Now a 'netstat' command is showing a buildup of ssh connections to a host that is, in reality, timing out. This is causing problems on our production box, so we've had to kill it for now.
I'm going to re-write this thing using sockets, and only resort to ssh if a socket connection is refused (from which, we can reasonably assume, the server is down - I can verify with an added ping call).
In other news, they have me writing reports to verify the validity of other reports. Managers - gotta love 'em.
