ebgp - add inactive routes to routing table

[pkc]

Hi,

For redudancy purposes I'd like to bring inactive eBGP routes (due to lower preference they don't show up).

I tried to activate multipath but it doesn't work. 

Did anyone already manage to have inactive routes shoing up in the routing table, so they can be exported in eBGP ?

thanks,. 

[emnoc]

A diagram would be helpful but if for generics, you have 2 eBGP and they are pushing the same BGP prefixes only one will be in  the FIB as the best path, but both will be in the BGP table. There's no such thing or term as "in active" it's best path and of paths if multiple paths are for the same exact prefixes and no filtering is involved. With multipath same prefixes can be enabled in  the FIB I believe it's still 4 max .

Can you draft a diagram of what your talking about? And yes  bgp multi-path will work but use with cautious due to asymetrical issues.

Also if you have numerious prefixes and unstable  peers or receive alot of updates, your CPU % could be taxed and run hot from our experience. So  use caution and monitor.

Ken

[pkc]

Diagram is really simplified.

Both fortigate clusters learn routes from bottom and from top, and import it with different preferences to avoid asymetric routing.

network engineer asks me if it is possible to have both routes in the routing table, so if one link goes down, the route is already there.

support doesn't seem to think it's supported. if anyone already tried this.

thanks.

[emnoc]

Will these 2 fortigates are in  two different    ASN for one, and yes it should work regardless. But the really question is the route originating from  both 2 different ASN? That's not normal in most  cases.

Also the locl_pref is not going to  do much if the  FGT ASN are different, locl_pref is not transitive attribute that local to that BGP router & ASN.

So if these 2 unique ASN are delivering the route, the locl_pref should be enough to select the HIGH# for the route we pull into the FIB. If you lab this up take a look at your bgp table and see if the 2nd path with inferior  ASN path is present.

Once again in public BGP, the prefix origination should be via one ASN and not 2  3 or 4 etc......

Ken