I haven’t run into the issue myself, but I did spot this after a quick search:
They’re using AAP’s single-node deployment, which is podman based, but if you’re seeing this in AWX on k8s, I’m not sure if you would see the same pause.pid file on a cluster node.