Upgrade problem exists from W->master where we go from runc to crun. Any changes to containers configs/systemd files causes the containers to be recreated/restarted and it hangs

[Bogdan Dobrelya]

There is an upgrade problem that exists from W->master where we go from runc to crun. Any changes to containers configs/systemd files causes the containers to be recreated/restarted and it hangs

[Bogdan Dobrelya (bogdando)]

IIUC, downstream there is no such issue, since there is only FFU paths remained? And since FFU goes from scratch - clean OS upgrade, what's the point to keep non-FFU upgrade jobs blocking upstream?

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.opendev.org/c/openstack/tripleo-ci/+/811720

[Bogdan Dobrelya (bogdando)]

requires a stable reproducer (it might be related to neutron/virtlogd wrappers only)

[Bogdan Dobrelya (bogdando)]

also we should not drop the UC upgrade job, since it will be useful for catching regressions like that

[Michele Baldessari (michele)]

https://review.opendev.org/c/openstack/tripleo-ansible/+/806481 is prolly what we want to start with, so at least we make sure we cannot change the runtime depending on what rpms get pulled in. (i.e. if crun is installed later podman version will use it as a default runtime)

[Marios Andreou (marios-b)]

@Bogdan - especially the undercloud upgrade we need to keep running.
For the FFU case, the undercloud is upgraded through 3 consecutive versions, i.e. we run undercloud upgrade 3 times. So the point to point upgrade needs to keep working in this case especially. Unless that has completely changed ;) but I'd ask the upgrades team to sign off on it - I think there will just have to be some workaround for this upgrade in particular where the runtime changes.

[OpenStack Infra (hudson-openstack)]

Change abandoned by "Bogdan Dobrelya <email address hidden>" on branch: master
Review: https://review.opendev.org/c/openstack/tripleo-ci/+/811720