If the route filters aren't implemented correctly, the BGP session won't function properly. The session might remain up, but it will not be able to exchange routes or critical routes will not get advertised. If you haven't done so already, look into the route filters section of this tutorial.
If you suspect that your route filters are blocking your neighbor's BGP announcements, use the following commands to add the soft-reconfiguration command to the BGP neighbor session in order to see what routes they are truely sending.
conf t
router bgp <as number>
neighbor x.x.x.x soft-reconfiguration
inbound
The above commands allow you to see what announcements are actually coming in without removing your route filter and getting flooded with route announcements. To see ALL the routes the neighbor is announcing to you, type the following command:
show ip bgp neighbor x.x.x.x received-routes
If you don't see the routes after setting up soft-reconfig and checking for received routes, THEY AREN'T BEING ANNOUNCED! The administrator of the neighboring router has some work to do.
Keep in mind that altering a BGP session configuration on a Cisco router with IOS prior to 12.0 will cause the BGP session to reset and cause the connection to drop. Purposely resetting the session does not count as a route flap in the local session between you and your ISP. However, your route may flap on the Internet while work on the BGP session is going on.
Practical experience has shown that routes tend to propagate to the major route servers after the session has been up and stable for about 15-20 minutes. --InetD