background reload secondary

.gitea/workflows/reload-act-runner.yaml

@@ -28,10 +28,24 @@ jobs:
       stack_name: gitea_act_runner_secondary
       # the primary runner executes the secondary restart webhook
       runs_on_label: runner-primary
+      # We use a detached webhook call so primary doesn't hang if the Gitainer stack takes 5 minutes to rebuild, and it avoids bridge network drop edge cases
+      delayed_silent: true
 
-  pull-cache-secondary:
+  wait-for-secondary-death:
     needs: [diff, reload-secondary]
     if: ${{ needs.diff.outputs.is_changed == 'true' }}
+    runs-on: runner-primary
+    steps:
+      - name: Wait for Gitainer to kill the secondary runner
+        run: |
+          echo "The detached webhook container sleeps for 10s before firing."
+          echo "We wait 30s here to guarantee the webhook fired and Gitainer destroyed the old secondary runner."
+          sleep 30
+
+  pull-cache-secondary:
+    needs: [diff, wait-for-secondary-death]
+    if: ${{ needs.diff.outputs.is_changed == 'true' }}
+    # Because the old secondary is dead, Gitea leaves this job in the queue until the NEW secondary runner boots up.
     runs-on: runner-secondary
     steps:
       - name: Force actions cache pull