connections to albatross are not cleanly closed

Yesterday I cleaned (killed) all albatross-tls-endpoint processes (there were around 50 of them). Today, there are again 7 processes up and running (started between 10:08 and 10:12 today).

It is crucial to properly clean these connections up and close the TCP connections, we need to investigate why they are hanging around.