> This doesn't sound believable to me. It sounds like the > discussion mangled some slightly related topics and applied > them to this situation. ;-) Or that someone misinterpreted the > output from "debug ip icmp." Or perhaps someone jumped to this > conclusion due to the fact that you have to map your serial > interface IP address to a DLCI to ping it in a Frame Relay > environment.
Priscilla, Thanks. Here is one link I crossed on CCO recently: http://www.cisco.com/warp/public/116/fr_faq.html#1 I have found other similar links and discussions over the years. In fact, I was thinking that it might have been you that once explained how it all worked. In any case, I have tried some of the same debug that you have in the past. I assumed that what was going on was somehow hidden from debug. > > To forward a frame a router looks into its routing information > base, (whether that be the routing table, fast cache, or > whatever), and determines how to forward the frame. In this > case, it would see that the destination network is directly > connected and from there it would presmumably check its running > config and see that the destination node is directly connected > also. So there's no need to send the frame out an interface. > > Anyway, I tested it. I tried various serial encaps, including > HDLC, Frame Relay, and PPP. I was consoled into two routers at > once, Router A and Router B, to make it easy to see the output > from "debug ip icmp" on both routers. The router serial > interfaces are connected back-to-back. When I pinged from > Router A's e0 to Router A's s0, Router B did not see the > packets. I don't believe they crossed the serial link. > > To be absolutely sure I would want to use a serial protocol > analyzer, but alas, those are too expensive for the > self-employed. But I'm 99% convinced by the testing that I did. > > Priscilla This is exactly why I have been looking into building my own WAN protocol analyzer. There are a few PCMCIA cards on the market that do synchronous, bit-oriented protocols. Ethereal apparently supports Cisco's perversion of HDLC. But those PCMCIA cards aren't exactly cheap either. In any case, this discussion stemmed from someone's problem whereby he was observing dropped packets on an extended ping from an ethernet interface to a serial interface on the same router. The router CPU and memory was apparently virtually untapped. Yet, he observed packet loss. He specifically asked if pings were required to cross the WAN. I remembered having read about it in the past. Is it possible that I have just misinterpreted it every single time I've read it? You bet! Any thoughts on what CCO is really saying? My brain hurts.... Thanks again, Scott Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=53154&t=53148 -------------------------------------------------- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
