From: Marcus Jensen (marcus@pobox.com)
Date: Sun Aug 03 2003 - 05:29:26 GMT-3
Shawn, this is not a frame relay issue. It is an underlying serial issue
with any encoding. Heck, the earliest IOS versions didn't even support
assigning ip addresses to serial interfaces.
Try this test to show self pings must exit the interface. Two serial
interfaces back to back any encoding. Ping yourself once. Then go to other
side, put an inbound ACL denying only ping replies using source and
destination of first router. Go back to first router and turn on debug and
ping yourself. You will see ping request exit and come back, then ping reply
exit, but not come back. Remove ACL and it will work again.
I don't know why serial is still this way, it seems Cisco could code a hole
for it, but it is not so entirely unnatural from an ip stack point of view.
IP over isdn/ethernet/atm have similar considerations. IP stack processing
is all the way to the bottom, and all the way to the top. It is all "in/out"
processed. Any deviation from this such as pinging yourself, ip routing,
fast switching, firewalling, is a feature coded into the stack interrupt
processes. Or a bug :)
Marcus
-----Original Message-----
From: nobody@groupstudy.com [mailto:nobody@groupstudy.com]On Behalf Of
Shawn Yang
Sent: Saturday, August 02, 2003 3:00 AM
To: yu chunyan; ccielab@groupstudy.com
Subject: RE: frame map ip: self address?
Hi, John and Yu,
Thanks for getting back to me. I understand the map/inarp issue,
but my point is when you ping, should IOS first check if the
destination is one of it's own addresses, and if it is, there
should be a "shortcut" somewhere in the code. The ping should
not even get to the bottom of the protocol stack. So Frame mapping
should never be an issue here.
And, why does not the same happen on ATM/ISDN, but only for FR.
Sounds like a bug on FR code for me, at least something that can
be optimized. It makes us have to type one more command.
Shawn
-----Original Message-----
From: nobody@groupstudy.com [mailto:nobody@groupstudy.com]On Behalf Of
yu chunyan
Sent: Thursday, July 31, 2003 3:04 PM
To: syang@bitfone.com; ccielab@groupstudy.com
Subject: Re: frame map ip: self address?
Hi,
In multipoint interface, if you donot define a map pointing to self ip
address, frame relay donot know which dlci to send the icmp, and will show
"encapsulation failure". In point-to-point interface, There is only one dlci
over there, frame relay just send the icmp over it. so, even you ping self
ip address, the icmp packet is send to the other end, which will send
response back.
Bin.
>From: "Shawn Yang" <syang@bitfone.com>
>Reply-To: "Shawn Yang" <syang@bitfone.com>
>To: <ccielab@groupstudy.com>
>Subject: frame map ip: self address? Date: Thu, 31 Jul 2003 12:50:24 -0700
>
>Hi, all,
>
>On FR interfaces, we need to map not only the "connected" ip address,
>but also the self ip address, is it right? What's the reason for
>that? I have found that if you don't do that you won't even able to ping
>your own interface.
>
>and this is true only for multipoint interfaces, but not point-to-point.
>
>Thanks for any info!
>
>Shawn
>
>
>
>
>R1 (s0) ------- R2
>
>On R1:
>int s 0
>ip addr 1.1.1.1
>frame map ip 1.1.1.1 102 <---------- without this, you won't able to ping
>1.1.1.1
>frame map ip 2.2.2.2 102 broadcast
>
>
>_______________________________________________________________________
>You are subscribed to the GroupStudy.com CCIE R&S Discussion Group.
>
>Subscription information may be found at:
>http://www.groupstudy.com/list/CCIELab.html
This archive was generated by hypermail 2.1.4 : Tue Sep 02 2003 - 18:53:52 GMT-3