Telnet Sessions getting killed on its own - HPUX 11.11

[Naveen Shetty]

Dear All,

I am coming across a typical problem in HPUX 11.11 OS on HP 9000 rp7410 & L2000 Servers.

Even though telnet sessions are in use & is used continously, sessions get killed on its own without any message.

even syslog.log does not indicate any message pertaining to this.

Any one who has come across this problem.

Server is with Bundles of Mar-03.

Thanks in advance.
Naveen Shetty.

[Keith Bevan_1]

Naveen,

You say the sessions are being used continuously !

The following advice only applies if the session remain idle for a certain time period so may or may not be relevant (worth checking this anyway):-

Check your .profile .login for the shell environment your are running for each session. Do you have an entry for timeout. It would appear in the .profile .login as a variable :-

TMOUT=3200

If this exists then you may wish to increase the value or remove or comment out the entry.

Keith

[Umapathy S]

Naveen,
Try to deduce the problem with different applications under different scenarios. Are there any patterns in the killing of applications like the same time or same part of code etc.

Try to run some applications with tusc.

HTH,
Umapathy

[Piergiacomo Perini]

Hi,
I had troubles like this (with other version of
HP-UX and hp box):
try to check if some applications are generating huge number of processes that
collapse your memory.

HTH

[Michael Tully]

If users get killed off for no reason (i.e. shell users) there must be some time out being used in either the users profile or globally in /etc/profile. Check the users shell, ksh TMOUT (.profile) and setlogout for csh. (.cshrc)

[yogesh_4]

Dear Naveen,
Telnet users are getting temporary directory from /var . If your application is generating logs and /var is getting full then also your telnet sessions can logoff automatically.

[Kevin O'Donovan]

Hi,

I've seen where idle sessions time out alright, this can be sorted out by putting

set autologout=0

in .profile or .login, not sure which it was, depending on the shell of course.

With sessions being used and dying off sounds a bit strange though... when its happening on a few servers, does it happen on a few different clients too?

Does it happen after a particular amount of time, or more randomly?

I wonder is there a problem in the network somewhere that breaks the connection? Maybe you could have a session open, and at the same time a ping going between the machines. If you start loosing a few pings the same time the sessions die I'd start looking at the network.

Another thing to check out, do all clients use the same application to telnet to the servers, e.g. Reflections or Windows telnet? Maybe try using a different application on one of the machines and see if that works any better, would eliminate/confirm the application in the list of suspects.

If the both above fail it would look like the server, and sounds like maybe a patch is needed...

hope that helps,
Kevin.

[Massimo Bianchi]

Hi,
the connection is direct or there are some network device in between ?

I'm thinking of: firewall timeouts, dynamic router changes of route-table and so on...

Massimo

[Massimo Bianchi]

Another good reason is a duplicate ip.

Check with arpa cache if the MAC address is always the same....

Massimo

[Rita C Workman]

You may need to do some 'narrowing down' of the problem.

First the obvious...are you patched okay.

Couple folks mentioned .. a timeout set in your .profile.
Check folks who are getting killed and see if this is the case.
Also check your /etc/profile....did somebody put a timeout in there?

Now if it's not a timeout being set..then I might start looking at:
What is my kernel parms set for tty connections? Probably not causing the
the killing, but might need increased.

Next - Who is getting killed? What were they doing when they were killed?
Here is where the real detective work begins...what maybe do they have in common????
Try a simple test....set up a telnet session for yourself....do just enough so it shows activity. See if it gets killed. If it doesn't - then you know that the other problem is application related. You will then need to figure out which application.
It may be something those folks are doing that is triggering this. Like an application error that has gone awry, so it kills the session.

Just a thought,
Rita

[Suresh Patoria]

Naveen,

i want u check the following things,

enable the logging services using the inetd -l command

have set the time out in /etc/inetd.conf file

After enabling inetd logging all the messages will show in syslog.log file

thanx

[okcunix]

Naveen,

You might take a look at your network also. I've had similar experience, especially if telnetd' into server, and exporting X display back. My GUI would get an intermittent "broken/killed" mesg. Determined it was the firewall/router. Network guys acknowledged it was their issue, told me the fix was to "live with it".

[binil]

Dear Naveen,
How are you? How is life going on...
Is this problem happening while continuously working also?
Server to server telnet is also giving problem?

I hope this could be misbehavior of patches unless there is some major changes in main network service configuration files and nittab file.

If you have any backup (Ignite)which is prior to the March2003 patching ,try to resore that and check.

Cheers..
Binil