| mnasiadka | fungi: did you manage to add that extra track? Can’t see it in https://ptg.opendev.org/etherpads.html | 10:05 |
|---|---|---|
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-systemd_networkd master: Restart systemd-udev on link changes https://review.opendev.org/c/openstack/ansible-role-systemd_networkd/+/954876 | 12:03 |
| fungi | mnasiadka: oh. sorry i missed your followup answer (i only have this channel to notify me on nick highlights and i had probably switched to another channel right after asking). i can add it now | 13:00 |
| mnasiadka | fungi: please do, thanks! :) | 13:01 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Set Rocky Linux distro jobs as NV https://review.opendev.org/c/openstack/openstack-ansible/+/964879 | 13:11 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Unfreeze roles after Flamingo Beta release https://review.opendev.org/c/openstack/openstack-ansible/+/963970 | 13:26 |
| noonedeadpunk | kind reminder - it's a ptg time now | 15:01 |
| noonedeadpunk | #link https://meetpad.opendev.org/oct2025-ptg-os-ansible | 15:01 |
| noonedeadpunk | jrosser: | 15:01 |
| noonedeadpunk | andrewbonney: | 15:03 |
| noonedeadpunk | mgariepy: | 15:03 |
| noonedeadpunk | damiandabrowski: | 15:03 |
| noonedeadpunk | NeilHanlon: :) | 15:04 |
| NeilHanlon | o/ thanks for ping | 15:05 |
| NeilHanlon | be there in a couple mins | 15:05 |
| mgariepy | sorry i do not have time today :/ | 15:05 |
| jrosser | andrew and i are double booked :( | 15:05 |
| noonedeadpunk | doh :( | 15:05 |
| jrosser | i think i have a timezone error - clocks changed yesterday | 15:06 |
| NeilHanlon | yea just ran into that on my end too heh | 15:09 |
| jrosser | i'm joined now | 15:46 |
| opendevreview | Merged openstack/openstack-ansible-os_keystone stable/2024.1: Fix package name for mod_auth_openidc https://review.opendev.org/c/openstack/openstack-ansible-os_keystone/+/961088 | 15:49 |
| opendevreview | Merged openstack/openstack-ansible-os_neutron master: Ensure neutron_use_uwsgi is treated a bool https://review.opendev.org/c/openstack/openstack-ansible-os_neutron/+/961340 | 17:29 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_neutron stable/2025.1: Ensure neutron_use_uwsgi is treated a bool https://review.opendev.org/c/openstack/openstack-ansible-os_neutron/+/964998 | 17:36 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_neutron stable/2024.2: Ensure neutron_use_uwsgi is treated a bool https://review.opendev.org/c/openstack/openstack-ansible-os_neutron/+/964999 | 17:37 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_neutron stable/2024.2: Ensure neutron_use_uwsgi is treated a bool https://review.opendev.org/c/openstack/openstack-ansible-os_neutron/+/964999 | 17:39 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_neutron stable/2024.1: Ensure neutron_use_uwsgi is treated a bool https://review.opendev.org/c/openstack/openstack-ansible-os_neutron/+/965000 | 17:39 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_neutron stable/2024.1: Ensure neutron_use_uwsgi is treated a bool https://review.opendev.org/c/openstack/openstack-ansible-os_neutron/+/965000 | 17:40 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Add debian trixie job definitions https://review.opendev.org/c/openstack/openstack-ansible/+/954616 | 17:44 |
| noonedeadpunk | well debian 13 fails on osa venv preparation: https://zuul.opendev.org/t/openstack/build/0c412322feb5444a841e689bf62aba64 | 18:59 |
| noonedeadpunk | but we're using some really weird pypi index | 19:00 |
| noonedeadpunk | --extra-index-url http://deb.debian.org/wheel/debian-13-x86_64 | 19:00 |
| jrosser | oh yes i saw that before | 19:00 |
| jrosser | that must come from NODEPOOL_WHEEL_MIRROR | 19:01 |
| noonedeadpunk | oh... could be.... | 19:01 |
| jrosser | this perhaps https://opendev.org/opendev/base-jobs/src/branch/master/roles/mirror-info/templates/mirror_info.sh.j2#L58 | 19:02 |
