ADSM-L

SCO TCP/IP

1996-11-11 08:49:51
Subject: SCO TCP/IP
From: REIDAR KARLSSON <reidar.karlsson AT SKANDIA DOT SE>
Date: Mon, 11 Nov 1996 14:49:51 +0100
--- Received from SK.KARREI  +46 8 7883302          96-11-11 14.48
  -> IN:ADSM-L(A)VM.MARIST.EDU
  -> IN:ADSM-L(A)VM.MARIST.EDU
  -: - - - - - - - - - - > COMPLEX NAME OF MHS USER
Hi.
Is there anybody out there with a solution for the following
problem:

Background
The server is on a MVS/S390 system located in Stockholm, the
clients are located in different areas in Sweden, Norway and
in Denmark. In Denmark we have problem with the backup process
for a SCO unix.
The TCP/IP connection fails. The dsmerror.log contain these
message:
TcpFlush: Error 108 sending data on Tcp/Ip socket 4.
SessSendVerb: Error sending Verb, rc: -50
TcpFlush: Error 32 sending data on Tcp/Ip socket 4.
TcpFlush: Error 9n sending data on Tcp/Ip socket 42949672950
ANS4017E Session rejected: TCP/IP connection failure.

I do understand the last message, but not the previous. It's not
that the backup process fails, we also need to re-boot the SCO
because the TCP/IP in it hangs.

We have these values for the parameters in the dsm.sys  for the
client: TXNBYTELIMIT        300
        TCPWINDOWSIZE       8
        TCPBUFFSIZE         4
        SLOWINCREMENTAL     YES
We had the default values earlier, and after assuming that the
ADSM were much quicker than the connection could manage, we
lowered the values.
We also changed two values for the server,
COMMTIMEOUT changed from 60 to 300
COMMTIMEOPEN changed from 10 to 100

After the changes above, the backup process fails after approx.
15 minutes, before the changes about 5 minutes.

So, if there is someone with the answer I would be Very Happy to
know.
Rgds /Reidar.Karlsson(a)Skandia.Se    8-)
<Prev in Thread] Current Thread [Next in Thread>
  • SCO TCP/IP, REIDAR KARLSSON <=