SSH/Telnet sessions timeing out over satellite link

[James P. Byrne]

We're having an issue with remote SSH/Telnet sessions timing out (after
approx 5 minutes) over a satellite communications link, but only to the
VMS nodes (running MultiNet v5.1) on our main network, and not to other
Solaris servers on the same network [as the VMS clusters]. It may be a
timing issue and I was wondering if there were any timer settings in
MultiNet's implementation of TCP/IP [as compared to Solaris
implementations of TCP/IP] that could be increased to resolve this
problem. After some preliminary analysis, it appears that a reset is
coming from somewhere in-between the client and the host, but it does
not affect SSH sessions to the Solaris hosts.

Jim Byrne
Montclair State University

[Dan Wing]

Post by James P. Byrne
We're having an issue with remote SSH/Telnet sessions timing out (after
approx 5 minutes) over a satellite communications link, but only to the
VMS nodes (running MultiNet v5.1) on our main network, and not to other
Solaris servers on the same network [as the VMS clusters]. It may be a
timing issue and I was wondering if there were any timer settings in
MultiNet's implementation of TCP/IP [as compared to Solaris
implementations of TCP/IP] that could be increased to resolve this
problem. After some preliminary analysis, it appears that a reset is
coming from somewhere in-between the client and the host, but it does
not affect SSH sessions to the Solaris hosts.

Something/someone is spoofing a TCP RST? My guess is the Solaris telnet
and/or ssh connections are doing keepalives to remind the box in the
middle that the session is still active, perhaps? Are the sessions that
are timing out on your MultiNet systems active during that 5 minute
interval, or inactive?

Are you perhaps running Watcher <http://www.madgoat.com/watcher.html>
or some other idle terminal program on the VMS box with different
rules for the incoming Telnet/ssh sessions than for your local
sessions?

-d