| opendevreview | Merged openstack/osprofiler master: Replace deprecated constant_time_compare https://review.opendev.org/c/openstack/osprofiler/+/931151 | 02:30 |
|---|---|---|
| opendevreview | Takashi Kajinami proposed openstack/osprofiler master: pre-commit: Bump version https://review.opendev.org/c/openstack/osprofiler/+/935591 | 02:50 |
| opendevreview | Takashi Kajinami proposed openstack/osprofiler master: Remove Python 3.8 support https://review.opendev.org/c/openstack/osprofiler/+/935592 | 02:52 |
| opendevreview | Takashi Kajinami proposed openstack/osprofiler master: Drop remaining pin of elasticsearch https://review.opendev.org/c/openstack/osprofiler/+/935593 | 02:54 |
| opendevreview | Takashi Kajinami proposed openstack/osprofiler master: pre-commit: Bump version https://review.opendev.org/c/openstack/osprofiler/+/935591 | 02:54 |
| opendevreview | Takashi Kajinami proposed openstack/osprofiler master: Drop redundant if-else block https://review.opendev.org/c/openstack/osprofiler/+/935594 | 02:59 |
| zigo | hberaud[m]: Hi there! I'm getting this rebuilding Nova (Dalmatian): | 07:45 |
| zigo | https://paste.opendev.org/show/b93dLmkywwVJDu0RIClo/ | 07:45 |
| zigo | This looks like an issue in oslo.service under Python 3.13, no? | 07:45 |
| hberaud[m] | looks like | 07:46 |
| zigo | It feels like the issue is in the LOG.info() | 07:47 |
| zigo | Should I patch it to use an f'' string? | 07:47 |
| zigo | What's your suggestion? | 07:47 |
| hberaud[m] | well, if it come from log, maybe we have a problem with https://opendev.org/openstack/oslo.log/src/branch/master/oslo_log/pipe_mutex.py | 07:48 |
| hberaud[m] | is your env a monkey patched env? | 07:48 |
| zigo | My env is whatever nova test discovery with stestr does, I believe it is monkey patched. | 07:49 |
| hberaud[m] | ok | 07:49 |
| zigo | (not sure during test discovery though...) | 07:49 |
| zigo | It's Debian Unstable, so Py 3.13 with packages and Dalmatian. | 07:49 |
| hberaud[m] | I'm on the way to release eventlet 0.38.0 https://github.com/eventlet/eventlet/issues/991 | 07:49 |
| hberaud[m] | and this new version of eventlet introduce the support of python 3.13 | 07:49 |
| zigo | Will this work with Dalmatian?!? | 07:50 |
| hberaud[m] | https://github.com/eventlet/eventlet/commit/4d48d10b990910cc87ec6bbeeaea63b295fe314d | 07:50 |
| zigo | Any breaking change? | 07:50 |
| zigo | I have added a few Py 3.13 patches already though. | 07:50 |
| hberaud[m] | Dalmatian do not support 3.13 | 07:50 |
| zigo | In Debian, it will ! :) | 07:51 |
| hberaud[m] | ah | 07:51 |
| zigo | In Eventlet 0.36.0 in Debian, I have these: | 07:51 |
| zigo | https://salsa.debian.org/openstack-team/third-party/python-eventlet/-/blob/debian/dalmatian/debian/patches/python3.13-support.patch?ref_type=heads | 07:51 |
| zigo | and | 07:51 |
| zigo | https://salsa.debian.org/openstack-team/third-party/python-eventlet/-/blob/debian/dalmatian/debian/patches/Spew_Correct_line_lookup_from_inspect.getsourcelines.patch?ref_type=heads | 07:51 |
| hberaud[m] | I see | 07:51 |
| zigo | You've merged Stefano Rivera's patches, right? :) | 07:52 |
| zigo | Anyways, my question remain: do you know any breaking change in Eventlet 0.38.0 that would break Dalmatian? | 07:52 |
| hberaud[m] | yes | 07:52 |
| hberaud[m] | no | 07:53 |
| hberaud[m] | the python 3.13 was the main breaking change | 07:53 |
| hberaud[m] | (patch) | 07:53 |
| zigo | Ok, I may try upgrading it and re-run tempest then, thanks for your input. | 07:53 |
| hberaud[m] | because it changed the attributes we monkey patch with threading and greenlet | 07:53 |
| zigo | Back to this PipeMutex: I don't understand what's going on... | 07:54 |
| hberaud[m] | in parallel of that my colleague Itamar also faced a problem with that yesterday https://github.com/eventlet/eventlet/issues/874#issuecomment-2483425977 | 07:54 |
| hberaud[m] | maybe you can ask him more details about his context | 07:54 |
