On Wed, 1 Dec 1999 01:41:42 -0500 , "Ogunsanya A.T."
<[log in to unmask]> wrote:

>Thanks for your reply,
>I do NOT have checksum on for tcp. Could this be the problem?

IF you have a noisy, problematic LAN, then bad packets will arrive.  IF
you do NOT have checksums on, these packets will not be rejected at the
front end, and will be permitted to "enter" the system, causing all
kinds of hard-to-find, strange errors.  So, I strongly urge everyone to
turn on checksums on both of the NMMGR screens on all systems.  (I urge
even more strongly that HP change the default!)  If this turns out to
help you, then you still need to determine why your LAN is noisy.  But
in the meantime, I prefer "slow" to "fast with corrupted data and
occasional hangs".

While we are on the subject, I always suggest a few other changes from
the default values as suggested by Eero Laurilla of HP CSY Networking
Lab in a posting on 16 Nov 1995.  In that posting Eero said:

"I think the TCP timeout default values on MPE/iX are way out of wack."

He suggests changing the default values:

   [4  ]    Retransmission Interval
   [180]  Maximum Time to Wait for Remote Response
   [5 ]    Initial Retransmission Interval
   [4 ]   Maximum Retransmissions per Packet

to:

   [1  ]    Retransmission Interval
   [300]  Maximum Time to Wait for Remote Response
   [2 ]    Initial Retransmission Interval
   [6 ]   Maximum Retransmissions per Packet

This probably has NOTHING to do with your problem.  But, while you are
in NMMGR, you might as well make these changes too.  And, if other
people see this and make these changes, there will be fewer hung
sessions in this world.

Steve Cooper