patch tags

README.md

@@ -25,3 +25,30 @@ The compose service uses tmpfs storage, giving each test run a clean database.
 
 - `ocrmypdf` (optional but recommended): Used by the OCR worker to extract text from PDFs when no embedded text layer is available. Ensure it is installed and available on the worker hosts if OCR is desired.
 - Quickwit (optional): The Quickwit indexer is used to ingest extracted text for search. Set `QUICKWIT_ENDPOINT` and `QUICKWIT_INDEX` in the environment when running workers if you want indexing jobs to run. The local compose file starts a Quickwit instance on `http://localhost:7280` and seeds the `documents` index automatically.
+
+## Running Migrations in Kubernetes
+
+The backend container image ships the `diesel` CLI, so schema migrations can be executed as a short-lived Job (or Helm hook) before rolling out new pods. Example manifest:
+
+```yaml
+apiVersion: batch/v1
+kind: Job
+metadata:
+  name: paperless-migrate
+spec:
+  template:
+    spec:
+      restartPolicy: OnFailure
+      containers:
+        - name: migrate
+          image: ghcr.io/example/paperless-backend:<TAG>
+          command: ["/usr/local/bin/diesel", "migration", "run"]
+          env:
+            - name: DATABASE_URL
+              valueFrom:
+                secretKeyRef:
+                  name: paperless-db
+                  key: DATABASE_URL
+```
+
+Run the Job manually (`kubectl apply -f migrate-job.yaml`) or configure it as a Helm pre-install/pre-upgrade hook so migrations run automatically on each deployment. Once the Job succeeds, deploy/update the backend `Deployment` as usual.