Re: Difference between ServerAliveInterval and TCPKeepAlive?
From: Richard E. Silverman (res_at_qoxp.net)
Date: 01/23/05
- Previous message: David Deutsch: "Difference between ServerAliveInterval and TCPKeepAlive?"
- In reply to: David Deutsch: "Difference between ServerAliveInterval and TCPKeepAlive?"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
Date: 23 Jan 2005 01:47:10 -0500
>>>>> "David" == David Deutsch <sptest@71broadway.info> writes:
David> Could someone give me a quick explanation of the difference
David> between these two (I have already read the man page)? I'm
David> trying to a) keep my NAT from timing out my connection, and B)
David> terminate the process asap if the NAT *does* time it out (so
David> that I can detect it and restart the process).
TCPKeepAlive is not SSH-specific; it simply sets the "keep-alive" option
on the underlying TCP connection. TCP keepalive is not intended to keep
anything alive, as in avoiding NAT timeouts. Rather, it detects and reaps
half-dead TCP connections to prevent them from building up over time. It
typically operates over too long a time scale to be useful in defeating
NAT timeout, and that scale is usually only tunable for the whole TCP
stack, not per-connection.
ServerAliveInterval controls keepalive messages which are sent within the
SSH protocol; since they are application-level and hence tunable
per-connection, this is what you want to use.
-- Richard Silverman res@qoxp.net
- Previous message: David Deutsch: "Difference between ServerAliveInterval and TCPKeepAlive?"
- In reply to: David Deutsch: "Difference between ServerAliveInterval and TCPKeepAlive?"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
Relevant Pages
|
