Nagios comes with a plugin named check_bgp that checks if a set of BGP sessions is up or not. It didn't work for me (Pete Siemsen), so I modified it to create check_bgp_new.

This will tell you that BGP is up, but not whether BGP is "sane" or not. We want to know if BGP routes are being correctly received/advertised by the session. For instance, suppose one of the FRGP's providers erroneously stops advertising routes to the FRGP, or suppose a network problem causes traffic flows to erroneously favor one provider over another.

To provide sanity checking, I further modified check_bgp_new to retrieve BGP counters from a router. This works only with routers that support the Juniper SNMP MIBs. The script retrieves the number of routes that are received from each BGP peer and the number of those routes that are being used. These values are compared against thresholds that we've defined for the peers. If the values exceed the thresholds, the script returns a Nagios WARNING value and text string.

John Hernandez further modified the script to correct performance issues and to use Nagios services.

When you suspect routing problems, run netserver:/usr/web/nets/tools/trcheck/CheckNcarTraceroutes.sh. It runs about 40 traceroute commands and compares the results of each one to the "normal" paths that packets take when the network is healthy. This script provides a quick snapshot of the health of our wide-area links, but must be run manually when you suspect a problem. See http://netserver.ucar.edu/frgp/internal/tools/trcheck/.