Friday, 2023-01-06

*** ysandeep is now known as ysandeep|ruck01:30
opendevreviewLuke Odom proposed openstack/diskimage-builder master: Map curl to curl-minimal for rocky 9
*** ysandeep|ruck is now known as ysandeep|ruck|afk04:14
opendevreviewLuke Odom proposed openstack/diskimage-builder master: Add swap support
opendevreviewLuke Odom proposed openstack/diskimage-builder master: Add swap support
*** akahat|rover is now known as akahat05:12
*** pojadhav- is now known as pojadhav05:40
*** ysandeep|ruck|afk is now known as ysandeep|ruck05:52
*** marios is now known as marios|rover06:04
*** ysandeep|ruck is now known as ysandeep|ruck|lunch08:24
*** ysandeep|ruck|lunch is now known as ysandeep|ruck10:03
*** ysandeep__ is now known as ysandeep|ruck10:12
*** rlandy|out is now known as rlandy11:12
*** pojadhav is now known as pojadhav|sick11:17
*** dviroel|afk is now known as dviroel11:24
*** ysandeep|ruck is now known as ysandeep|ruck|afk12:50
*** dasm|off is now known as dasm14:14
*** ysandeep|ruck|afk is now known as ysandeep|ruck14:18
fricklerslaweq: there's still a held node for 865822, I assume that it can be removed now?14:23
fricklerykarel: same question for 867609, are the held nodes still needed?14:24
ykarelfrickler, no not needed, sorry i forgot about them 14:24
fricklerykarel: np, will clean up, thx14:29
ykarelfrickler, also slaweq is on PTO today, but considering the lib/neutron patches merged and root cause was vexxhost nodes for other case too, should be safe to cleanup14:33
fricklerit's not urgent, I'll follow up next week14:33
ykarelfrickler, okk14:34
ykarelfrickler, btw you recall issue with cirros 0.6.1 and host-passthrough?14:35
ykarelseeing in
fricklerykarel: I think there were similar issues when I tried to bump cirros for devstack, but I'd need to look closer to be sure it's the same thing14:39
ykarelfrickler, k Thanks14:42
ykarelwill try without host-passthrough and see how it goes14:47
*** ysandeep|ruck is now known as ysandeep|out15:01
clarkbykarel: frickler: host-passthrough is going to vary by cloud because each cloud presents us with a different cpu model16:31
clarkbI would avoid using host passthrough in CI for this reason if consistency is needed. IIRC devstack defaults to some core 2 like cpu now?16:31
*** dviroel is now known as dviroel|lunch16:39
opendevreviewLuke Odom proposed openstack/diskimage-builder master: Add swap support
*** marios|rover is now known as marios|out16:44
clarkbThey can even vary within a single provider which created problems for the initial implementation of live migration testing16:52
clarkbfungi: if you have time for thats a pretty straightforward nox conversion for bindep and the parent change fixes a small bug in bindep too16:54
clarkbNot necessarily looking to land the nox change right away either, just hoping for eyes on the conversions so that others can see what they look like and provide feedback if they have any16:59
fungiclarkb: on a related note, stephenfin has a couple of pbr fixes related to tox v4 which need looking at (both passing tests now): 869190 and 86908217:01
clarkboh cool I reviewed some of those yesterday let me pull the stack up again17:01
clarkbfungi: stephenfin: worries me beacuse new pbr needs to continue working with old distros and setuptools. Is there a reason we are dropping that testing?17:03
clarkbI think we can add to the list just fine we we should be very careful removing support for old systems in pbr as it we can't easily constrain it17:03
stephenfinclarkb: Yeah, those versions of pip aren't supported on python3.10 (because of the collections -> move)17:04
stephenfinso if we want to test those, we need to go back to testing on 20.0417:04
stephenfin...or pull in additional python versions on 22.0417:04
stephenfinyou'll see the failed jobs for the previous PS'17:04
clarkbI think we should run the job on 20.04 and 22.04 in that case personally. Just because it is super easy to make a change to master pbr and break all the old stable openstack for example17:04
stephenfinThat's fair. I don't really want to have to do that myself though. I've already burned this week and much of last fixing tox 4 issues :) Could we merge this as-is and retrospectively re-add those tests?17:06
clarkbI mean I'm ok from it for all of the software I run as none of it runs on old python at this point I don't think17:07
clarkbbut it is a very real risk to stable openstack.17:07
fungiwhat's involved in running it more than once? just job variants for each nodeset we want?17:08
clarkbfungi: and the test code there would need to select test cases based on the underlying python version17:09
fungiahh, right17:09
fungiso a bit of additional complexity in the test selection17:09
stephenfinyeah, skipif decorators17:09
clarkbstephenfin: re tox I'm on the stop doing anything tox related full steam ahead on nox train. It takes time too but feels like less of a treadmill17:10
clarkbI guess the alternative for those old platforms is to update pip and setuptools to newer versions than the distro ships with if we were to break them somehow17:14
clarkbor fix the testing at that point in time17:14
clarkbthe testscenarnios readme has a hacky simple way of doing this. Let me try and push that up as a followup and if that looks good we can land the whole bunch17:21
fungiclarkb: i left you some comments on 868004, but in particular any idea why we don't get the subunit2html report on the nox-based unit test jobs? specifically we don't appear to collect testr_results.html nor testrepository.subunit17:26
clarkbfungi: I think that was a bug in zuul-jobs that has since been fixed17:27
clarkbfungi: a recheck should confirm17:27
fungiahh, will do. thanks!17:27
clarkbthe tooling to collect that info looked for a tox venv to generate the content17:27
clarkbnow it looks for a nox venv too17:27
fungimake sense17:27
fungier, makes sense17:27
clarkboh that change will need to sort out the opendev-nox-docs stuff. I'm always super confused by those jobs and corvus set things up for zuul. Maybe corvus can point me in the right direction17:38
clarkbchances are low I'll get to that today, and even if I did not sure I'd trust myself with the results. Next week however I should start there17:40
*** dviroel|lunch is now known as dviroel17:50
*** dviroel is now known as dviroel|pto21:12
clarkbfungi: it does look like the rechecked jobs on the bindep nox change include those log files now so the fix I wrote to zuul-jobs was sufficient just a timing thing21:17
clarkbFor OpenDev I've put on my todo list for monday getting a nox-opendev-docs job sorted out then we can look at porting things properly for other bits. I would like to move opendev just to get past the failure cycle train21:21
*** dasm is now known as dasm|off21:37

Generated by 2.17.3 by Marius Gedminas - find it at!