problem with centos-stream mirror?15:14
fungifserucas: maybe start with describing what problem you're observing15:17
fungibut it looks like the centos-stream mirror volume ran out of quota around 13 days ago and hasn't been updating15:18
fungiare you seeing stale content?15:18
fungihttps://grafana.opendev.org/d/9871b26303/afs?orgId=1&from=now-30d&to=now&timezone=utc&viewPanel=panel-3615:18
fungii'll work on extending the quota now15:19
fungi#status log Increased the mirror.centos-stream AFS volume quota from 400000000 to 500000000 since it filled up around 2025-10-1015:25
fserucasis seems outdated for 13 days ago15:26
fserucasthank you15:27
fungiyes, i15:27
fungii'm manually running the update script now to catch it up15:28
fungilooks like our statusbot is also not in here any longer, i'll look into that next15:28
fungi#status log Increased the mirror.centos-stream AFS volume quota from 400000000 to 500000000 since it filled up around 2025-10-1015:30
opendevstatusfungi: finished logging15:34
stephenfinClark[m]: https://review.opendev.org/c/openstack/pbr/+/96471617:18
stephenfinjfyi17:18
stephenfinthat's as far as I've gotten17:18
stephenfin*got17:18
clarkback (sorry I'm on my irc client now for maximum confusion)17:20
clarkbI'm slowly reviewing the bottom of that stack. Want to diff the vendored code before I +2 that change17:20
clarkbstephenfin: in https://review.opendev.org/c/openstack/pbr/+/960029/3/pbr/_compat/commands.py a longer term followup may be to drop the localinstallscripts as a general thing anyway since pip apparently does that now without the inefficient setuptools imports and entrypoints17:23
opendevreviewJames E. Blair proposed opendev/system-config master: static: redirect zuulci.org to zuul-ci.org  https://review.opendev.org/c/opendev/system-config/+/96429617:27
clarkbstephenfin: I think I found a bug on https://review.opendev.org/c/openstack/pbr/+/96003117:34
clarkbstephenfin: ok I think I'm mostly caught up on reviewing that stack. I think things are mostly good until 960031 but did leave comments here and there on the whole stack if you wnt to read them before updating anything18:05
fungidid i happen to review those? it's been a while and i can't remember (trying to dig myself out from under paperwork and post-summit backlog still)18:09
clarkbfungi: no I'm the only reviewer on the latest versions I think18:12
fungiokay, i'll try to remember to get to those, thanks18:13
opendevreviewClark Boylan proposed zuul/zuul-jobs master: Find tox/nox siblings with normalized package names  https://review.opendev.org/c/zuul/zuul-jobs/+/96472418:17
clarkbinfra-root the openinfra ptg is next week. I'd like to put the meetpad servers in the emergency file to avoid any unexpected container updates to the service during the ptg. Any objections to me doing that now?18:27
fungino, please do!18:28
opendevreviewClark Boylan proposed zuul/zuul-jobs master: Find tox/nox siblings with normalized package names  https://review.opendev.org/c/zuul/zuul-jobs/+/96472418:29
clarkb#status log Added jvb02 and meetpad02 to the emergency file to avoid updates during the PTG18:31
opendevstatusclarkb: finished logging18:31
clarkbrosmaita: fungi I just tested meetpad recording in an up to date chrome installation and it captured both my audio and video18:36
clarkbit does look like you have to ensure that you tell it to capture teh correct tab but it was selected by default when the pop up to allow it came up18:36
rosmaitaclarkb: ack18:37
clarkbI guess the other thing to note is that I'm using X windows. I know that wayland impacts screen capturing which I think may eb related to recording as well18:37
clarkbI think the underlying mechanism for sharing your screen is very similar to the mechanism used to record the video18:37
rosmaitaok, i'll see if jbernard can run a test with me and try it out18:38
opendevreviewClark Boylan proposed zuul/zuul-jobs master: Find tox/nox siblings with normalized package names  https://review.opendev.org/c/zuul/zuul-jobs/+/96472418:44
clarkbthis is what i get for trying to think while jet lagged. Apologies for the noise on issues that shouldn't even happen in the first palce18:45
opendevreviewClark Boylan proposed opendev/system-config master: Force gitea http(s) connectivity through the load balancer  https://review.opendev.org/c/opendev/system-config/+/96472821:18
clarkbfeedback welcome this is something we discussed as I was trying to head to the summit. I think the change itself is straightforward but we should consider what the impacts to our ability to manage gitea rae and whether or not the tradeoff is worthwhile21:18
tonybclarkb: +2 from me22:02
tonybclarkb: Even with 964724, I'm seeing https://zuul.opendev.org/t/openstack/build/27b40f3997a0431499e5502e0188ddb4/log/job-output.txt#662-68022:10
tonybclarkb: Do you have time to look at it or should I?22:11
clarkbtonyb:  Ithink that is a separate issue with logging?22:43