| jrosser | is it that there are not wheels built for trixie so it goes bad | 19:03 |
| noonedeadpunk | which kinda looks reasonable considering this: https://opendev.org/zuul/zuul-jobs/src/branch/master/roles/configure-mirrors/defaults/main.yaml#L11-L12 | 19:03 |
| noonedeadpunk | but I guess `mirror_fqdn` is not really on AFS | 19:03 |
| noonedeadpunk | fungi: don't you accidentally know anything about mirrors for Debian 13? | 19:03 |
| noonedeadpunk | jrosser: I think we just expect it to be infra mirrors, but it appers to be set to deb.debian.org instead | 19:04 |
| noonedeadpunk | which makes trixie to be unique | 19:05 |
| jrosser | yes i was just trying to understand where that comes from | 19:05 |
| jrosser | maybe its a fallback when there is not a deb mirror configured for an OS | 19:05 |
| fungi | we stopped building wheels for newer distros a while back because these days most stuff openstack depends on publishes usable wheels to pypi | 19:05 |
| jrosser | and thats leaked into the wheel mirror config | 19:05 |
| jrosser | trixie seems to be the first place thats blown up completely though, we get away with whatever config we had before on the other OS | 19:06 |
| noonedeadpunk | fungi: I think that trixie has `mirror_fqdn: deb.debian.org` | 19:07 |
| fungi | looking at https://static.opendev.org/mirror/wheel/ we didn't build wheels for debian 12 either | 19:07 |
| noonedeadpunk | in here: https://opendev.org/opendev/base-jobs/src/branch/master/roles/mirror-info/templates/mirror_info.sh.j2#L17 | 19:07 |
| noonedeadpunk | but I think we indeed might need to drop then wheels from infra mirrors | 19:08 |
| noonedeadpunk | I'm just thinking there could be more interesting things then, as there're also doesn't seem to be debian 13 infra mirrors at all? | 19:08 |
| fungi | not yet, no | 19:10 |
| jrosser | hmmm - so are we getting wrong info on the node in the case that a mirror does not exiat for a particular OS? | 19:11 |
| jrosser | *exist | 19:11 |
| fungi | doesn't look like anyone's proposed a change to opendev/system-config to include debian trixie in the reprepro config yet | 19:12 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Remove load_nodepool_pip_opts function https://review.opendev.org/c/openstack/openstack-ansible/+/965007 | 19:14 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Add debian trixie job definitions https://review.opendev.org/c/openstack/openstack-ansible/+/954616 | 19:14 |
| noonedeadpunk | ok, cool, thanks for the inptut! | 19:14 |
| noonedeadpunk | I guess I will look into adding mirrors then shortly | 19:14 |
| fungi | looking at the actual fatal error for https://zuul.opendev.org/t/openstack/build/0c412322feb5444a841e689bf62aba64 i think it's a problem with the pip version? | 19:17 |
| fungi | wheel mirror urls are a red herring, it should skip any that don't exist/404 | 19:17 |
| noonedeadpunk | well. yes. but I saw smth like that with some pip versions | 19:18 |
| fungi | oh, right it's the --index-url override | 19:18 |
| fungi | so it's not actually checking pypi at all, that explains it | 19:19 |
| noonedeadpunk | yeah, I wonder how it was working for noble though | 19:19 |
| noonedeadpunk | https://zuul.opendev.org/t/openstack/build/f856822eb2e046c281ea62ba7fd11c7d/log/job-output.txt#4450 | 19:19 |
| fungi | the play should output the pip install command it used, so ought to be easy to compare | 19:20 |
| noonedeadpunk | But given that at least some index exists, it could be just workarounding things | 19:20 |
| fungi | well, we don't publish a wheel mirror for noble either | 19:20 |
| noonedeadpunk | as it's passed as extra but at least index is there | 19:21 |
| fungi | oh, right it's going to http://mirror.bhs1.ovh.opendev.org/pypi/simple which is an unversioned proxy to pypi | 19:22 |
