Hello cisc0.ameer,
my previous example using only route tags may not work when BGP is involved. It is good for sure for mutual redistribution between two IGPs.
For BGP the BGP community attribute can take the place of the route tag.
There was a previous post thread about this specific issue.
BGP community is similar to a route tag with the added benefit that multiple BGP communities can be added to a single BGP prefix.
The logic to create the two route-maps is still the same but on the BGP side we are going to set a BGP community instead of a route tag.
router ospf 100
redistribute bgp 100 subnets route-map BGP-into-OSPF2
!
router bgp 100
redistribute ospf 100 route-map OSPF-into-BGP2
We will use BGP community 100:89 to signal BGP routes that have been originated by redistribution of OSPF into BGP.
We cannot match a community value directly so we need to define a community-list a special ACL for BGP standard communities:
ip community-list 1 permit 100:89
The configuration of the new route-maps becomes the following:
route-map OSPF-into-BGP2 deny 10
match tag 179
route-map OSPF-into-BGP2 permit 20
set community 100:89
the logic is reversed for the route-map in the opposite redistribute direction.
Now we match on community using the community list 1 defined above.
route-map BGP-into-OSPF2 deny 10
match community 1
route-map BGP-into-OSPF2 permit 20
set tag 179
This approach should work when BGP is involved.
Paul Driver has provided a clear explanation of the logic under the two route-maps.
Here there is just an update to use BGP community on the BGP side in case matching or setting a route tag is not supported or not effective.
The idea is that we deny = not allow all routes that have been originated on the protocol that should receive the redistributed routes using a match tag for OSPF routes when examined for possible redistribution into BGP, and using a match community when examining BGP routes for possible redistribution into OSPF.
The same configuration has to be performed on both R1 and R2 as the main reason to control mutual redistribution is the presence of two ASBR routers both performing mutual redistribution.
As noted in previous post a similar configuration can be created for routers R3,R4 where we just need to remember that EIGRP requires the setting of a default metric using the five metric components to make redistribution effective to be able to create the seed metric for EIGRP external routes.
Edit:
the use of the route-maps is really necessary if R1 and R2 have also an iBGP session between them.
If the iBGP session is not present R1 and R2 are not going to accept the routes originated by the other one because going via R3 or R4 the AS path BGP attribute will contain the BGP AS 100.
However, because also R3 and R4 are performing mutual redistribution between BGP and EIGRP without any control R3 or R4 could re-inject an original OSPF route via eBGP taking the prefix from EIGRP and in this case the BGP AS path attribute would contain only AS 200 and would be accepted on R1 or R2.
So actually the use of the route-maps is highly recommended/necessary even if no iBGP sessions between R1,R2 and R3,R4 are configured.
Hope to help
Giuseppe
