Instead of shutting down my transit BGP neighbor, I was updating my route-maps from and to my transit with this, so that I would send 0 prefixes from me and receive 0 prefixes from him.
route-map TRANSIT-IN deny 10 route-map TRANSIT-OUT deny 10 my BGP config is like this: neighbor x.x.x.x remote-as 1234 neighbor x.x.x.x route-map TRANSIT-IN in neighbor x.x.x.x route-map TRANSIT-IN out After I did these 2 deny lines, my router has gone nuts, starting to drop many many BGP sessions with various peers and customers, mostly with this message: %BGP-3-NOTIFICATION: sent to neighbor y.y.y.187 4/0 (hold time expired) 0 bytes OSPF was going down and up as well. This kept going all the time until after about 1 hour I removed both route-map IN/OUT deny 10 lines, then after a few minutes, everything became stable again. CPU was obviously at 100%: CPU utilization for five seconds: 100%/10%; one minute: 99%; five minutes: 96% PID Runtime(ms) Invoked uSecs 5Sec 1Min 5Min TTY Process 442 34802356 21731575 1601 58.38% 56.35% 56.52% 0 BGP Router 329 938516 1208660 776 15.99% 17.63% 15.58% 0 IP RIB Update 340 227608 1498205 151 3.88% 7.43% 6.58% 0 XDR mcast 563 38626436 284432 135801 3.88% 3.41% 3.53% 0 BGP Scanner 273 5178956 43762732 118 0.85% 0.99% 0.92% 0 IP Input All I wanted to do was to "mute" the BGP session with one of my transits, for testing purpose, without shutting down the BGP session. Router: 6504 with sup720-3bxl on IOS SXI3 What did I do wrong here? I cannot imagine that a simple route-map deny line can do such harm...? Andy _______________________________________________ cisco-nsp mailing list [email protected] https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
