You've got 2 different issues. (1) What is causing the disconnects?
(2) Why is the telnet daemon not recognizing that the client has
disappeared?

I'd bet issue (1) is due to client problems rather than server abends
as you speculate. Check the XP event logs.

I've seen issue (2) with some applications. It's been a while since
I've looked at Appgen; don't remember if that was one of troublesome
ones. It depends on the line discipline, what jobs are attached to the
session, etc. There are some kernel settings and telnet timeouts you
can set to get rid of stale sessions, but those generally are treating
symptoms rather than causes.

Generally the SSHD does better. I'd switch.

--RR

James_Szabadics wrote:
> Various users have report random events where their telnet session very
> abruptly terminates leaving an orphan process.
>
> There doesnt seem to be any pattern of user or location or time or
> specific part of the application where this event occurs and usually
> only one user of the 40 is impacted. Sometimes the user has multiple
> open sessions and only one session ends. Anecdotally it may happen to
> only the active session at the time but we are uncertain of this.
>
> We thought that we might find some messages in a log somewhere in the
> unix host that might point us in the right direction. We looked at
> /usr/adm/messages but there was no indication of any related message.
> Where should we look for relevant messages regarding abnormal session
> termination on the server . What is a good troubleshooting method for
> this kind of issue? We are thinking of turning on debugging on the
> telnet daemon.
>
> Our system is SCO Openserver 5.0.6 and clients are Windows XP using
> Appgen Client 5.1 (vt220 emulator telnet).

James_Szabadics wrote (on Wed, Jul 26, 2006 at 12:50:55AM -0700):
> Various users have report random events where their telnet session very
> abruptly terminates leaving an orphan process.
>
> There doesnt seem to be any pattern of user or location or time or
> specific part of the application where this event occurs and usually
> only one user of the 40 is impacted. Sometimes the user has multiple
> open sessions and only one session ends. Anecdotally it may happen to
> only the active session at the time but we are uncertain of this.
>
> We thought that we might find some messages in a log somewhere in the
> unix host that might point us in the right direction. We looked at
> /usr/adm/messages but there was no indication of any related message.
> Where should we look for relevant messages regarding abnormal session
> termination on the server . What is a good troubleshooting method for
> this kind of issue? We are thinking of turning on debugging on the
> telnet daemon.

Because it's kind of icky :-( running tcpdump on SCO, perhaps try
running Ethereal on the windows (right?) clients.

Or, interpose a switch (or machine) which lets you grab packets and look
at them.

Or, run scotruss on telnetd.

Or, heck, d/l an open source telnetd and replace the stock one.

Or just load ssh (an adventure in itself on SCO).

--
_________________________________________
Nachman Yaakov Ziskind, FSPA, LLM awacs@ziskind.us
Attorney and Counselor-at-Law http://ziskind.us
Economic Group Pension Services http://egps.com
Actuaries and Employee Benefit Consultants

Thanks for the pointers. I will take a close look at XP events next
time.

I'm interested in looking at dealing with stale sessions as my band-aid
while i research the other options with ssh and the appgen client. We
have some mobile wireless users that could benefit from attention to
this.

What kernel settings and telnet timeouts are configureable to control
how these are handled?


ThreeStar wrote:
> You've got 2 different issues. (1) What is causing the disconnects?
> (2) Why is the telnet daemon not recognizing that the client has
> disappeared?
>
> I'd bet issue (1) is due to client problems rather than server abends
> as you speculate. Check the XP event logs.
>
> I've seen issue (2) with some applications. It's been a while since
> I've looked at Appgen; don't remember if that was one of troublesome
> ones. It depends on the line discipline, what jobs are attached to the
> session, etc. There are some kernel settings and telnet timeouts you
> can set to get rid of stale sessions, but those generally are treating
> symptoms rather than causes.
>
> Generally the SSHD does better. I'd switch.
>
> --RR
>
> James_Szabadics wrote:
> > Various users have report random events where their telnet session very
> > abruptly terminates leaving an orphan process.
> >
> > There doesnt seem to be any pattern of user or location or time or
> > specific part of the application where this event occurs and usually
> > only one user of the 40 is impacted. Sometimes the user has multiple
> > open sessions and only one session ends. Anecdotally it may happen to
> > only the active session at the time but we are uncertain of this.
> >
> > We thought that we might find some messages in a log somewhere in the
> > unix host that might point us in the right direction. We looked at
> > /usr/adm/messages but there was no indication of any related message.
> > Where should we look for relevant messages regarding abnormal session
> > termination on the server . What is a good troubleshooting method for
> > this kind of issue? We are thinking of turning on debugging on the
> > telnet daemon.
> >
> > Our system is SCO Openserver 5.0.6 and clients are Windows XP using
> > Appgen Client 5.1 (vt220 emulator telnet).

James_Szabadics wrote:
> Thanks for the pointers. I will take a close look at XP events next
> time.
>
> I'm interested in looking at dealing with stale sessions as my band-aid
> while i research the other options with ssh and the appgen client. We
> have some mobile wireless users that could benefit from attention to
> this.
>
> What kernel settings and telnet timeouts are configureable to control
> how these are handled?

For info on keepalive see:

http://wdb1.sco.com/kb/showta?taid=107691

John