This is from the archive:
The author is Mike.Golovach@vtmednet.org
If you’re on AIX, this is the solution. It sure saved me
a lot of heartache.
>>>>>>>>
We had this same problem and I tried everything including letting an
alert create a keep alive message. But the vendor would not agree to
doing the same thing on their end of the interface for their outbound
thread … which was a good thing. It forced me to keep digging.
The issue was that our TCP/IP (AIX 5.2) was configured with all of the
default timeout values. The TCP_keepidle parameter needed to be set to
an interval shorter that Cisco timeout value. We set ours to 3600 and
the problem was solved.
Michael
>>>>>>>>
