cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
891
Views
10
Helpful
3
Replies

BGP Multi-Homing Default route Issues

Meddane
VIP
VIP

The customer is connected to an ISP through two links to provide redundancy through two provider routers.
To achieve primary and backup path requirement, the customer decides to use a default route on the primary path R2 and a FLOATING default route on the backup R3, then redistribute these default route into the IGP that the customer run within its network.

To ensure the return traffic will always go through the primary path, the Provider decides to use a static route toward the customer public address space 209.165.200.0/24 as shown in the topology and redistribute these static routes into BGP :
The static route on R5 has an AD 254, making the redistributed IBGP route toward 209.165.200.0/24 from R4 better, so that R5 will always send the return traffic to R4 through the primary path.

After throubleshooting, R4 detects a link failure of the prim path, R5 now decide to insert the static floating route in its routing table as expected.
A few hours after this failure, the primary path R1-R4 comes back, after verification you notice after the floating route is inserted, it is never removed from the routing table even if the primary link comes back which make this issue undesirable for customer.

How to explain this behabvior and how to fix it?

Screen Shot 2019-05-30 at 6.28.23 PM.png

 

 

1 Accepted Solution

Accepted Solutions

we need to reduce the weight 32768 of the locally sourced route on R5 to be equal to 0 because the IBGP route received from R4 does not have a weight value, and then set the local preference to be lower than the default local preference of the IBGP route received from R4.

View solution in original post

3 Replies 3

Giuseppe Larosa
Hall of Fame
Hall of Fame

Hello @Meddane ,

in BGP a locally generated route either via network command either via redistribute has weight 32,768.

So R5 BGP route has this weight and it is preferred to the iBGP route coming from R4.

 

solution:

under router bgp

neighbor   <R4>  weight 40000

 

Hope to help

Giuseppe

 

we need to reduce the weight 32768 of the locally sourced route on R5 to be equal to 0 because the IBGP route received from R4 does not have a weight value, and then set the local preference to be lower than the default local preference of the IBGP route received from R4.

Hello @Meddane ,

your solution is not so different then mine. By the way I tested it 20 years ago and it works and it is simpler then yours.

Next time if you have already the answer for your question don't open a thread or provide your suggested solution in the first post.

 

Edit: multiple solutions are possible for the same problem, but the root cause is the default weight value of locally generated routes in BGP.

 

Hope to help

Giuseppe

 

Review Cisco Networking for a $25 gift card