clarkbtonyb: it assumes the venv for the env exists. Does that target maybe not use the same venv name?22:45
clarkbI think there is autodetection for that by reading the tox config but maybe something around that is broken?22:45
nhicher[m]tonyb, clarkb: I've proposed a review to use linux bridge instead openvswitch ansible module, could be good if you can review it https://review.opendev.org/c/zuul/zuul-jobs/+/959393 thanks =)22:46
tonybcan see that /home/zuul/src/opendev.org/openstack/requirements/.tox/py310-check-uc/ exists, but make the log subdir /home/zuul/src/opendev.org/openstack/requirements/.tox/py310-check-uc/log/ doesn't?22:47
tonybnhicher[m]: I did a while ago, but I don't have +2 there22:47
tonybclarkb: I guess I'm trying to figure a) what makes that dir, and b) what changed to stop it doing so22:48
clarkbtonyb: I think tox does (or did) create that log dir for its own logs22:50
clarkbtonyb: maybe tox is failing to do so due to a failure or it just logs elsewhere now? When we did the nox port for these roles we just stopped doing the logging to disk as nox writes everything to stdout/stderr iirc22:51
tonybHmm I'll look at nox as a comparison point22:52
clarkbit certainly looks like tox ran successfully without running any tests22:53
tonybYeah, there was a tox release but that was 3 weeks ago so I suspect we would have seen this befoer now22:54
clarkbhttps://opendev.org/openstack/requirements/src/branch/master/tox.ini I don't see any envlogdir or envdir configs where requirements is overriding things22:54
clarkbhttps://tox.wiki/en/latest/config.html#env_log_dir seems to imply that tox should still be creating that dir22:55
clarkbmaybe it is a permissions thing? though I woudl expect a different error I think for that22:55
clarkbtonyb: for debugging I think you can add a task between the steps that does an ls of the tree under .tox to see what is there22:56
clarkbtonyb: basically after tox without tests runs and before siblings happens22:56
tonybYeah.  I'll do that.22:56
clarkbnhicher[m]: I'll try to take a look tomorrow /me makes a note now22:58
clarkbwe can probably make the tox role more robust by trying to open the file in a try except and if it can't be opened then oh well emit a warning rather than an error situation23:01
nhicher[m]clarkb: sure, no rush, I've proposed it few weeks ago =)23:01
tonybclarkb: and/or add an os.makedirs(os.path.dirname(log_file)) or similar23:02
tonybIt's only really fatal because it's in a finally clause23:03
tonybIt's gotta be something specific to the requirements repo/check-uc toxenv or we'd be seeing red on every tox job right?23:23
clarkbthats what I'm wondering. I note that you have this set: https://opendev.org/openstack/requirements/src/branch/master/tox.ini#L16-L17 and maybe that causes it to skip enough work between no installs and no tests so it doesn't log anything to the log file?23:24
clarkbtonyb: in aprticular I think those log files may predominantly record the installation steps23:24
tonybAh, yeah I guess that could be an unexpected side-effect23:25
clarkbtonyb: https://zuul.opendev.org/t/openstack/build/9736bd34a5b64d32bb4c97b1408745b4/log/tox/functional/1-install_package.log here is an example from the change rosmaita was checking23:25
tonybThat's a solid lead23:26
clarkbtonyb: so maybe in that deps list instead of deps = empty string and no install you could test dropping the no install and setting deps = setuptools or something along those lines23:26
tonybthanks clarkb 23:26
clarkbto force it to do something noopy but just enough work to create logs?23:26
clarkbtonyb: I checked that glance job and tox doesn't seem to log when it creats the log dir so hard to narrow it down that way23:36
tonybclarkb: I'm confident that you're very close with that guess.  It's a fairly substantial difference from all the other tox jobs we run23:38
tonybclarkb: https://zuul.opendev.org/t/openstack/status?change=964731 hopefully that's helpful.  As we have several check jobs I tried a few different things23:39
clarkbcool I guess we wait and see now23:42
tonybYup we need to install *something*, now to make it the least possible23:45
tonybclarkb: I think you should look at https://review.opendev.org/c/opendev/system-config/+/963642 .. re docker not being present23:49
Clark[m]tonyb: aha thanks. I actually think that is better for us as we may be able to drop service disable if the package doesn't install at all. But we need to figure out the paths to the socket and if that works without docker installed. I'll take a look in the morning. Off to dinner now23:55
Clark[m]Oh also the command for compat so ya a few things to figure out before we do real cleanup23:56
tonybWe need /usr/bin/docker or all the ansible tasks where we use 'docker pull' or 'docker inspect' fail23:58
tonybI tried the "right" thing first and discovered the fallout23:59
tonybenjoy dinner23:59

Generated by irclog2html.py 4.0.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!