Problem: Job ran on every ArgoCD sync, creating duplicate runners in Gitea. Solution: - initContainer gets token + saves to PVC (/data/.registration-token) - Runner container registers once, persists .runner file in PVC - Subsequent restarts skip registration (idempotent) - PVC runner-data (1Gi) persists registration across pod restarts - Removed register-job.yaml, moved RBAC to rbac.yaml - Runner waits for DinD before starting - Stable runner name: test-env-runner - Labels: edt + ubuntu-latest
35 lines
789 B
YAML
35 lines
789 B
YAML
# RBAC for runner registration initContainer
|
|
# Allows reading gitea-admin secret and listing pods in gitea namespace
|
|
---
|
|
apiVersion: v1
|
|
kind: ServiceAccount
|
|
metadata:
|
|
name: runner-registrar
|
|
namespace: test-env
|
|
---
|
|
apiVersion: rbac.authorization.k8s.io/v1
|
|
kind: ClusterRole
|
|
metadata:
|
|
name: test-env-gitea-reader
|
|
rules:
|
|
- apiGroups: [""]
|
|
resources: ["secrets"]
|
|
resourceNames: ["gitea-admin"]
|
|
verbs: ["get"]
|
|
- apiGroups: [""]
|
|
resources: ["pods"]
|
|
verbs: ["get", "list"]
|
|
---
|
|
apiVersion: rbac.authorization.k8s.io/v1
|
|
kind: ClusterRoleBinding
|
|
metadata:
|
|
name: test-env-gitea-reader
|
|
subjects:
|
|
- kind: ServiceAccount
|
|
name: runner-registrar
|
|
namespace: test-env
|
|
roleRef:
|
|
kind: ClusterRole
|
|
name: test-env-gitea-reader
|
|
apiGroup: rbac.authorization.k8s.io
|