I have an MPLS router that connects to the core network.
This router distributes (per route maps) routes from OSPF into BGP and from BGP into OSPF.
The OSPF Process conencts a 6509 to the 7206 MPLS router. There are some routes in the OSPF process that I have filtered out of the 6509. They do not show up inthe 6509 at all and this is the only way they can be getting into the 7206.
Checking the 6509 database, this route is gone, but it stays in the 7206 until I clear the route manually. The result is the route still gets distributed into MPLS.
7206VXR-01#sh ip route 10.10.10.3
Routing entry for 10.10.10.3/32
Known via "ospf 1", distance 110, metric 3020, type extern 1
Redistributing via bgp 65001
Advertised by bgp 65001 route-map OSPF_to_BGP
Last update from 10.1.68.254 on GigabitEthernet0/3, 00:32:21 ago
Routing Descriptor Blocks:
* 10.1.1.254, from 172.16.254.1, 00:20:00 ago, via GigabitEthernet0/3
Route metric is 3020, traffic share count is 1
One thing to note, there are two possible OSPF paths the route gets into OSPF, one of them, the route is filtered with distribute-list on the 6509, which means it is still in the database, so it is still in the 7206 database, and still get distributed into BGP on the 7206, correct?
Perhaps you can give us some additional details about this? When you say that you have filtered out routes on the 6509 what do you mean?
Where did the route come from and how are you filtering it out?
If the route is being learned and you are using something like a distribute list on the 6509 to deny the inbound advertisement then you are encountering one of the interesting behaviors of OSPF. The distribute list is preventing the route from getting into the routing table of the 6509. But it gets into the OSPF data base (you can verify this with sh ip ospf database) and it does get advertised to OSPF neighbors.
OSPF routes cannot be filtered from entering the OSPF database. If you use this command (distribute list like in your case) for OSPF, it only filters routes from the routing table; it does not prevent link-state packets from being propagated.
However, as far as I know, in order to redistribute a route should be in the routing table. In your case, you have two OSPF routes for the same prefix which the one of these routes is on the routing table is redistributed to the BGP.
The OSPF route in database is not redistributed since can not be installed to the routing table.
Vasilis has rightly pointed out that the ospf filtering only stops the routes being installed into the RIB and not the LSDB. You dont't filter the LSDB's as that forms the fundamental for SPF calculation. If the LSDB's are tampered then the very basic idea of topology state routing is defeated. Does this make sense?
You see the prefix on the 7206 because ospf has run its SPF on the LSDB on the 7206 and found the best path for it and hence installs it in the routing table and then redisttributes into BGP.
We are pleased to announce availability of Beta software for 16.6.3. 16.6.3 will be the second rebuild on the 16.6 release train targeted towards Catalyst 9500/9400/9300/3850/3650 switching platforms. We are looking for early feedback from custome...