RE: bgp metric

From: stephen skinner (stephenski@gmail.com)
Date: Tue Aug 19 2008 - 03:41:12 ART


Mr Hobbs,

 

Thanks for labbing it up ,

 

I have been playing with your suggestions and it is working for me too now .

 

I did have the "max-paths ibgp 2" command in ,but at the time it wasn't
working for me.

 

I changed the OSPF P-2-MP back to an Broadcast type( as you said) and it
worked fine .

 

I didn't use the loopbacks to peer , I just used the local FA interfaces.

 

I was playing with BGP at the time . And came across this issue ,

 

Weird one , I had not seen that metric statement before !

 

Many thanks for your help ,

 

I owe you one mate J

 

 

#From: Hobbs [mailto:deadheadblues@gmail.com]
Sent: 18 August 2008 02:00
To: stephen skinner
Subject: Re: bgp metric

 

It is interesting, I think this is why:

When you have a point-to-multipoint connection is OSPF, the local router
installs a /32 route for the neighbor. if you look on R4 you should have a
/32 route towards your OSPF point-to-multipoint neighbor. The metric will be
1. The ethernet neighbor will have no such route.

btw, I don't think ospf would really be necessary unless you were peering
with loopbacks in AS 300. Is this what you are doing?

If you want to load balance IBGP then use this, make sure you have the ibgp
keyword.

maximum-paths ibgp 2

I have labbed your setup, this works for me.

hope that helps!

On Sun, Aug 17, 2008 at 12:37 AM, stephen skinner <stephenski@gmail.com>
wrote:

Mr. Hobbs ,

 

Thanks very much for your offer ,

 

Here goes

 

The idea here is that R4 wants to have two identical paths to the networks
that are advertised by R1 ( originally) and passed on By R2 and R3

 

 

 R1 (which originates the routes ) is connected to R2 and R3 via FR

 

R2 and R3 are connected to R4 over Ethernet

 

Sort of like this , there is only 1 R1 but I have put it in twice to
represent the BGP peering

         
         R1 R1

         R3----R4----R2

 

R1 is in AS200

R2, R3, R4 are all in AS300

 

AS300 underlying IGP is OSPF

 

Ethernet Link R3-R4 is configured as "ip ospf network broadcast"

Ethernet Link R2-R4 is configured as " ip ospf network point-to-multipoint
non-broadcast"

 

The bgp statement showing the "metric" of 1 associated with the R2-R4 link ,
but not the R3-R4 link .

 

Also ,what I can't understand is why the metric is 1 ,

 

Is that a dynamic metric , or is it simply set to 1 whenever a metric is
found ?

 

What causes the metric to be assigned to this route ??

is it just the bgp multipath command ?

 

if it was to take it directly from the IGP , it would be more , so why did
it set the metric to 1 only .

 

I apologize in advance for all the questions , this just got me good and
proper the other day .

 

Many thanks

 

From: Hobbs [mailto:deadheadblues@gmail.com]
Sent: 17 August 2008 10:01

To: stephen skinner
Cc: Cisco certification
Subject: Re: bgp metric

 

Well the metric was 1, so the only way you could have an IGP metric of 1
would be RIP with a hop count of 1 or ospf over fastethernet. It would be
hard to make eigrp have a metric of 1 :)

I don't know any links off hand that explain it, but remember you wouldn't
modify this metric in bgp config, it would be in your igp config that you
would alter it. the show command for bgp, I assume, just looks it up in the
route table and displays it for you for informational purposes.

also, if you use "next-hop-self" this metric would probably disappear
because your next hop would be directly connected...but that's just an
assumption right now.

btw, what is your ibgp load balancing scenario? maybe post your topology and
explanation and maybe i can help.

On Sat, Aug 16, 2008 at 7:16 PM, stephen skinner <stephenski@gmail.com>
wrote:

Ahh,

 

Thanks for that ,

 

I am "trying" to do load balancing over IBGP and its not working to well.

 

Yes , I am running OPSF over fast Ethernet ,

How could you tell that from the BGP output , I would really like to know ?

 

Also

I have looked for a link to explain where and how to manipulate this metric

 

Do you know of a link that might help me understand this better ?

 

Many thanks

 

From: Hobbs [mailto:deadheadblues@gmail.com]
Sent: 16 August 2008 23:21
To: stephen skinner
Cc: Cisco certification
Subject: Re: bgp metric

 

That's your IGP metric to the BGP next-hop. It looks like maybe your doing
RIP or OSPF over fast ethernet?

On Sat, Aug 16, 2008 at 3:04 AM, stephen skinner <stephenski@gmail.com>
wrote:

Hello,

Can anyone help ,

Ref the statement below

The metric statement in brackets , next to the neighbour peer ,

Can someone please tell me where this metric comes from ?

   131.10.25.2 (metric 1) from 131.10.25.2 (131.11.122.1)

     Origin IGP, metric 100, localpref 100, valid, internal,
atomic-aggregate

TIA

steve

Blogs and organic groups at http://www.ccie.net



This archive was generated by hypermail 2.1.4 : Mon Sep 01 2008 - 08:15:31 ART