fix: recreate buildkitd when not running (config drift)

The name-exists guard skipped recreation when a stopped container
from a previous failed run existed — e.g. one created before the
port mapping was added, leaving 127.0.0.1:8375 with no listener.

New logic: recreate the container whenever it is not running;
keep it (warm cache) when it is already up.

action.yml

@@ -36,14 +36,19 @@ runs:
     - name: Start BuildKit container (with labels)
       shell: bash
       run: |
-        docker ps -a --format '{{.Names}}' | grep -qx '${{ inputs.container_name }}' || \
+        # Recreate if a stopped/old container exists (config drift, e.g. port
-        docker run -d --name ${{ inputs.container_name }} --privileged \
+        # mapping changes, would otherwise be skipped by the name guard).
-          --network ${{ inputs.network }} \
+        # If it is already running with the right config, keep it (warm cache).
-          -p 127.0.0.1:${{ inputs.port }}:${{ inputs.port }} \
+        if [ "$(docker inspect -f '{{.State.Running}}' ${{ inputs.container_name }} 2>/dev/null)" != "true" ]; then
-          --restart unless-stopped \
+          docker rm -f ${{ inputs.container_name }} >/dev/null 2>&1 || true
-          --label com.centurylinklabs.watchtower.monitor-only=${{ inputs.monitor_only }} \
+          docker run -d --name ${{ inputs.container_name }} --privileged \
-          --label dockhand.notify=${{ inputs.dockhand_notify }} \
+            --network ${{ inputs.network }} \
-          ${{ inputs.image }} --addr tcp://0.0.0.0:${{ inputs.port }}
+            -p 127.0.0.1:${{ inputs.port }}:${{ inputs.port }} \
+            --restart unless-stopped \
+            --label com.centurylinklabs.watchtower.monitor-only=${{ inputs.monitor_only }} \
+            --label dockhand.notify=${{ inputs.dockhand_notify }} \
+            ${{ inputs.image }} --addr tcp://0.0.0.0:${{ inputs.port }}
+        fi
 
     - name: Set up Docker Buildx
       uses: docker/setup-buildx-action@v4