| opendevreview | Maksim Malchuk proposed openstack/diskimage-builder master: Revert "Fallback to persistent netifs names with systemd" https://review.opendev.org/c/openstack/diskimage-builder/+/838863 | 12:53 |
|---|---|---|
| opendevreview | Merged openstack/diskimage-builder master: Update gentoo python version to 3.9 https://review.opendev.org/c/openstack/diskimage-builder/+/836237 | 13:56 |
| opendevreview | Dr. Jens Harbott proposed openstack/diskimage-builder master: Test building jammy https://review.opendev.org/c/openstack/diskimage-builder/+/836228 | 14:51 |
| clarkb | TheJulia: rpittau stevebaker[m] fyi https://review.opendev.org/c/openstack/diskimage-builder/+/838863 reverts a change that ya'll needed/wanted. I left a comment on a possible fix that doesn't ened to be reverted | 15:59 |
| rpittau | clarkb: I'm pretty sure I tested the sdandard way and it didn't work | 16:09 |
| clarkb | rpittau: well at least on centos 9 stream matching "move" in addition to "add" saw the events | 16:10 |
| clarkb | none of that is super well documented but if you dig around in the kernel enough eventually you find it emitting the move action for those changes | 16:11 |
| rpittau | well considering that if the revert is merged we'll have to find a way or our CI will be inevitably broken, again | 16:19 |
| rpittau | I guess we'll have to dig very deep | 16:19 |
| clarkb | rpittau: do you consume dib from source or from releases? But ya I think breaking everything but bullseye is a good reason to revert | 16:33 |
| rpittau | clarkb: we consume DIB from source | 16:39 |
| rpittau | AFAIK DIB supports only the current Debian stable, bullseye, not older releases, and that's not broken | 16:39 |
| rpittau | I haven't seen any breakage on ubuntu either, but probably because we lack coverage on that | 16:39 |
| clarkb | we use dib to build bsuter I would hope it continues to support oldstable too | 16:40 |
| clarkb | it continues to support centos 7 and 8-stream despite 9-stream being present etc | 16:40 |
| clarkb | and ya I think most of the testing uses glean | 16:40 |
| clarkb | (glean will default to dhcp as well fwiw) | 16:40 |
| rpittau | well centos 7 is mentioned as supported | 16:41 |
| rpittau | https://docs.openstack.org/diskimage-builder/latest/user_guide/supported_distros.html#testing-and-targets | 16:41 |
| clarkb | right I'm just saying that dib historically continues to support valid releases until the distro eols them | 16:41 |
| clarkb | that means buster is supported | 16:41 |
| rpittau | is buster broken with that patch? | 16:44 |
| clarkb | rpittau: that is what the person pushing the change has reported | 16:44 |
| clarkb | we use glean so our buster builds are ok | 16:44 |
| rpittau | I tested with buster too in our environment, it was working fine, and it was not using predictable interface names | 16:45 |
| clarkb | anyway I strongly suspect that if you update dhcp-all-interfaces-udev.rules as I updated the glean udev rules it may just work without needing to nuke the distro config | 16:46 |
| clarkb | I think that is preferable because users are reporting the nuking breaks them, but it also diverges from the distro further | 16:46 |
| clarkb | oh and it is backward compatible because older kernels just don't emit a move action | 16:48 |
| clarkb | they do the add and its fine | 16:48 |
| opendevreview | Dr. Jens Harbott proposed openstack/diskimage-builder master: Add a job to test building jammy https://review.opendev.org/c/openstack/diskimage-builder/+/836228 | 18:49 |
| stevebaker[m] | ianw: could you take a look at this review? It fixes overcloud-full being bootable when the kernel is updated during the image build | 22:56 |
| stevebaker[m] | ianw: https://review.opendev.org/c/openstack/diskimage-builder/+/838792 | 22:56 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!