- CronJob at 0 4 1 * * restores the latest openwebui-manifests backup into a transient restore-test-<ts> namespace, diffs core resource-kind counts against the live openwebui namespace, and deletes the target namespace. - ServiceAccount + ClusterRole scoped to velero.io + namespace + configmap ops. Script enforces the 'restore-test-' name prefix before deleting a namespace. Cluster-wide 'view' bound for the diff. - Failure surfaces via kube_job_status_failed (DEV-574 alert stack) and via ConfigMap velero-restore-drill-last-result written on every run. - Optional Paperclip webhook Secret is sealed but unreachable from Hetzner right now (home Paperclip is LAN-only); wired as optional secretKeyRef so the drill runs regardless. Connectivity is a follow-up. - Argo Application velero-restore-drill points at velero/restore-drill/. Co-Authored-By: Paperclip <noreply@paperclip.ing>
3.2 KiB
Velero restore drill
Monthly Kubernetes CronJob that exercises the Velero backup chain by restoring
the openwebui namespace into a transient restore-test-<ts> namespace,
diffing resource counts against the live namespace, and cleaning up.
Wired by DEV-575 under the DEV-569 Velero cluster-reconstruction backup layer.
What it does
- Picks the latest
Completedbackup for scheduleopenwebui-manifests(envDRILL_BACKUP_OVERRIDEshort-circuits this — used by the failure drill). - Creates a
RestoreCR with namespace mappingopenwebui: restore-test-<ts>. - Waits for the
Restoreto reach a terminal phase (Completed,PartiallyFailed,Failed,FailedValidation) or times out at 30 min. - Counts core kinds (deployments, statefulsets, services, configmaps, PVCs, …) in source vs target and requires strict equality on the manifest-carrying kinds to pass.
- Deletes the
restore-test-<ts>namespace and theRestoreCR. - Writes the final report to ConfigMap
velero/velero-restore-drill-last-resultand logs it to stdout. - Optionally POSTs a JSON payload to the Paperclip routine webhook if
the
velero-restore-drill-webhookSecret is present (currently unreachable from the cluster — see follow-up ticket).
Failure signalling
The Job exits non-zero on any failure. The alerting stack (DEV-574) will fire
KubeJobFailed on the velero-restore-drill-* job name and route the alert
through existing Alertmanager receivers. Ops can inspect the last result via:
kubectl -n velero get cm velero-restore-drill-last-result -o jsonpath='{.data.result}'
kubectl -n velero get cm velero-restore-drill-last-result -o jsonpath='{.data.log}'
Manual runs
Kick a drill on demand:
kubectl -n velero create job --from=cronjob/velero-restore-drill drill-adhoc-$(date -u +%Y%m%d-%H%M%S)
Force a failure to test the alert wiring (uses a bogus backup name):
kubectl -n velero create job drill-adhoc-fail-$(date -u +%s) \
--image=alpine/k8s:1.30.5 \
--dry-run=client -o yaml \
| kubectl -n velero apply -f - # only if you need a scratch, prefer:
kubectl -n velero create job --from=cronjob/velero-restore-drill drill-fail-$(date -u +%Y%m%d-%H%M%S) --dry-run=client -o yaml \
| sed 's/value: "openwebui-manifests"/value: "openwebui-manifests"\n - name: DRILL_BACKUP_OVERRIDE\n value: "no-such-backup"/' \
| kubectl apply -f -
For a simpler failure drill, apply the CronJob-derived Job then patch env before scheduling — see the parent DEV-575 comment for the exact runbook.
RBAC scope
velero.io/backups,restores: get/list/create/delete in-cluster (namespace scope is enforced by resource type — Backups/Restores live invelero).namespaces: cluster-wide get/list/create/delete. Cannot be name-scoped in RBAC; the drill script enforces therestore-test-prefix before ever deleting a namespace.configmaps: get/list/create/update/patch (only used to write the result ConfigMap invelero— pod-level namespace scoping via ServiceAccount).- Cluster-wide
viewClusterRole for the counting diff.