Tuesday, 2023-01-03

*** mtreinish_ is now known as mtreinish00:59
*** pojadhav is now known as pojadhav|ruck04:33
*** ysandeep is now known as ysandeep|lunch09:10
*** pojadhav|ruck is now known as pojadhav|lunch09:37
*** pojadhav|lunch is now known as pojadhav|ruck10:22
opendevreviewMerged openstack/project-config master: Fix tox4 error  https://review.opendev.org/c/openstack/project-config/+/86852610:52
*** rlandy|out is now known as rlandy10:53
*** ysandeep|lunch is now known as ysandeep11:29
*** dviroel_ is now known as dviroel11:39
opendevreviewRafal Lewandowski proposed openstack/diskimage-builder master: Fix for cloud-user home being owned by root  https://review.opendev.org/c/openstack/diskimage-builder/+/86907712:06
stephenfinclarkb: fungi: Figured out the pbr-tox v4 incompatibility issue. It's this https://github.com/openstack/pbr/blob/master/pbr/core.py#L89-L9712:50
stephenfinbecause tox v4 isn't using a separate subprocess for each step, we end up calling into setuptools/pbr multiple times12:51
stephenfini.e. get_requires_for_build_sdist, prepare_metadata_for_build_wheel, build_sdist12:52
stephenfinit looks like the call to prepare_metadata_for_build_wheel launches us into that code, which sets in_use, and then the later call to build_sdist doesn't go through that path and we end up using the setuptools.command.sdist.sdist command instead of pbr.packaging.LocalSDist12:54
*** ysandeep is now known as ysandeep|brb12:55
stephenfinif I remove that, all the issues I reported in https://github.com/tox-dev/tox/issues/2712 go away12:55
stephenfinso we need a better canary than that12:57
fungistephenfin: ahh, yeah i added that in https://review.opendev.org/827931 (pbr 5.8.1)13:11
fungii guess we need another regression test reproducing the problem tox v4 exposed13:13
fungiso we can make sure we catch recursion13:14
stephenfinfungi: was there a bug report that led to that change or did you spot it in action?13:16
fungistephenfin: it was fixing a regression observed at https://meetings.opendev.org/irclogs/%23opendev/%23opendev.2022-02-04.log.html#t2022-02-04T10:00:02 which seemed to have been introduced by https://review.opendev.org/817794 "Allow PEP517 without setup_requires"13:26
fungibasically any sdists created by pbr 5.8.0 were lacking pbr.json files13:28
fungibasically sdists created between 2021-11-18 and 2022-02-0713:30
fungistephenfin: thankfully that didn't span an openstack coordinated release, but it did impact a number of libraries in yoga since it spanned from milestone 1 to a couple weeks before milestone 313:33
fungipbr 5.8.1 fixed it about a week before the non-client lib final release deadline for yoga13:36
stephenfinfungi: lmk what you think of https://review.opendev.org/c/openstack/pbr/+/869082. I'm not yet sure how to test that without providing our own pep-517 implementation or adding a dependency of pyproject-api13:36
*** ysandeep|brb is now known as ysandeep13:37
fungistephenfin: yeah, that does seem challenging to test. maybe clarkb will have a suggestion once he wakes up/is around13:40
fungii agree we'd basically need some instrumentation to make multiple pbr calls in one process13:41
fungistephenfin: -1 that's more of a -0.5 on that change, just some cleanup you missed13:44
gthiemongeHi Folks, how can we hold a node for troubleshooting CI issues?13:52
fungigthiemonge: let us know the job name, the project that job is failing on, and optionally the change you're rechecking to reproduce the issue. also need a brief synopsis of what you're investigating so we can include it in the comment field for the hold for future reference13:56
gthiemongefungi: ok thanks! the change is https://review.opendev.org/c/openstack/octavia/+/868765 , the job: openstack-tox-functional-py39 (it fails quikcly, we can recheck it at any time)13:58
fungii'll get it set in a moment13:58
gthiemongefungi: we are still investigating the issue we discussed yesterday in the Octavia jobs: "tox.tox_env.python.api.NoInterpreter: could not find python interpreter matching any of the specs python3.10"13:59
fungigthiemonge: and you're not able to reproduce it locally i guess?14:02
gthiemongefungi: nop14:02
fungigthiemonge: did you try removing the basepython=python3 in the general [testenv] section? that's fixed it elsewhere14:03
gthiemongefungi: yes, tweining did it in patchset 914:04
tweiningI tried it in a jammy vm with tox 4.1.2 and it worked fine14:05
*** dasm|off is now known as dasm14:06
fungigthiemonge: tweining: autohold is set. recheck 868765 and once the openstack-tox-functional-py39 job fails let me know what ssh key(s) you want to be able to log into the node with14:07
Clark[m]stephenfin fungi I don't have any great ideas right now. I'm also not feeling well which means the idea machine is running poorly.14:10
Clark[m]tweining: did you run python3.9 and not 3.10 on your VM?14:10
fungiClark[m]: i've had some of my best ideas on cold meds ;)14:10
tweiningClark[m]: 3.1014:12
Clark[m]tweining that job is for 3.9 and failing to find 3.10. I don't think that reproduction is valid14:15
tweiningClark[m]: thanks. I'll try with 3.9.14:16
*** ysandeep is now known as ysandeep|brb14:22
tweiningmmh, the standard jammy repo seem to only have py3.10 and py3.1114:24
Clark[m]Oh right 3.9 jobs should run on focal. It's possible your jobs have been running under python3.10 due to use of basepython=3 and ignoring conflicting versions14:25
Clark[m]This has likely been the case since the default nodeset was changed to jammy14:26
tweiningI installed python3.9 from another repo, created a py3.9 venv and installed tox there. it still works with that14:29
*** ysandeep|brb is now known as ysandeep14:47
gthiemongefungi: Ok I think we have found something: removing basepython from the functional-py* section seems to work (but not the basepython from [testenv])15:16
gthiemongefungi: I will propose a new change, can we remove the hold?15:16
*** dviroel is now known as dviroel|lunch15:32
fungigthiemonge: yep, removed now15:37
fricklerthat's just what I said yesterday, maybe I should have been more explicit about what I meant, sorry for that15:41
opendevreviewJeremy Stanley proposed opendev/system-config master: Feature our cloud donors on opendev.org  https://review.opendev.org/c/opendev/system-config/+/86909116:25
opendevreviewJeremy Stanley proposed opendev/system-config master: Feature our cloud donors on opendev.org  https://review.opendev.org/c/opendev/system-config/+/86909116:26
*** marios is now known as marios|out16:32
*** ysandeep is now known as ysandeep|brb16:34
*** ysandeep|brb is now known as ysandeep16:42
*** dviroel|lunch is now known as dviroel16:45
*** ysandeep is now known as ysandeep|out17:19
fungiamorin: probably not urgent, but today we received notification that "the payment could not be taken for the invoice with the ID FR54172080"20:51
fungii can't recall, who are we supposed to reach out to in order to get the voucher renewed?20:51
*** dviroel is now known as dviroe|out21:17
fricklerbtw the issue with https://vexxhost.com/.well-known/openstack/api seems to have been resolved, maybe mnaser has an rca for us? 21:48
*** rlandy is now known as rlandy|out23:38

Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!