IBM SG24-5131-00 사용자 설명서

© Copyright IBM Corp. 1999

This chapter covers all aspects of monitoring and managing an existing 
HACMP cluster. This includes a description of the different monitoring 
methods and tools available, how to start and stop the cluster, changing 
cluster or resource configurations, applying software fixes, user 
management, and other things.

By design, HACMP for AIX compensates for various failures that occur within 
a cluster. For example, HACMP for AIX compensates for a network adapter 
failure by swapping in a standby adapter. As a result, it is possible that a 
component in the cluster could have failed and that you would be unaware of 
the fact. The danger here is that, while HACMP for AIX can survive one or 
possibly several failures, a failure that escapes your notice threatens a 
cluster’s ability to maintain a highly available environment.

HACMP for AIX provides the following tools for monitoring an HACMP cluster:

  • The 

/usr/sbin/cluster/clstat 

utility, which reports the status of key 

cluster components—the cluster itself, the nodes in the cluster, and the 
network adapters connected to the nodes.

  • The 

HAView

 utility, which monitors HACMP clusters through the NetView for 

AIX graphical network management interface. It lets users monitor 
multiple HACMP clusters and cluster components across a network from a 
single node.

  • The 

SMIT Show Cluster Services

 screen, which shows the status of the 

HACMP for AIX daemons

  • The following log files: the

 

/var/adm/cluster.log file, which tracks cluster 

events, the /tmp/hacmp.out file, which records the output generated by 
configuration scripts as they execute, the 
/usr/sbin/cluster/history/cluster.mmdd log file, which logs the daily cluster 
history, and the /tmp/cspoc.log file, which logs the status of C-SPOC 
commands executed on cluster nodes.

When you monitor a cluster, use the 

clstat

 utility to examine the cluster and 

its components. Also, constantly monitor the /tmp/hacmp.out file. Use the 
SMIT Show Cluster Services screen to make sure that the necessary HACMP 
for AIX daemons are running on each node. Finally, if necessary, examine the 
other cluster log files to get a more in-depth view of the cluster status.