Bug#256916: More info
I am going to eat my own words now...
The process which bound the original tcp socket 2345 is still on
the system as a zombie child of init. Presumably its parent exited and
for whatever reason init failed to reap the reparented child when it
exited. This indicates that there could be a race condition in init,
but there is still the possibility of a kernel bug. Is there any
mechanism for asking init to scan/reap any zombies? Respawning init
does nothing.
To make matters more frustrating, ps doesn't display zombie processes
when called with -L or -T.
Interestingly, it is still possible to establish a connection to the
wedged socket. All input is queued as presumably is the connection for
acceptance by the zombie...
Regards,
Matt
--
Matt Kern
http://www.undue.org/
Reply to: