Hi, Attached is a patch against Ganglia monitor-core 2.5.1 which adds support to gmetad for tracking the state of hosts and triggering an alarm when a host goes down or comes up again.
Two new configuration parameters are added to gmetad.conf; death_threshold, which is the timeout for considering a host down, and hostalarm_handler, which specifies an executable file that is called with cluster name, hostname and host state as arguments when a host changes state. No default alarm handler is supplied, but the one I'm using right now is basically: #! /bin/sh echo "$1/$2 is $3!" | mail -s "Host alarm!" [EMAIL PROTECTED] -- Leif Nixon Systems expert ------------------------------------------------------------ National Supercomputer Centre Linkoping University ------------------------------------------------------------
hostalarm.patch.gz
Description: GNU Zip compressed data