IBM SG24-5131-00 User Manual

Cluster Management and Administration 

prevents unpredictable behavior from corrupting the data on the shared 
disks. See the 

clexit.rc

 man page for additional information.

Use the 

/usr/sbin/cluster/etc/rc.cluster 

script or the 

startsrc

 command to 

start 

clinfo

 on a client, as shown below:

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

You can also use the standard AIX 

startsrc

 command: 

startsrc -s clinfo

Use the standard AIX 

stopsrc

 command to stop 

clinfo

 on a client machine:

stopsrc -s clinfo

8.2.4.1  Maintaining Cluster Information Services on Clients
In order for the clinfo daemon to get the information it needs, you must edit 
the /usr/sbin/cluster/etc/clhosts file. As installed, the clhosts file on an 
HACMP client node contains no hostnames or addresses. HACMP server 
addresses must be provided by the user at installation time. This file should 
contain all boot and service names or addresses of HACMP servers from any 
cluster accessible through logical connections with this client node. Upon 
startup, 

clinfo

 uses these names or addresses to attempt communication 

with a 

clsmuxpd

 process executing on an HACMP server.

An example list of hostnames/addresses in a clhosts file follows:

n0_cl83

#  n0 service

n2_cl83

#  n2 service

n3_cl83

#  n3 service

For more detailed information on the clinfo command refer to Chapter 2, 
“Starting and Stopping Cluster Services”, 

HACMP for AIX, Version 4.3: 

Administration Guide, SC23-4279.

Never use the 

kill -9

 command on the 

clstrmgr

 daemon. Using the 

kill

 

command causes the 

clstrmgr

 daemon to exit abnormally. This causes the 

SRC to run the /usr/sbin/cluster/utilities/clexit.rc

 

script which halts the 

system immediately, causing the surviving nodes to initiate failover.