One cheesy but easy way to monitor via SNMP could be memory utilization. In
the original scenario, decreasing the routing table from +100k to a single
default route would have freed up a discernible amount of memory. Assuming
your router is dedicated to edge traffic, utilization should be relatively
constant and you can define low utilization thresholds to alarm in SNMP.
-----Original Message-----
--At 10:50 03/07/2001 -0400, Alan Halachmi wrote: >pfs>bgp neighbor max-prefix makes this a lot easier to do now. As soon >pfs>as you go over 75% of the threshold set, the router sends messages >pfs>to syslog. And it will optionally tear the peering down once you >pfs>hit the threshold you set. I know a lot of folks monitor syslog for >pfs>strange events, so adding a monitor for the BGP threshold would be >pfs>easy enough to do. Better than cron logging into the router every >pfs>5/15/60 minutes... > >OK... This suggestion solves the upper-bound issue... What about when your >provider accidentally "inverts" an ACL and starts sending you 1 route >instead of blocking a default. Basically, I just want to be notified if my >BGP sessions drop below 100,000 prefixes. Does IOS currently have this >type of functionality? If not, I might run a perl or shell script. >Either way, it is cron based. > >Best! >Alan > >-- >Alan Halachmi >Wide Area Network Specialist >Ingram Entertainment Network Services >mailto:alan@halachmi.net >http://www.ingramentertainment.com
This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:12:43 EDT