icmp node monitor and the connection table

Why do icmp monitors create connection table entries, with a default timeout of 300 seconds? With 40+ nodes and icmp every 5 seconds, this produces 2000+ spurious connection table entries, artificially bloating statistics.

 

 

I've been unable to figure out a way to configure a idle timeout for an icmp connection table entry. I get that it might be necessary to have an icmp entry in a connection table if you route all traffic for a node, but we aren't set up that way and would be happy without any icmp entries at all.

 

 

Why I shouldn't `b conn protocol icmp delete` every second, or how can I change the default connection table timeout for icmp?

 

 

Kernel:

 

Linux 2.4.21-9.4.4.65.0smp

 

Package:

 

BIG-IP Version 9.4.4 65.1

 

Final Edition