The nodereboots works of the SNMP agent restarting, which if you are managing servers is probably the wrong thing.
I've posted a UnDP and alert definition to use the HOST-RESOURCES MIB
hrSystemUptime.UnDP
hrSystemReboot.AlertDefinition
I'd like NPM to poll for this OID as well as sysUptime
if hrSystemUptime is present then
if hrSystemUptime less than the previous then the alert should be "Node rebooted"
if sysUptime is less than the previous then the alert should be "SNMP agent restarted"
else
if sysUptime is less than the previous then the alert should be "Node rebooted"
this should preserve the type of alerts for older devices that do not support the host-resource MIB and make the alert more meaningful for devices like servers