Henrik Nordstrom wrote:
Are you sure? I see some ARP traffic even when there is TCP, and can find no trace of this extending of the neighbor entries in the TCP code (but this does not mean it isn't there)..
I'm not really sure what code is being used and what is not. It seems to me that if eth_rebuild_header in net/ethernet/eth.c is called, it will update the used-parameter of the arp entry (done by the call to arp_find), which will delay expiry. But I'm not sure if this is actually called anywhere.
Probably hasn't been considered important to implement for ICMP, if it is implemented for TCP. But I do have a feeling both ICMP and TCP behaves the same in this regard, and my testing indicates this is the case.
I've been testing this with just two nodes and a crossover cable (fairly silent network), and logging with ethereal. It is most visible when the other machine is a Windows box, since they have a far longer arp timeout (with its own problems). When I just ping, I get a new arp request every 30-45 seconds. When I start a management application that uses a http connection (with keepalive, it's the same TCP session all the time) every 20 seconds or so, the arp requests disappear (both with and without the ping in parallell).
Maybe your problem is that the traffic is simply too infrequent causing the entry to expire entirely before the next ICMP is sent?
Nope, I ping every second, and the timeout is far more infrequent. -Harald -- Harald Nordgård-Hansen, AXXESSIT ASA - : send the line "unsubscribe linux-net" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
