Date: Sat, 15 Oct 94 04:30:01 PDT From: Advanced Amateur Radio Networking Group Errors-To: TCP-Group-Errors@UCSD.Edu Reply-To: TCP-Group@UCSD.Edu Precedence: List Subject: TCP-Group Digest V94 #230 To: tcp-group-digest TCP-Group Digest Sat, 15 Oct 94 Volume 94 : Issue 230 Today's Topics: ARP grumbles (3 msgs) Unsubscirbe Send Replies or notes for publication to: . Subscription requests to . Problems you can't solve otherwise to brian@ucsd.edu. Archives of past issues of the TCP-Group Digest are available (by FTP only) from UCSD.Edu in directory "mailarchives". We trust that readers are intelligent enough to realize that all text herein consists of personal comments and does not represent the official policies or positions of any party. Your mileage may vary. So there. ---------------------------------------------------------------------- Date: Fri, 14 Oct 1994 09:34:42 -0800 (PDT) From: jmorriso@bogomips.ee.ubc.ca (John Paul Morrison) Subject: ARP grumbles Why doesn't the ARP cache code in JNOS (1.08): 1. learn the hardware address from incoming packets. it's pretty silly to receive an incoming packet (say an ICMP echo request) then turn around and broadcast an ARP request when the hardware address is right under its nose 2. why doesn't it reset the ARP cache timers when there are incoming packets, instead of letting the timers run down, and have to retransmit an ARP request. In other words, JNOS can hear a packet a few ticks before the ARP timer runs down, but it doesn't reset the timer, so it's forced to do a useless ARP request. Life's bad enough at 1200bps, I'm going to add static ARP entries for the important hosts. --------------------------------------------------------------------------- BogoMIPS Research Labs -- bogosity research & simulation -- VE7JPM -- jmorriso@bogomips.ee.ubc.ca ve7jpm@ve7jpm.ampr.org jmorriso@ve7ubc.ampr.org --------------------------------------------------------------------------- ------------------------------ Date: Fri, 14 Oct 1994 16:30:53 -0400 From: "Louis A. Mamakos" Subject: ARP grumbles > Why doesn't the ARP cache code in JNOS (1.08): > > 1. learn the hardware address from incoming packets. it's pretty > silly to receive an incoming packet (say an ICMP echo request) then > turn around and broadcast an ARP request when the hardware address > is right under its nose Because you don't know if the packet arrived from the source address in the packet or not. It may be been forwarded from an intermediate router. > 2. why doesn't it reset the ARP cache timers when there are incoming > packets, instead of letting the timers run down, and have to > retransmit an ARP request. In other words, JNOS can hear a packet > a few ticks before the ARP timer runs down, but it doesn't reset the > timer, so it's forced to do a useless ARP request. For the same reason above. You don't know what arp entry it's associated with, especially if you play proxy ARP games. ARP caches are not usually designed to be searched in that direction, either. > Life's bad enough at 1200bps, I'm going to add static ARP entries for > the important hosts. Oh well. You could us a V.32bis modem on a dial-up connection and get better throughput until we get better radios and modems. louie wa3ymh ------------------------------ Date: Fri, 14 Oct 1994 11:45:18 -1000 (HST) From: Antonio Querubin Subject: ARP grumbles On Fri, 14 Oct 1994, John Paul Morrison wrote: > Why doesn't the ARP cache code in JNOS (1.08): > > 1. learn the hardware address from incoming packets. it's pretty > silly to receive an incoming packet (say an ICMP echo request) then > turn around and broadcast an ARP request when the hardware address > is right under its nose There is a limited 'arp eavesdrop' feature in the latest versions of JNOS. But to blindly map every incoming IP address to an ARP entry isn't wise. Suppose you were the only host on a net talking to a single default router. Now suppose you had several dozen other hosts connecting to you all through that router. Do you create several dozen ARP entries? At what point does the IP routing table became useless and NOS becomes a bridge rather than a router? Suppose instead of several dozen hosts connecting to you it was several thousand over a period of days or weeks? > 2. why doesn't it reset the ARP cache timers when there are incoming > packets, instead of letting the timers run down, and have to > retransmit an ARP request. In other words, JNOS can hear a packet > a few ticks before the ARP timer runs down, but it doesn't reset the > timer, so it's forced to do a useless ARP request. The 'arp eavesdrop' feature does this for you. But remember that it depends on arp responses heard on the net to reset timers. Antonio Querubin tony@mpg.phys.hawaii.edu / ah6bw@uhm.ampr.org ------------------------------ Date: Fri, 14 Oct 1994 19:36:00 -0400 (EDT) From: Admin Subject: Unsubscirbe unscribe ------------------------------ End of TCP-Group Digest V94 #230 ******************************