Yes, exactly.

You eventually can delete all containers on a recently upgraded control plane and re-setup them, which should prevent the outage. You can also use the playbook [1] to mark all backends on this node as DOWN with gracefully draining traffic from them.
But still, you'd need to check the upgrade guide docs for some tips, as while deleting containers you still need to find out which controller is considered as "main" one, in order to execute playbooks with slightly different flags based on that.

Eventually, it is also possible to execute `do-release-upgrade` inside of the containers, and upgrade each container OS independently. But as for me - doing such an upgrade is a waste of time and OS reinstall is simpler and faster in case all customizations are automated as well.


[1] https://opendev.org/openstack/openstack-ansible-ops/src/branch/master/ansible_tools/playbooks/set-haproxy-backends-state.yml 

чт, 24 апр. 2025 г. в 11:42, Rambo Rambo <ram.ramb2412@gmail.com>:
Hi Dmitriy

Thanks for your response.
I am now concerned on the rabbitmq and galera LXC containers.
We have 3 control nodes and on each we have galera and rabbitmq LXC containers.

After OS upgrade is done on the control nodes.
So can we destroy and recreated them one at a time to prevent outage?

control-01 rabbitmq-1 -->
control-02 rabbitmq-2-->
control-03 rabbitmq-3

Galera 1-->2--->3

Regards


On Wed, Apr 23, 2025 at 10:18 AM Dmitriy Rabotyagov <noonedeadpunk@gmail.com> wrote:

So lxc containers will remain intact on OS upgrades.
So pretty you will need to destroy the existing containers on the node and create the new ones.

You can destroy containers using a playbook containers-lxc-destroy.yml, with providing a proper --limit which will point either to a specific container or all containers on the lxc host.

Then they can be re-created with containers-lxc-create.yml playbook.

The guide I pointed to though, assumes OS re-setup rather then upgrade, as if you did not do manual changes or interventions, then re-setup is simply faster and easier to execute then an upgrade.


On Wed, 23 Apr 2025, 11:12 Rambo Rambo, <ram.ramb2412@gmail.com> wrote:
Hi Dmitriy

Thanks for the feedback.
We have Openstack services implemented as LXC containers on the control nodes.
So do we need to have special considerations for the LXC containers while doing the operating system upgrade following the shared procedure.

Regards
Rambo

On Wed, Apr 16, 2025 at 1:07 PM Dmitriy Rabotyagov <noonedeadpunk@gmail.com> wrote:
Hey,

Have you checked this guide?
https://docs.openstack.org/openstack-ansible/2023.1/admin/upgrades/distribution-upgrades.html

I'd guess it should mainly work, despite being written for Antelope (2023.1)

ср, 16 апр. 2025 г. в 13:59, Rambo Rambo <ram.ramb2412@gmail.com>:
>
> Hi Team
>
> We are currently on below release of Openstack:
>
> Openstack Release: Zed
> OS Release: Ubuntu 20.04.2 LTS
>
> I am looking for the tested procedure/suggestions/experience for doing the operating system upgrade from 20.04 to 22.04 on this openstack release.
>
> Regards
> Rambo
>