[c-nsp] converting mp-ibgp full mesh to dual redundant route reflectors

Aaron aaron1 at gvtc.com
Thu Apr 19 17:30:25 EDT 2012


Is this normal/expected when converting from full mesh mp-ibgp to dual
redundant rr's ?

this is output from one of my pe's learning routes from my dual hub pe's (.1
and .2)   (.4 and .8.2 are just two other pe's)

sh bgp vpnv4 u al sum
...

10.101.0.1      4        65000    2540    2700    11707    0    0 1d16h
150
10.101.0.2      4        65000    2454    2689    11707    0    0 1d16h
179
10.101.0.4      4        65000    2660    2929    11707    0    0 1d20h
1
10.101.8.2      4        65000    1451    1442    11707    0    0 21:41:46
3

 

---- now after config'ing dual RR at dual hub pe's (.1 and .2) I see the
following...

 

sh bgp vpnv4 u al sum
...

10.101.0.1      4        65000      40      18     6116    0    0 00:11:04
333
10.101.0.2      4        65000      35      13     6116    0    0 00:07:33
333

 

 

150 + 179 + 1 + 3 = 333

 

So is it normal to see all available routes *from both* of my rr's?

 

Also, is this the only thing needed on the rr side?  I did this on both hub
pe's (.1 and .2)  I just want to make sure that adding the cluster id
globally to bgp is the right thing to do.  ( pretty much only use my pe's
for vpnv4 mpls l3vpn ... currently not doing any ipv4 native between my
pe's)

 

router bgp 65000

bgp cluster-id 1

commit

 

(of course also I specified on the two hub pe's that all my neighbors are
route-reflector-client)

 

Aaron

 

 



More information about the cisco-nsp mailing list