IBM SG24-5131-00 사용자 설명서

Cluster Management and Administration 

are started in sequential order - not in parallel. The output of the command 
run on the remote node is returned to the originating node. Because the 
command is executed remotely, there can be a delay before the command 
output is returned.

8.2.2.1  Automatically Restarting Cluster Services
You can optionally have cluster services start whenever the system is 
rebooted. If you specify the 

-R

 flag to the 

rc.cluster 

command, or specify 

restart or both in the 

Start Cluster Services SMIT screen, the 

rc.cluster 

script adds the following line to the /etc/inittab file. 

hacmp:2:wait:/usr/sbin/cluster/etc/rc.cluster -boot> /dev/console 2>&1

# Bring up Cluster

At system boot, this entry causes AIX to execute the 

/usr/sbin/cluster/etc/rc.cluster

 script to start HACMP Cluster Services.

8.2.2.2  Starting Cluster Services with IP Address Takeover Enabled
If IP address takeover is enabled, the

 /usr/sbin/cluster/etc/rc.cluster

 script 

calls the 

/etc/rc.net 

script to configure and start the TCP/IP interfaces and to 

set the required network options.

You stop cluster services on a node by executing the HACMP 

/usr/sbin/cluster/etc/clstop

 script. Use the 

HACMP for AIX Stop Cluster 

Services SMIT screen to build and execute this command. The 

clstop

 script 

stops an HACMP daemon or daemons. The 

clstop

 script starts all the cluster 

services or individual cluster services by calling the SRC command 

stopsrc

. 

Using the C-SPOC utility, you can stop cluster services on a single node or on 
all nodes in a cluster by executing the C-SPOC 

/usr/sbin/cluster/utilities/cl_clstop

 command on a single node. The 

C-SPOC 

cl_clstop

 command performs some cluster-wide verification and 

then calls the 

clstop

 command to stop cluster services on the specified 

nodes. The nodes are stopped in sequential order—not in parallel. The output 
of the command that is run on the remote node is returned to the originating 

Be aware that if the cluster services are set to restart automatically at boot 
time, you may face problems with node integration after a power failure and 
restoration, or you may want to test a node after doing maintenance work 
before having it rejoin the cluster.