Redeploy the Forgejo Runnerhost VM with latest changes #473

Closed
opened 2026-03-25 09:40:02 +00:00 by dkirwan · 2 comments
Member

Summary

Redeploy the Forgejo Runnerhost VM with latest changes

Background

As we now have more of the Runnerhost installation/configuration automated, its time to reinstall and test these changes end to end.

Details

As we now have more of the Runnerhost installation/configuration automated, its time to reinstall and test these changes end to end.

Summary

Redeploy the Forgejo Runnerhost VM with latest changes

### Summary Redeploy the Forgejo Runnerhost VM with latest changes ### Background As we now have more of the Runnerhost installation/configuration automated, its time to reinstall and test these changes end to end. ### Details As we now have more of the Runnerhost installation/configuration automated, its time to reinstall and test these changes end to end. ### Summary Redeploy the Forgejo Runnerhost VM with latest changes
dkirwan added this to the Sprint 17 project 2026-03-25 09:40:03 +00:00
Member

The VM creation and runner services are now fully automated.
We tested the changes end-to-end tearing down the staging forgejo-runnerhost-vm and deploying it via infra/ansible forgejo playbook.
Confirmed that the VM gets created, podman socket gets set, ansible-pull clones the ansible-role-forgejo-runner and runs it, which installs a watcher service that watches the oc secret with runner configs, fetches them on the VM and runs runner systemd services accordingly.

Related PRs:

infra/ansible@baef7ad4e7
infra/ansible@3febfa7233
infra/ansible@d11d12061a
infra/ansible@cde0bea197
infra/ansible@98625936dc

Next step: do the same for production.
Will need to announce outage.

The VM creation and runner services are now fully automated. We tested the changes end-to-end tearing down the staging forgejo-runnerhost-vm and deploying it via infra/ansible forgejo playbook. Confirmed that the VM gets created, podman socket gets set, ansible-pull clones the ansible-role-forgejo-runner and runs it, which installs a watcher service that watches the oc secret with runner configs, fetches them on the VM and runs runner systemd services accordingly. Related PRs: https://forge.fedoraproject.org/infra/ansible/commit/baef7ad4e7fdf1bbe8d08f37a70a83c8c8669617 https://forge.fedoraproject.org/infra/ansible/commit/3febfa7233eec39617dbcfe5c68a032fb2c56434 https://forge.fedoraproject.org/infra/ansible/commit/d11d12061a30953b94d7faeabdb8000ac51c0aa8 https://forge.fedoraproject.org/infra/ansible/commit/cde0bea197b0a5fe0fb97017fe1724308dd27e12 https://forge.fedoraproject.org/infra/ansible/commit/98625936dc664404cea02a496dad146810b339ef Next step: do the same for production. Will need to announce outage.
Member

Runners on production are now up and running with the refactored automation.

During the deployment we discovered a discrepancy between the staging and production virtualization setting in openshift => production did not have the volume snapshots available.
We found the problem and fixed that. Volume snapshots are now available for all projects on production openshift, not only forgejo.

This makes it more a 5 point issue than 2 point.

Runners on production are now up and running with the refactored automation. During the deployment we discovered a discrepancy between the staging and production virtualization setting in openshift => production did not have the volume snapshots available. We found the problem and fixed that. Volume snapshots are now available for all projects on production openshift, not only forgejo. This makes it more a 5 point issue than 2 point.
lenkaseg 2026-04-23 13:52:33 +00:00
Sign in to join this conversation.
No milestone
No project
2 participants
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
forge/forge#473
No description provided.