Quote:
Originally Posted by
Balaram
The Cluster node's performance.
The Cluster functionality and communication.
Monitoring a cluster node's performance is not different from monitoring a standalone system. Hence I won't elaborate on that. Search this group to find thousands of posts regarding performance monitoring in AIX.
Monitoring the cluster's functionality is different. You need to know how your cluster works which might be different depending on the HACMP version and release you use. E.g. you could use SNMP easily till HACMP 5.2 but that won't work as well from HACMP 5.3 onwards.
If you want to monitor a cluster by its own means you'd probably start by using clstat. This tool comes as an ASCII and a MOTIF version. If you want to monitor several clusters you could monitor them all in clstat if thoses clusters share a network so that clstat gets the information of local and other clusters.
IIRIC from HACMP 5.2 on you could use
WebSMITto monitor your cluster. This allows to monitor the cluster with the help of a web browser and at the same time administrate the cluster just as with SMIT.
Furthermore you could use monitoring tools other admins wrote and make available via the internet. E.g. check this website:
A. Abderrazag's custom HACMP monitor
Then there are those monitoring tools that cost money: IBM Director or BMC Patrol to name two of the usual suspects. They pretend to be more comfortable but IMHO if you can make your monitor send email to a trouble ticket system you don't need to spend that money.