| fungi | so the problem is the trixie job is replacing the mirror hostname with deb.debian.org even though the mirrors host other stuff the job relies on besides just distro packages | 19:23 |
| noonedeadpunk | and with debian it tries just http://deb.debian.org which was confusing | 19:23 |
| noonedeadpunk | yeah | 19:23 |
| noonedeadpunk | and it replacing through sourcing /etc/ci/mirror_info.sh | 19:24 |
| noonedeadpunk | But this should be fixing it: https://review.opendev.org/c/openstack/openstack-ansible/+/965007/1/scripts/scripts-library.sh#b93 | 19:24 |
| noonedeadpunk | as it's still kinda pointless to carry | 19:24 |
| noonedeadpunk | crap.... | 19:26 |
| noonedeadpunk | there's still smth pulling in index... | 19:27 |
| noonedeadpunk | I wonder if this what also contributes t obreaking pip: https://zuul.opendev.org/t/openstack/build/0ed44df1ef9c407eb0146614630b42c3/log/job-output.txt#351-352 | 19:31 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Add debian trixie job definitions https://review.opendev.org/c/openstack/openstack-ansible/+/954616 | 19:33 |
| noonedeadpunk | fungi: I think it actually does contribute to the problem: https://opendev.org/zuul/zuul-jobs/src/branch/master/roles/configure-mirrors/templates/etc/pip.conf.j2 | 19:35 |
| noonedeadpunk | as obviously pypi_mirror resolves to deb.debian.org anyway | 19:36 |
| noonedeadpunk | and there's actually no way to avoid it from execution: https://opendev.org/zuul/zuul-jobs/src/branch/master/roles/configure-mirrors/tasks/main.yaml#L8 | 19:37 |
| fungi | any idea where mirror_fqdn is getting overridden to deb.debian.org? | 19:38 |
| fungi | doesn't look like it's being set in opendev/base-jobs | 19:39 |
| noonedeadpunk | I think it's coming from here somehow: https://opendev.org/zuul/zuul-jobs/src/branch/master/roles/configure-mirrors/defaults/main.yaml#L2 | 19:39 |
| noonedeadpunk | but where zuul_site_mirror_fqdn is coming from.... | 19:40 |
| noonedeadpunk | I'd guess it should be coming from here: https://opendev.org/openstack/project-config/src/branch/master/zuul/site-variables.yaml#L4-L9 | 19:41 |
| noonedeadpunk | oh | 19:42 |
| noonedeadpunk | /o\ | 19:42 |
| noonedeadpunk | ah, no, it's not contributing to the issue I guess. Just found this vars file we have, but it should be just reflecting what Zuul has: https://opendev.org/openstack/openstack-ansible/src/branch/master/zuul.d/playbooks/templates/osa-job-vars.yml.j2#L4 | 19:43 |
| noonedeadpunk | yup, pip.conf is totally screwed | 19:44 |
| noonedeadpunk | https://zuul.opendev.org/t/openstack/build/2e169242253440a6a899f69791031a3f/log/job-output.txt#4412-4418 | 19:44 |
| noonedeadpunk | The only thing I can propose right now, is to change condition on https://opendev.org/zuul/zuul-jobs/src/branch/master/roles/configure-mirrors/tasks/main.yaml#L8 | 19:45 |
| noonedeadpunk | to also include "mirror_fqdn | length > 0" | 19:45 |
| noonedeadpunk | and then we can likely set vars for the debian to be `mirror_fqdn: ""` | 19:45 |
| noonedeadpunk | or well, `zuul_site_mirror_fqdn: ""` | 19:46 |
| fungi | should be able to test that via depends-on | 19:47 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Add debian trixie job definitions https://review.opendev.org/c/openstack/openstack-ansible/+/954616 | 19:52 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Add debian trixie job definitions https://review.opendev.org/c/openstack/openstack-ansible/+/954616 | 20:15 |
| noonedeadpunk | ok, I've figured it out.... | 20:26 |
| noonedeadpunk | now we're progressing :) | 20:26 |
| noonedeadpunk | `msg: No package matching 'software-properties-common' is available` | 20:33 |
| noonedeadpunk | https://opendev.org/openstack/openstack-ansible-openstack_hosts/src/branch/master/vars/debian.yml#L68 | 20:34 |
Generated by irclog2html.py 4.0.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!