RE: BGP neighbor connections

From: Jason Graun (jgraun@xxxxxxxxxx)
Date: Mon Nov 26 2001 - 16:43:47 GMT-3


   
Just to clarify the ebgp-multihop doesnt effect neighboring in the same
AS. It is only for ebgp hence the name of the command.

Thanks

Jason

-----Original Message-----
From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
Vega, Juan R, SOBUS
Sent: Monday, November 26, 2001 10:08 AM
To: 'Matt Smith'; Ben-Shalom, Omer; ccielab@groupstudy.com
Subject: RE: BGP neighbor connections

Exactly,
It is obvious that you have configured the iBGP neighbor statements to
the
peer routers loopback interface. I know this because if you used the ip
address of the physically connected link you would not need a default
route.
You would have a directly connected route. Since you are using the
loopback
address you must specify the neighbor x.x.x.x update-source lo0 command.

-JV-

-----Original Message-----
From: Matt Smith [mailto:matt-n-donna@cablespeed.com]
Sent: Saturday, November 24, 2001 8:20 PM
To: Ben-Shalom, Omer; ccielab@groupstudy.com
Subject: Re: BGP neighbor connections

Both iBGP and eBGP require that both sides of the relationship have a
specific neighboir statement for the other side. By specific I mean a
neighbor X.X.X.X where X.X.X.X mathes exactly the source address in the
TCP
packet it recieves from the other side. If the routers have more than 1
way
to route to one another than you must keep in mind that by default the
TCP
messages leaving each router will be created with a source address equal
to
the exiting interface. So if you enter a neighbro statement for a
remote
router and enter an IP address different than the exact address that the
remote router will replay with the neighbor relationship will not form.
I
hope this provides some help.

Matt Smith
----- Original Message -----
From: "Ben-Shalom, Omer" <omer.ben-shalom@intel.com>
To: <ccielab@groupstudy.com>
Sent: Saturday, November 24, 2001 8:34 AM
Subject: BGP neighbor connections

> Hi all.
>
> I am doing ccbootcamp lab8 and ran across a problem that I solved but
I am
> not sure why the problem is there to begin with (although I have a
feeling
I
> am missing something elementary).
>
> When forming ibgp peering between two routers the adjacency is never
formed
> , debugging BGP gets me
>
> >From R1:
> BGP: 137.20.60.1 open active, local address 137.20.25.1
> BGP: 137.20.60.1 open failed: Connection refused by remote host
>
> >From R6:
> 2d02h: BGP: 200.200.200.1 went from Idle to Active
> 2d02h: BGP: 200.200.200.1 open active, delay 7272ms
> 2d02h: BGP: 200.200.200.1 open active, local address 137.20.64.6
> 2d02h: BGP: 200.200.200.1 open failed: Connection refused by remote
host
> R6 is an RR client of R1
>
> after the change
>
> R1:
> BGP: 137.20.60.1 open active, local address 200.200.200.1
> BGP: 137.20.60.1 went from Active to OpenSent
> BGP: 137.20.60.1 sending OPEN, version 4
> BGP: 137.20.60.1 OPEN rcvd, version 4
>
>
> R6:
> 2d02h: BGP: 200.200.200.1 rcv message type 1, length (excl. header) 10
> 2d02h: BGP: 200.200.200.1 rcv OPEN, version 4
> 2d02h: BGP: 200.200.200.1 went from Connect to OpenSent
> 2d02h: BGP: 200.200.200.1 sending OPEN, version 4, my as: 2
> 2d02h: BGP: 200.200.200.1 rcv OPEN w/ OPTION parameter len: 0
> 2d02h: BGP: 200.200.200.1 went from OpenSent to OpenConfirm
>
>
> changing the BGP update source on R1 to the loopback fixes
everything,
why
> ?
> I did not see any reference to any requirement on the ip from which a
BGP
> speaker tries to form iBGP connections.
>
>
> Any insight is welcome.



This archive was generated by hypermail 2.1.4 : Fri Jun 21 2002 - 06:45:23 GMT-3