IBM SG24-5131-00 User Manual

IBM Certification Study Guide  AIX HACMP

At this point in time, the cluster is not yet started. So the cluster manager has 
to be started first. To check whether the cluster manager is up, you can either 
look for the process with the 

ps

 command:

ps -ef | grep clstr

or look for the status of the cluster group subsystems:

lssrc -g cluster

or look for the status of the network interfaces. If you have IP Address 
Takeover (IPAT) configured you should see that the network interface is on its 
boot address with the 

netstat -i 

command. 

Then start HACMP through 

smit clstart. 

In the panel that appears, choose 

the following parameters and press Enter:

1. start now

2. broadcast message true

3. start cluster lock services false

4. start cluster information daemon true

Reissue either the 

ps

 command (see above) or look for the interface state 

with the 

netstat -i 

command. Now, you should see that the boot interface is 

gone in favor of the service-interface.

You also would like to check whether a takeover will work, so, you have to 
bring up HACMP on all cluster nodes through 

smitty clstart

 and check 

whether the cluster gets into a stable state. Use 

clstat

 for this purpose.

When the cluster is up and running, stop one of the node’s cluster managers 
with 

smitty clstop

 and choose graceful with takeover. One possibility to 

check whether the takeover went through smoothly is to look at the 
/tmp/hacmp.out file during the takeover, preferably on the takeover node. You 
can use the 

tail -f /tmp/hacmp.out

 command for this. 

After the cluster has become stable, you might check the 

netstat -i

 output 

again to verify that the takeover node has acquired the IP address of the 
“failed” node.