# services/mailu/mailbox-watchdog.yaml --- apiVersion: v1 kind: ServiceAccount metadata: name: mailu-mailbox-watchdog namespace: mailu-mailserver --- apiVersion: rbac.authorization.k8s.io/v1 kind: Role metadata: name: mailu-mailbox-watchdog namespace: mailu-mailserver rules: - apiGroups: [""] resources: ["pods"] verbs: ["get", "list", "delete"] - apiGroups: [""] resources: ["pods/exec"] verbs: ["create"] --- apiVersion: rbac.authorization.k8s.io/v1 kind: RoleBinding metadata: name: mailu-mailbox-watchdog namespace: mailu-mailserver roleRef: apiGroup: rbac.authorization.k8s.io kind: Role name: mailu-mailbox-watchdog subjects: - kind: ServiceAccount name: mailu-mailbox-watchdog namespace: mailu-mailserver --- apiVersion: apps/v1 kind: Deployment metadata: name: mailu-mailbox-watchdog namespace: mailu-mailserver spec: replicas: 1 selector: matchLabels: app: mailu-mailbox-watchdog template: metadata: labels: app: mailu-mailbox-watchdog spec: serviceAccountName: mailu-mailbox-watchdog automountServiceAccountToken: true nodeSelector: hardware: rpi5 node-role.kubernetes.io/worker: "true" containers: - name: watchdog image: registry.bstein.dev/bstein/kubectl:1.35.0 imagePullPolicy: IfNotPresent command: ["/bin/sh", "-ec"] env: - name: HOME value: /tmp args: - | failures=0 last_pod="" sleep 60 while true; do pod="$(kubectl get pods \ -l app.kubernetes.io/component=dovecot,app.kubernetes.io/instance=mailu \ --field-selector=status.phase=Running \ -o jsonpath='{.items[0].metadata.name}' 2>/dev/null || true)" if [ -z "${pod}" ]; then failures=0 last_pod="" sleep 30 continue fi if [ "${pod}" != "${last_pod}" ]; then failures=0 last_pod="${pod}" echo "watching ${pod}; allowing startup grace" sleep 60 continue fi if timeout 15s kubectl exec "${pod}" -c dovecot -- \ doveadm mailbox status -u brad@bstein.dev uidnext INBOX >/dev/null 2>&1; then if [ "${failures}" -gt 0 ]; then echo "mailbox health recovered after ${failures} failed checks" fi failures=0 else failures=$((failures + 1)) echo "mailbox health check failed (${failures}/4) for ${pod}" >&2 fi if [ "${failures}" -ge 4 ]; then echo "recreating ${pod} to remount mailbox storage after persistent failures" >&2 kubectl delete pod "${pod}" --wait=false failures=0 last_pod="" sleep 180 continue fi sleep 30 done resources: requests: cpu: 10m memory: 32Mi limits: cpu: 100m memory: 128Mi securityContext: allowPrivilegeEscalation: false capabilities: drop: ["ALL"] readOnlyRootFilesystem: true runAsNonRoot: true runAsUser: 65532 volumeMounts: - name: tmp mountPath: /tmp imagePullSecrets: - name: harbor-regcred volumes: - name: tmp emptyDir: {}