Monday, 2026-02-23

tkajinamstephenfin, hi. it'd be nice if you have time to review https://review.opendev.org/c/openstack/oslo.privsep/+/977534 . I'm honestly not very confident about the change in oslo_privsep/tests/unit/test_daemon.py and am wondering if you have a better idea ?10:34
opendevreviewTakashi Kajinami proposed openstack/oslo.metrics master: Fix broken log format  https://review.opendev.org/c/openstack/oslo.metrics/+/97761010:44
PavloKostianov[m]Hi everyone! I want to share I have recently built and released to allow developers patch multiple repositories with the same change at the same time. If you need to apply same change on all  or even just more than 1 oslo repo it will help.11:35
PavloKostianov[m]Here is the PyPI page: https://pypi.org/project/changes-roller/11:35
PavloKostianov[m]Feel free to open new issues with bugs/issue requests on the GH repo: https://github.com/k-pavlo/changes-roller11:35
PavloKostianov[m]*a tool11:35
opendevreviewMerged openstack/oslo.metrics master: Fix broken log format  https://review.opendev.org/c/openstack/oslo.metrics/+/97761013:11
opendevreviewMerged openstack/oslo.privsep master: Fix mypy errors  https://review.opendev.org/c/openstack/oslo.privsep/+/97753413:33
opendevreviewMerged openstack/oslo.privsep master: Replace usage of undocumented logging.WARN  https://review.opendev.org/c/openstack/oslo.privsep/+/97753513:33
tkajinamAttributeError: Can't pickle local object '_ConfigFileOpt.__init__.<locals>.<lambda>'14:18
tkajinamhm14:18
tkajinamdamani[m] hberaud[m]  I'm a bit confused now seeing the above error in heat. I don't think this is quite specific to heat and we expect any conf instance, loaded by oslo.config may encounter the same14:21
tkajinamI mean, I guess https://review.opendev.org/c/openstack/oslo.service/+/965208 may never work actually and 'fork' context may be always used... Is that what you expected ?14:24
tkajinamgmaan, you might be interested in ^^^ , too14:26
opendevreviewTakashi Kajinami proposed openstack/castellan master: tox: Fix missing functional-vault-py310 env  https://review.opendev.org/c/openstack/castellan/+/97767814:36
opendevreviewTakashi Kajinami proposed openstack/castellan master: Move vault functional tests to noble  https://review.opendev.org/c/openstack/castellan/+/97768114:38
tkajinamidk why this has not been caught ... https://review.opendev.org/c/openstack/castellan/+/97767814:42
tkajinamprobably recent tox release revealed it14:42
opendevreviewTakashi Kajinami proposed openstack/castellan master: Bump vault used in tests  https://review.opendev.org/c/openstack/castellan/+/97513514:59
opendevreviewTakashi Kajinami proposed openstack/castellan master: Use more explicit import  https://review.opendev.org/c/openstack/castellan/+/97539914:59
opendevreviewTakashi Kajinami proposed openstack/castellan master: barbican: Drop redundant full url composition  https://review.opendev.org/c/openstack/castellan/+/97549714:59
opendevreviewTakashi Kajinami proposed openstack/castellan master: Check context instance type directly  https://review.opendev.org/c/openstack/castellan/+/97537514:59
opendevreviewTakashi Kajinami proposed openstack/castellan master: barbican: Pass down domain/system scope  https://review.opendev.org/c/openstack/castellan/+/97537614:59
fricklertkajinam: isn't that what https://review.opendev.org/c/openstack/oslo.service/+/977241 is supposed to fix? 15:01
tkajinamfrickler, yes it "fixed" but eventually all services use fork context15:01
tkajinamthat _select_service_manager_context was added to use spawn context where it can be used, but spawn context may be never used due to conf instance not picklable15:03
opendevreviewMerged openstack/castellan master: tox: Fix missing functional-vault-py310 env  https://review.opendev.org/c/openstack/castellan/+/97767815:44
opendevreviewTakashi Kajinami proposed openstack/oslo.cache master: Fix remaining access to [cache] memcache_username/password  https://review.opendev.org/c/openstack/oslo.cache/+/97768715:46
opendevreviewTakashi Kajinami proposed openstack/castellan stable/2025.2: tox: Fix missing functional-vault-py310 env  https://review.opendev.org/c/openstack/castellan/+/97768815:52
opendevreviewMerged openstack/castellan master: Move vault functional tests to noble  https://review.opendev.org/c/openstack/castellan/+/97768115:56
cardoetkajinam: https://review.opendev.org/c/openstack/oslo.utils/+/977568 you mind looking at that since I know you're really good at looking at technical Python-y things?15:57
opendevreviewMerged openstack/castellan master: Use more explicit import  https://review.opendev.org/c/openstack/castellan/+/97539916:08
opendevreviewMerged openstack/castellan master: Bump vault used in tests  https://review.opendev.org/c/openstack/castellan/+/97513516:10
gmaantkajinam: yes that change fixed it and release 4.5.1 will be ok. and yes until oslo.cofig is not picklable, services will be using 'fork' method which is same as before. to start using spawn, oslo.config as well services must be picklable16:11
opendevreviewMerged openstack/futurist master: Remove MANIFEST.in  https://review.opendev.org/c/openstack/futurist/+/97598016:12
gmaanralonsoh: you mentioned about nova cpu also in this, is there any failureI missed to check https://review.opendev.org/c/openstack/requirements/+/97769116:17
opendevreviewTakashi Kajinami proposed openstack/oslo.utils master: Refactor force_reraise  https://review.opendev.org/c/openstack/oslo.utils/+/97769216:26
tkajinamcardoe, I'll recheck it once CI post results but it looks good to me... though I'm wondering if we can refactor the method a bit ^^^ to avoid duplicating same error message16:27
tkajinamgmaan, ok so that's what needs to be addressed during next cycle, right ?16:29
tkajinamI just want to make sure that won't be abandoned in the current incomplete status with that noisy warning always appearing.16:29
gmaantkajinam: I am not sure as someone should be working on that. as you know 'fork' is deprecated one and might be removed in future. spawn is not going to be default so it will not break us at least in near future16:31
gmaanbut if spawn can work then this is the end goal for threading mdoe16:31
gmaantkajinam: I think damani[m] idea was to provide the interface to use spawn and projects start fixing and using it. I do not think it was 100% workable in current implementation of services and oslo.config16:32
opendevreviewMerged openstack/castellan master: Check context instance type directly  https://review.opendev.org/c/openstack/castellan/+/97537516:32
opendevreviewMerged openstack/castellan master: barbican: Pass down domain/system scope  https://review.opendev.org/c/openstack/castellan/+/97537616:32
opendevreviewMerged openstack/castellan master: barbican: Drop redundant full url composition  https://review.opendev.org/c/openstack/castellan/+/97549716:32
tkajinamok16:33
gmaanwe just need to make sure oslo.service does not use the spawn if servies are not yet ready16:33
gmaanand that was the bug where oslo.service was selecting spawn even service instance and oslo.config used in that service instance was not ready16:33
tkajinamI probably have to talk with damani[m] about his next plan.16:35
tkajinamhonestly speaking I feel like we've been doing too many break-and-fix-partially stuff regarding threading for some time and I'm a bit lost about the goal16:36
tkajinamthat fallback, effectively voiding the context selection, is one of these16:37
opendevreviewMerged openstack/oslo.cache master: Fix remaining access to [cache] memcache_username/password  https://review.opendev.org/c/openstack/oslo.cache/+/97768720:24
*** Callum0270 is now known as Callum02720:42

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