Tuesday, 2026-06-02

opendevreviewmelanie witt proposed openstack/tempest master: Wait for volume "in-use" in test_encrypted_cinder_volumes_resize_revert  https://review.opendev.org/c/openstack/tempest/+/99099900:25
elodillesnote that recent upper-constraints bump ( https://review.opendev.org/c/openstack/requirements/+/990790 ) broke stable/2025.1 (epoxy) ubuntu jammy based tempest jobs, as rpds-py 2026.5.1 version removed py310 support, which is jammy's default python version :/14:26
elodilleswhat is the nice approach to fix this?14:27
clarkbis this usually when we switch to a tagged tempest release?14:49
elodillesclarkb: not yet. the 2025.1-last tag should be introduced when 2025.1 is not maintained anymore (goes to Unmaintained or EOL)14:52
clarkbelodilles: but tempest will tag early when master is no longer compatible with stable branches so that stable branches can use the tags14:55
clarkb(since tempest is not branched)14:55
mnasiadkaelodilles: should we add rpds-py version with python version constraint?14:59
mnasiadkaBut another thing is we should have a mechanism that uses stable branch UC in stable branch tempest tests?15:00
elodillesclarkb mnasiadka : i'm not sure, hence i'm asking what is the practice here o:)15:00
elodillesmnasiadka: actually, tempest is installed in its own venv, with master constraints. this worked so far i think15:01
elodillesprobably so far, packages haven't been dropping old python versions so aggressively :/15:02
mnasiadkaMight be, I think the fastest solution is to fix that in requirements - but maybe tempest should start using stable u-c on stable branches15:03
mnasiadkaelodilles: something like https://review.opendev.org/c/openstack/requirements/+/991126 ?15:07
mnasiadkafrickler: ^^ fyi15:07
elodillesmnasiadka: ah, yes, you are right, that will be it! thanks o/15:08
fricklerno, we got rid of py3.10 constraints intentionally15:09
mnasiadkaYeah, that’s true - just trying to find a quick solution which we can revert once the correct solution lands15:10
frickleriiuc the solution is to use tempest 46.3.0, which is pinned to stable/2026.1 reqs15:11
mnasiadkaSo override-checkout in all projects? That’s a fun task15:12
mnasiadkaProposed this in neutron-t-p - let’s see if that helps: https://review.opendev.org/c/openstack/neutron-tempest-plugin/+/99112915:15
timburkei was expecting to be able to set TEMPEST_VENV_UPPER_CONSTRAINTS (as in https://review.opendev.org/c/openstack/swift/+/990959), but that didn't actually seem to do much (the job still wound up using master constraints)15:15
gmaanthere are some more place we need to pin it but pinning based on job is possible with TEMPEST_VENV_UPPER_CONSTRAINTS but that needs to be done in plugin side also like neutron-tempest-plugin did15:16
gmaanelodilles: mnasiadka frickler if requirement change to have py3.10 constraints is not acceptable then I need to drop the py3.10 support from tempest and release a new tag15:18
gmaanand use that tag for the 2025.1 jammy testing. that is proper solution for that and we did in past15:19
gmaanlet me start the work for that today15:19
elodillesgmaan: ah, good to know. i did not remember that15:19
gmaanbefore releasing the tempest, we need to do some pre-work on tempest side on tox.ini constraints. I will propose those after my morning meetings15:23
mnasiadkaThanks gmaan15:24
opendevreviewMaor Blaustein proposed openstack/devstack master: Devstack OVN support for resource_provider_* configuration  https://review.opendev.org/c/openstack/devstack/+/99113615:25
gmaanwe usually keep old python version support in tempest as long as they are working. once they are broken (mostly due to constraints ) then we start that version support drop process, like this case.15:25
timburkeagain, i don't think TEMPEST_VENV_UPPER_CONSTRAINTS actually works for this -- it causes UPPER_CONSTRAINTS_FILE and TOX_CONSTRAINTS_FILE to get exported during run-devstack, but we seem to have a fresh environment when we later get to run-tempest, so it still uses constraints from master15:28
gmaanyeah, that is coming from the tempest role run-tempest, we pin the constraints threre but that is per branchs not per job/distro15:31
sean-k-mooneythe requireemt repo can technnly expre per python verion requireemnt so you coudl pin rpds-py to the prvious version fo py 3.1017:49
sean-k-mooneybut your aslo corerct that on master we should not need to do that17:49
fricklerthe reqs repo no longer runs py3.10 checks, so we would really like to avoid that. we could revert the rpds-py bump for a week if that would be really needed17:52
sean-k-mooneyya im not a fan of that solution eithe rfor what its worth17:53
sean-k-mooneythe per stable branch version fo the tempet job on master could just point to an older requirements file17:53
sean-k-mooneyanyways im sure gmaan's got this17:54
sean-k-mooneyit sound like its not the first time this has happended17:55
gmaanyeah, this is usually happen when we move the min supported python version17:55
opendevreviewGhanshyam Maan proposed openstack/tempest master: Drop py3.10 from Tempest  https://review.opendev.org/c/openstack/tempest/+/99116117:57
sean-k-mooneygmaan: so regarding 2025.1 jammy was only included in https://github.com/openstack/governance/blob/master/reference/runtimes/2025.1.rst as part fo the addtional testing for smoth upgrade so i woudl expect this mainly to have impacted the greade jobs correct18:01
sean-k-mooneythere may have been some other jobs that had not moved to noble in some cases too18:01
gmaanyes, that is the only stable jobs failing. rest all on 2025.1 run on noble and green18:02
opendevreviewGhanshyam Maan proposed openstack/tempest master: Set upper-constraints if TEMPEST_VENV_UPPER_CONSTRAINTS is set  https://review.opendev.org/c/openstack/tempest/+/99118019:10
opendevreviewGhanshyam Maan proposed openstack/whitebox-tempest-plugin master: Drop python 3.10  https://review.opendev.org/c/openstack/whitebox-tempest-plugin/+/99118319:19

Generated by irclog2html.py 4.1.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!