Main BGP route and second static route

[reloaded]

Hello guys,

I am testing righ now the hybrid network on one of my branches.

For exemple, i have on the branch 192.168.1.0/24:
- Main accès: On MPLS (Operator A)
- 2nd accès: Standby link over VPN IPsec (Internet)

When the main access is down on the branch 192.168.1.0/24, the access is switched to the 2nd access (IPSec) and it is working good.

On my router on data center, i have 2 IP route to access the branch 192.168.1.0/24:
- 1st route: BGP route learned from Operator A MPLS
192.168.1.0/24 [20/0] via 10.29.52.20
- 2nd route: Static route added manually with a higher metric (25)
192.168.1.0/24 [25/0] via 10.29.52.30

Consequently, on the datacenter router, the 2nd route becomes actif only when the main access is down and the 1st route no more broadcasted by the operator A to my data center router.

However when the main access is established again i am not able to retrieve my first BGP route broadcasted by the operator A:
192.168.1.0/24 [20/0] via 10.29.52.20

The 1st route come back again only when i delete my 2nd route manually from the router. which is very strange beacause i would like to automated this part.

Did i miss something in my router configuration ?

Thank you for you help and advices.
Bassel

[Julio E. Moisa]

Hi

I think you could configure fall-over on the BGP peering. 

[reloaded]

all what i want is to priorise the BGP route in face of secondary static route when the MPLS access is restablished. i dont know very well the fall over on BGP but i will take a look.
Thank you.

[Jon Marshall]

Are you redistributing static into BGP on your DC router ?

Jon

[reloaded]

Yes off course, the static route is redistributed to the operator through BGP to permit trafic between branchs. 

Should i change something here ?  

[Jon Marshall]

It is to do with weight and the higher the weight the more preferred the route. 

Routes received from an EBGP peer have a weight of 0 whereas locally generated routes have a weight of 32768. 

The static is redistributed into BGP and gets a weight of 32768.

When the the branch comes back up and advertises the BGP route the DC gets it but it has a weight of 0 which means the DC router prefers the redistributed static. 

Usual way way to fix this is to make the weight of the received route(s) from the MPLS PE device > 32768. 

Jon

[reloaded]

so if i understand, a change should be done on the operator side on the PE MPLS in order to change the weight of the sended BGP routes.

Right now all my recieved BGP routes on my DC router look like this:
B 10.26.39.0/24 [20/0] via 10.29.52.20, 4w5d
B 10.26.40.0/24 [20/0] via 10.29.52.20, 1w0d

With 0 as default weight, no ?

[Jon Marshall]

You need to do a "sh ip bgp" to see the weight and it is a change you make on your router not on the PE device. 

Jon