diff --git a/.forgejo/workflows/deploy.yml b/.forgejo/workflows/deploy.yml
index 8ff68af..9bc878b 100644
--- a/.forgejo/workflows/deploy.yml
+++ b/.forgejo/workflows/deploy.yml
@@ -64,12 +64,12 @@ jobs:
       - name: Health checks with rollback
         run: |
           echo "Waiting for services to start..."
-          sleep 20
+          sleep 30
 
           BACKEND_OK=false
-          for i in 1 2 3 4 5; do
+          for i in 1 2 3 4 5 6 7 8 9 10; do
             if docker run --rm --network bilhej-prod_default curlimages/curl:8.5.0 \
-               -s http://bilhej-backend-prod:8080/api/vehicles/ABC123 > /dev/null; then
+               -sf http://bilhej-backend-prod:8080/api/payment/swish-info > /dev/null; then
               echo "Backend is healthy"
               BACKEND_OK=true
               break
@@ -93,12 +93,25 @@ jobs:
           if [ "$BACKEND_OK" != "true" ] || [ "$FRONTEND_OK" != "true" ]; then
             echo ""
             echo "═══════════════════════════════════════════════════"
-            echo "  HEALTH CHECK FAILED — ROLLING BACK DEPLOYMENT"
+            echo "  HEALTH CHECK FAILED — DIAGNOSTICS"
+            echo "═══════════════════════════════════════════════════"
+            echo ""
+            docker compose -p bilhej-prod -f docker-compose.prod.yml ps
+            echo ""
+            echo "--- Backend logs ---"
+            docker logs bilhej-backend-prod 2>&1 | tail -80 || true
+            echo ""
+            echo "--- Postgres logs ---"
+            docker logs bilhej-postgres-prod 2>&1 | tail -30 || true
+            echo ""
+            echo "═══════════════════════════════════════════════════"
+            echo "  ROLLING BACK DEPLOYMENT"
             echo "═══════════════════════════════════════════════════"
             echo ""
             docker compose -p bilhej-prod -f docker-compose.prod.yml down
             echo ""
             echo "Rolled back. Containers stopped. DB volume preserved."
+            echo "Read Backend logs above to find the root cause before redeploying."
             exit 1
           fi
 
diff --git a/README.md b/README.md
index c3fb243..e4f2a66 100644
--- a/README.md
+++ b/README.md
@@ -311,6 +311,20 @@ Before the first deploy, complete these steps on the production server (`srvr.nu
 3. Enter a version tag (e.g., `v0.1.0`).
 4. Click **Run workflow**.
 
+### Deploy failed (backend health check)
+
+If the job passes the frontend check but the backend never becomes healthy:
+
+1. Open the failed job log and read **Backend logs** (printed before rollback).
+2. Match the error to a fix — do not guess:
+   - **`password authentication failed`** — DB credentials in the running stack do not match
+     what Postgres was initialized with; fix credentials or Postgres password to match (only
+     wipe the volume if you accept losing prod data).
+   - **`Production requires ADMIN_EMAIL and ADMIN_PASSWORD`** — add those Forgejo secrets.
+   - **Flyway / migration errors** — fix schema or migration history before redeploying.
+3. **DBeaver from your laptop** — prod Postgres binds to `127.0.0.1:5433` on the server only.
+   Use an SSH tunnel, then host `localhost` port `5433` (not `192.168.0.59` directly).
+
 ### What Happens
 
 | Step | Action |