The runner egress proxy has no PodDisruptionBudget #785

New issue

Open

opened 2026-08-11 23:35:12 +00:00 by coilyco-ops · 0 comments

coilyco-ops commented

2026-08-11 23:35:12 +00:00

Owner

kubectl get pdb -n forgejo returns no resources. The egress proxy runs two
replicas with no disruption protection, and both pods restarted 13h before this
was filed.

Today a proxy blip breaks whichever CI steps opted in. If the proxy becomes
always-on, it becomes a hard dependency for every job's egress, and a node
drain or rollout takes all CI with it.

Fix

Add a PodDisruptionBudget with minAvailable: 1. Consider whether the runners
themselves want one too, since the namespace has none at all.

`kubectl get pdb -n forgejo` returns no resources. The egress proxy runs two replicas with no disruption protection, and both pods restarted 13h before this was filed. Today a proxy blip breaks whichever CI steps opted in. If the proxy becomes always-on, it becomes a hard dependency for every job's egress, and a node drain or rollout takes all CI with it. ## Fix Add a PodDisruptionBudget with `minAvailable: 1`. Consider whether the runners themselves want one too, since the namespace has none at all.

coilyco-ops added the

role/sysadmin

priority

labels

2026-08-25 08:02:14 +00:00

coilyco-ops added the

autonomy

async-consult

label

2026-08-25 09:31:02 +00:00

coilyco-ops added this to the Forgejo runners milestone

2026-08-29 05:40:56 +00:00