Add PodDisruptionBudgets for critical services #31

New issue

Open

opened 2026-07-25 19:37:17 +00:00 by lab · 0 comments

lab commented

2026-07-25 19:37:17 +00:00

Owner

No PDBs exist. During node drain or maintenance, all replicas of a service could go down simultaneously.

Services needing PDBs

Authelia (DaemonSet — already one per node, lower priority)
Caddy (2 replicas on master-ingress nodes — HIGH priority)
Forgejo (if replicas > 1)
MySQL (if replicas > 1)
Prometheus (if replicas > 1)

Example PDB for Caddy

apiVersion: policy/v1
kind: PodDisruptionBudget
metadata:
  name: caddy-pdb
  namespace: caddy
spec:
  minAvailable: 1
  selector:
    matchLabels:
      app: caddy

Steps

Audit which services have replicas > 1
Create PDB manifests for each
Apply via GitOps workflow

No PDBs exist. During node drain or maintenance, all replicas of a service could go down simultaneously. ## Services needing PDBs - **Authelia** (DaemonSet — already one per node, lower priority) - **Caddy** (2 replicas on master-ingress nodes — HIGH priority) - **Forgejo** (if replicas > 1) - **MySQL** (if replicas > 1) - **Prometheus** (if replicas > 1) ## Example PDB for Caddy ```yaml apiVersion: policy/v1 kind: PodDisruptionBudget metadata: name: caddy-pdb namespace: caddy spec: minAvailable: 1 selector: matchLabels: app: caddy ``` ## Steps 1. Audit which services have replicas > 1 2. Create PDB manifests for each 3. Apply via GitOps workflow