Have a Digi CM 32
connected as a console server for Solaris 8 and 10 Sun Blade 2000/2500 workstations
Getting the port access menu via ssh works fine
and logging in to the Solaris machine works fine.
But if the TCP connection is lost, the Solaris login session is not.
Connecting to that same port again grants access to the previous sessions prompt.
It there a way to configure this such that a lost TCP connection will cause the login session to terminate and subsequent port connects get supplied with a fresh login prompt?
This DigiCM is configured such that all access is to be via ssh -p 7000 digihostname and then select the serial port from the connection menu.
Here is a way to do it:
eeprom #diag boot causes OBP messages to go to serial console
diag-level=min
diag-device=disk
diag-switch?=true #default tty port settings
ttya-rts-dtr-off=false
ttya-ignore-cd=true
ttya-mode=9600,8,n,1,- #change console to ttya
input-device=ttya
output-device=ttya
saf #here is the key piece #reconfigure ttymon to turn off software carrier detect
pmadm -r -p zsmon -s ttya
pmadm -a -p zsmon -s ttya -f u -i root -v ttyadm -V
-m “ttyadm -T tvi925 -I -d /dev/term/a -l console -m ldterm,ttcompat \ -p \"console login: \" -s /usr/bin/login -S n”
DigiCM
set timeout on console menu
set timeout on port inactivity
set DTR behavior to High when open
Now the console menu will be lost and the ssh session to the DigiCM will close after a timeout. An active connection will also timeout after a period of inactivity (both directions inactive) and the ssh session to the DigiCM will close. And anytime the ssh session to the DigiCM is lost any active session from a console port to the associated Solaris machine will be logged off.