tphipps
asked on
Mysterious loss of telnet session connections
Got Solaris 8 (with latest 8_Recommended patches) running on a Sun Blade 100. Mysterious thing is that on this box only (there are a couple other nearly identical machines where I don't have this problem), I'm having a very hard time keeping idle telnet sessions open. Once left unattended, the telnet session abruptly disconnects after somewhere between 10 minutes and 5 hours of idle time. I haven't been able to pinpoint what causes the variance in time, but it definitely happens each time even though the actual timing is inconsistent.
The users in question have /bin/sh as their shell and no TIMEOUT or similar environment variable set.
Any suggestions as to why these sessions are being disconnected and how to trace back the cause of the disconnect? Happens when directly connected to the machine via the local ethernet segment, remote connection via dialup and via VPN.
The users in question have /bin/sh as their shell and no TIMEOUT or similar environment variable set.
Any suggestions as to why these sessions are being disconnected and how to trace back the cause of the disconnect? Happens when directly connected to the machine via the local ethernet segment, remote connection via dialup and via VPN.
ahoffmann
Solaris has a TCP-timeout for idle connections which is 2 hours, usually. Not shure if this applies to Solaris 8 too.
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
Thanks for the suggesions yuzh.
My /etc/saf/zsmon/_pmpipe file only references the hardwired serial ports /dev/ttya and /dev/ttyb. It doesn't reference any of the pseudo ttys created for telnet sessions. I was under the assumption that entries in this file are only for physically connected devices. Am I wrong? Should I try to create entries for the pts/n devices?
Here's my contents of the file
# VERSION=1
ttya:u:root:reserved:reser
m,ttcompat:ttya login\: ::tvi925:y:#
ttyb:u:root:reserved:reser
m,ttcompat:ttyb login\: ::tvi925:y:#
My /etc/saf/zsmon/_pmpipe file only references the hardwired serial ports /dev/ttya and /dev/ttyb. It doesn't reference any of the pseudo ttys created for telnet sessions. I was under the assumption that entries in this file are only for physically connected devices. Am I wrong? Should I try to create entries for the pts/n devices?
Here's my contents of the file
# VERSION=1
ttya:u:root:reserved:reser
m,ttcompat:ttya login\: ::tvi925:y:#
ttyb:u:root:reserved:reser
m,ttcompat:ttyb login\: ::tvi925:y:#
It looks like the timeout is set in somewhere else on your system, please don't edit the /etc/saf/*/_pmtab. We'll try
to find the source of the configuration setting.
Have a look at your /etc/default/login file for the shell timeout, if you see
TIMEOUT=300 (or other value), comment it out, and then reboot your system, see how it goes.
to find the source of the configuration setting.
Have a look at your /etc/default/login file for the shell timeout, if you see
TIMEOUT=300 (or other value), comment it out, and then reboot your system, see how it goes.
> .. then reboot your system,
This is UNIX, not M$. You don't need to reboot ;-)
This is UNIX, not M$. You don't need to reboot ;-)
ASKER
I think I've solved this. It appears to be two different factors:
1) Whe connected to the machine (it's at home) from work, the dropped sessions can be blamed on a 600 second idle TCP timeout that our firewall seems to have in place for all outbound sessions.
2) When connected to the machine from home on the local segment, the dropped connections can most often be traced back to VPN connect/disconnects. It seems that about 50% of the time I initiate a VPN (Cisco VPN client) connection, it drops existing TCP connections to local hosts.
Thanks all for the suggions. Giving points to yuzh given that I learned a lot about the /etc/saf* tree in the process!
1) Whe connected to the machine (it's at home) from work, the dropped sessions can be blamed on a 600 second idle TCP timeout that our firewall seems to have in place for all outbound sessions.
2) When connected to the machine from home on the local segment, the dropped connections can most often be traced back to VPN connect/disconnects. It seems that about 50% of the time I initiate a VPN (Cisco VPN client) connection, it drops existing TCP connections to local hosts.
Thanks all for the suggions. Giving points to yuzh given that I learned a lot about the /etc/saf* tree in the process!
ASKER
Thanks Yuzh... This wasn't the solution, but some good info.