Move backup resources from separate 'backup' namespace to 'forgejo': - Change CronJob namespace from 'backup' to 'forgejo' - Reuse existing forgejo-postgres-secret instead of duplicate backup secret - Remove backup-namespace.yaml (no longer needed) - Remove forgejo-backup-secret-sealed.yaml (using existing secret) This simplifies the architecture by keeping all Forgejo-related resources in a single namespace and eliminating credential duplication. Co-Authored-By: Paperclip <noreply@paperclip.ing>
2 KiB
2 KiB
Dozzle Deployment Progress
Completed
-
✅ Created deployment manifests in
apps/dozzle/directory:- Namespace, ServiceAccount with cluster-wide RBAC for pod log access
- PVC (1Gi, hcloud-volumes-encrypted, ReadWriteOnce with Recreate strategy)
- Deployment with oauth2-proxy sidecar for Pocket ID OIDC authentication
- Service routing to oauth2-proxy (port 4180)
- Ingress with TLS for
dozzle.basicstack.de
-
✅ Created ArgoCD application (
app-dozzle.yaml) with automated sync -
✅ Created Pocket ID OIDC client:
- Client ID:
179c13f2-d251-4e1e-b1a0-c070df350c4e - Callback URL:
https://dozzle.basicstack.de/oauth2/callback - Sealed secret with client credentials and oauth2-proxy cookie secret
- Client ID:
-
✅ Configured oauth2-proxy sidecar (running successfully)
-
✅ All manifests committed and pushed to repository
Current Blocker
Dozzle container is failing to start with error: "Could not connect to any Docker Engine"
Root cause: Dozzle v10.6.10 is not auto-detecting Kubernetes mode despite:
- Proper RBAC permissions (ClusterRole with pods, pods/log, namespaces access)
- Service account correctly mounted at
/var/run/secrets/kubernetes.io/serviceaccount/ - ClusterRoleBinding connecting service account to ClusterRole
Troubleshooting attempted:
- Tried various command-line flags (
--agent,--no-analytics) - Verified RBAC permissions are correct
- Checked service account token is mounted
- oauth2-proxy sidecar is working correctly (1/2 containers ready)
Next steps:
- Research Dozzle Kubernetes deployment examples/Helm charts for missing configuration
- Consider alternative: Try older Dozzle version or different container log viewer
- May need to explicitly configure Kubernetes API endpoint or use different deployment approach
Current pod status:
dozzle-f4d557466-qh28j 1/2 CrashLoopBackOff
All infrastructure is in place; just need to resolve Dozzle's Kubernetes detection issue.