| zigo | Is he on IRC ? | 07:55 |
| hberaud[m] | yes but on libera | 07:55 |
| zigo | Oh... in what channel? :) | 07:55 |
| hberaud[m] | libra/libera I don't remember the right spelling | 07:55 |
| zigo | libera is right, I believe. | 07:56 |
| zigo | I'm on both OFTC and Libera. | 07:56 |
| hberaud[m] | at itamarst | 07:56 |
| hberaud[m] | IIRC | 07:56 |
| hberaud[m] | IMO that's not a coincidence if you both faced pipemutex more or less at the same time | 07:57 |
| hberaud[m] | (pipemutex error) | 07:57 |
| zigo | I believe I was able to /query with him. | 07:58 |
| hberaud[m] | else fell free to reply on his latest comment on the eventlet issue | 07:58 |
| zigo | Ok. | 07:58 |
| zigo | Thanks a lot for your input. | 07:58 |
| hberaud[m] | no problem | 07:58 |
| hberaud[m] | you are welcome | 07:59 |
| hberaud[m] | @zigo one last thing, I don't know if you are aware or not of that, but we created the #openstack-eventlet-removal channel, feel free to also join us there | 08:00 |
| zigo | Done! | 08:01 |
| zigo | Thanks for leading this much much needed effort. | 08:01 |
| zigo | FYI, we tried using uwsgi for swift and ... managed to undertand it was completely broken. | 08:01 |
| zigo | Hum... moving this discution to that channel. | 08:02 |
| *** elodilles_pto is now known as elodilles | 08:03 | |
| zigo | For this: | 08:07 |
| zigo | TypeError: 'PipeMutex' object does not support the context manager protocol | 08:07 |
| zigo | issue, it doesn't appear when running oslo.log unit tests, only when attempting to run Nova unit tests. | 08:07 |
| hberaud[m] | ok | 08:30 |
| opendevreview | Merged openstack/osprofiler master: Drop remaining pin of elasticsearch https://review.opendev.org/c/openstack/osprofiler/+/935593 | 09:35 |
| opendevreview | Artem Goncharov proposed openstack/pbr master: add python 3.12 support https://review.opendev.org/c/openstack/pbr/+/924216 | 09:38 |
| opendevreview | Takashi Kajinami proposed openstack/oslo.log master: Add missing contextmanager interfaces to PipeMutex https://review.opendev.org/c/openstack/oslo.log/+/935620 | 10:00 |
| tkajinam | hberaud[m], zigo ^^^ | 10:00 |
| hberaud[m] | 👍️ | 10:01 |
| tkajinam | it may not appear in unit tests because PipeMutex is part of the dirty hack for eventlet | 10:02 |
| zigo | Oh, thanks ! :) | 10:04 |
| zigo | Trying immediately if that fixes my Nova build. | 10:04 |
| damani[m] | hi | 10:05 |
| hberaud[m] | o/ | 10:05 |
| damani[m] | testing matrix :) | 10:06 |
| damani[m] | seems working and nice | 10:06 |
| frickler | zigo: for neutron and eventlet 0.37.0, see https://bugs.launchpad.net/neutron/+bug/2088453 | 10:08 |
| zigo | frickler: Thanks, though right now, I'm still with 0.36.0 + some Py 3.13 patches. | 10:09 |
| * zigo bookmarks the bug | 10:09 | |
| zigo | tkajinam: I confirm that your patch repairs Nova's test discovery ! :) | 10:12 |
| zigo | 25 failures under Py 3.13 still ... :/ | 10:16 |
| hberaud[m] | zigo: 0.38.0 is now out with the python 3.13 patch embedded | 10:37 |
| hberaud[m] | I published it one or two hours ago | 10:37 |
| zigo | gr8 ! :) | 10:50 |
| zigo | I'm on nova unit test failures with py3.13 right now. | 10:50 |
| damani[m] | gmann, can you please take a look https://review.opendev.org/c/opendev/irc-meetings/+/934648?usp=dashboard | 10:51 |
| zigo | Most flavor tests are failing with the mocking of get_flavor_by_id seemingly broken. | 10:51 |
| zigo | https://bugs.launchpad.net/nova/+bug/2088999 | 10:51 |
| frickler | oh, now I see where the damani nick comes from :-D | 11:29 |
| damani[m] | frickler, nice :) | 11:30 |
| damani[m] | not everyone found it | 11:32 |
| opendevreview | Takashi Kajinami proposed openstack/osprofiler master: Drop redundant if-else block https://review.opendev.org/c/openstack/osprofiler/+/935594 | 14:08 |
| gmann | damani[m]: done, thanks for handling it | 18:10 |
| damani[m] | gmann, thanks a lot | 19:10 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!