| jrosser | o/ morning | 08:47 |
|---|---|---|
| noonedeadpunk | o/ | 09:07 |
| noonedeadpunk | I've also realized that installation of dynamic-inventory as a console_script fails with python3.12 :( | 09:19 |
| noonedeadpunk | (https://zuul.opendev.org/t/openstack/build/2fcfe8c96fff4d7b91f6eba82e81bef8) | 09:20 |
| noonedeadpunk | as now it's not including the file as a whole, but just tries to import it, which fails as it's not installed | 09:20 |
| noonedeadpunk | jrosser: where should I be able to see capi logs for https://review.opendev.org/c/openstack/openstack-ansible-ops/+/936626 ? | 09:44 |
| noonedeadpunk | as either I look in the wrong place or they're not in CI results now? | 09:45 |
| jrosser | regarding CAPI logs, it only grabs them if it fails | 09:46 |
| jrosser | i was thinking about this and perhaps we should grab them always? | 09:46 |
| jrosser | i also did made some progress with the connection plugin + molecule, it failing because of apparmor on noble host, but not on jammy | 09:48 |
| noonedeadpunk | oh, I can recall you did some things ragarding lxc and apparmor when adding Noble at the first place | 10:02 |
| jrosser | yes that was inside the docker container | 10:13 |
| jrosser | but now this is an apparmor failure outside - centos9 container sshd does something when authing the user which causes an apparmor denied error, only on noble | 10:15 |
| jrosser | anyway i an a bit stuck with that now | 10:53 |
| jrosser | i've not found what actually is causing the apparmor error, though i suspect that if we mess with the PAM config in the centos container perhaps it can be worked around | 10:54 |
| jrosser | also disabling apparmor on the host works just fine, but we cant do that directly in the molecule ansible because in the zul molecule jobs, sudo is not allowed | 10:55 |
| jrosser | but that could go in a pre- playbook somewhere, before the sudo is revoked | 10:55 |
| jrosser | or we just run molecule/centos on jammy | 10:56 |
| jrosser | but all these things are kind of a bit hacky | 10:56 |
| noonedeadpunk | running on jammy is not really future-proof. we can run them on CentOS though? | 11:17 |
| noonedeadpunk | which make some sense if to think about it... | 11:18 |
| noonedeadpunk | ok, I need to sort out couple of things first... | 11:28 |
| noonedeadpunk | jrosser: sooo..... where logs are here? https://zuul.opendev.org/t/openstack/build/585acf56dac748eebc81f24989ca0f68/logs | 11:29 |
| jrosser | sorry -EMEETING, can look a bit later | 11:31 |
| noonedeadpunk | ++ | 11:47 |
| jrosser | oh well its actually broken here https://zuul.opendev.org/t/openstack/build/585acf56dac748eebc81f24989ca0f68/log/job-output.txt#11244 | 12:29 |
| jrosser | becasue i rebased my fork of magnum-cluster-api i think | 12:30 |
| jrosser | errr what https://github.com/vexxhost/ansible-collection-kubernetes/commit/471eaec113c9c1fbed68368cf8511e44d26e667a | 12:32 |
| noonedeadpunk | um | 12:33 |
| noonedeadpunk | sorry for that then | 12:33 |
| jrosser | not sure i understand how removing a default and leaving a var undefined makes things better | 12:33 |
| noonedeadpunk | given that it still accepts only "explicit" versions for which SHAs are defined? | 12:34 |
| noonedeadpunk | and all other defaults are still defined? | 12:34 |
| noonedeadpunk | yeah, me neither | 12:34 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-ops master: Define clusterctl_version explicitly https://review.opendev.org/c/openstack/openstack-ansible-ops/+/941114 | 14:49 |
| noonedeadpunk | jrosser: there this as well: https://github.com/vexxhost/ansible-collection-kubernetes/commit/1b51e526a7839a684a41b6878bef82d49a424462 | 16:33 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-ops master: Define clusterctl_version explicitly https://review.opendev.org/c/openstack/openstack-ansible-ops/+/941114 | 16:34 |
| jrosser | i guess this means we did break everyone by rebasing my branch | 16:44 |
| jrosser | doh | 16:44 |
| noonedeadpunk | yeah... | 16:44 |
| * noonedeadpunk now recalls why pinning to the branch is not good | 16:45 | |
| jrosser | i honestly did think that we could merge support for virtualenvs pretty quickly | 16:46 |
| jrosser | as it's really a non invasive change if they're not used | 16:46 |
| noonedeadpunk | it's indeed a weird thing to deal with | 16:48 |
| jrosser | mnaser: do you do any atmosphere deployments on noble yet? | 16:54 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible-ops master: Define clusterctl_version explicitly https://review.opendev.org/c/openstack/openstack-ansible-ops/+/941114 | 17:28 |
| noonedeadpunk | probably we should replace branch with SHA on stable branches and write a release note? | 18:56 |
| noonedeadpunk | release note for integrated repo... | 18:56 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!