[wellylug] SSH timeouts (again).

[Simon Antliff]

On 10/5/05, Cliff Pratt <> wrote:
> Still getting the ssh timeouts on RedHat, but not in a
> parallel situation with Debian.
>
> Someone mentioned earlier that there is a fixed session
> timeout (for TCP/IP?) and that the answer was to set the ssh
> client to send keepalives at periods less than the timeout.
>
> Where is the TCP/IP timeout (if that is what it is) set? If
> I set the client to send keepalives at intervals under the
> timeout what happens when the number of keepalives is reached?
>
> Say the TCP/IP timeout is set to 5 mins (300 seconds). If I
> set the keepalive interval to 60 seconds and the number to
> 5, what happens after 5 keepalives have been sent? What
> kicks off the cycle again?

My apologies, I must not have been clear enough (if you've confused
client with server).

Add (on the server side)

ClientAliveInterval <time in seconds>
ClientAliveCountMax <default is 3>

to your /etc/ssh/sshd_config on the server (Note the ssh*d* bit). If
you dont have permission to this file, then... dunno.


On the client side TCPKeepAlive should be default to 'yes' anyway, so
unless you've explicitly set it to 'no' it wont make a difference.


'man sshd_config' for more sshd config help. It will explain
ClientAliveInterval and ClientAliveCountMax better than I can.


One other thing you should try is to ssh in from somewhere else on the
Redhat servers LAN (if you can) and see if the timeouts are still
occuring.

Simon