SLRP uses a post-stop hook to ask the container daemon to clean up the endpoint socket after its plugin container is terminated. However, if the container daemon is destructed while waiting for the container it monitors before the container itself is terminated, the socket file will remain there, making SLRP unable to recover.
There might be two solutions:
1. During SLRP recovery, check if the plugin container is still running.
2. Start the container daemon in the waiting phase.