You can bypass haview entirely by adding an enterprise id for HACMP and then
defining traps for the cluster states as described in the HACMP MIB. In
effect, this treats HACMP just like an ordinary but virtual "device" such as
Oracle DBMS and its MIB. Remember to enter the trap destination ipaddr for
NetView in the /etc/snmp.conf file for the HACMP hosts.
Note that in case of a node failover in an HACMP cluster that NetView will show
an alarm avalanche of 26 traps, 13 traps from the node going down and 13 traps
for the standby node coming up.
Also, on HACMP nodes, you can configure an error notification method(s) for
HACMP cluster state events so that only certain cluster state changes send SNMP
traps. This alternative can eliminate the alarm avalanche, i.e. configure an
error notification method just for node down and node up.
Regards,
ken
|