telnet - strange issue

[joe_91]

We are having a strange issue. we have two identical nclass boxes in which an
application launches a telnet script which launches the application on a remote HP box.
I have verified that the 3 boxes (2 nclass and the remote HP) have sufficient telnet
kernel parameters enabled. this was working fine until recently the application added
more users (around 100 from 50) and suddenly the application from one nclass server
would not launch, but continues to be fine from other nclass. apart from the three telnet
based kernel parameters (npty, nstrtel, nstrpty) are there anything i should check
on the failing nclass? please advice.

Thanks

Joe.

[A. Clay Stephenson]

Have you seen anything in syslog? Have you seen any can't fork messages? It's possible that you are hitting other kernel limits such as nproc or nfile. Are many of these sessions using the same login? If so, you are probably hitting maxuprc. It's also possible you are having problems with a full filesystem or insuffiecient swap space. You really don't provide sufficient data for anything more than using "The Force".

[Marvin Strong]

As Clay, said we need a bit more information.
And exact error message from either syslog, or perhaps the application logs would help. somewhere there is a error being created that will should give an idea as to the problem.

If any of your kernel parameters are being hit you should see that in the syslog.

Did you reboot or manually create the ptys after setting the kernel parameters larger? You can do this with mknod if you haven't.

[Tim Nelson]

After increasing npty, nstrtel, nstrpty did you execute insf -n xxxx

You can count the installed telnets by ls -ld /dev/tty* and ls -ld /pty*/*

If you are short on pty the only error you would see is at the client side, he who started the telnet session, the message is quick but says "cannot create pty".