[487] in Moira
Another idea about the DCM hang
daemon@ATHENA.MIT.EDU (Richard Basch)
Sat Nov 28 11:28:23 1992
Date: Sat, 28 Nov 1992 11:28:06 -0500
To: bug-moira@Athena.MIT.EDU
Cc: network@Athena.MIT.EDU
From: "Richard Basch" <basch@MIT.EDU>
I wasn't feeling all that well when Ted asked me about the aliases
update, so I wasn't thinking all too clearly.
Anyway, obviously, testing for ECONNRESET isn't an answer.
What might be the answer is if we use the KEEPALIVE option on the
socket; that way if the TCP connection really does drop, it will be
detected, and in that case we test for ECONNRESET, or short reads, or
whatever. I believe we were probably blocking on the socket waiting for
input and we had not made any attempts to "ping" the remote TCP
connection. If I recall, keepalive isn't standard.
-R
