Simpler explanation of tcp keepalive parameters

[Ed Hon]

I read S1100002433B and am confused. Is a simpler explanation possible? I understand tcp_keepalive_interval, but I get lost on the secondary timers. For instance, suppose tcp_keepalive_interval = 15 and tcp_ip_abort_interval = 10. Does that mean a total of 25 minutes must past before a connection is determined invalid and is terminated? TIA

[Berlene Herren · Accepted solution]

If a TCP packet was retransmitted again and again, it first triggers the R1 timer (tcp_ip_notify_interval).
If the retransmission goes on it will reach the R2 timer which will close the whole connection.
This is done due to the fact that a huge amount of retransmissions indicates that the complete connection is not working anymore.
R2 must always be bigger than R1, otherwise we would drop the connection before we could trigger the Dead Gateway Detection.

Berlene

[Ed Hon]

What parameter sets the interval between the individual retransmissions?