IBM SG24-5131-00 User Manual

© Copyright IBM Corp. 1999

Before you start to test the HACMP configuration, you need to guarantee that 
your cluster nodes are in a stable state. Check the state of the:

  • Devices

  • System parameters

  • Processes

  • Network adapters

    • LVM

    • Cluster

  • Other items such as SP Switch, printers, and SNA configuration

Here is a series of suggested actions to test the state of a node before 
including HACMP in the testing.

  • Run 

diag -a

 in order to clean up the VPD.

  • Look in the errorlog for unusual errors by issuing the command 

errpt | 

more

 or 

errpt -a | more

.

  • Check that all devices are in the available state (

lsdev -C | more

).

  • Check that the SCSI addresses of adapters on shared buses are unique 

(

lsattr -E -l ascsi0

).

  • If you are using target mode SCSI networks, check the connection by 

issuing 

cat < /dev/tmscsi#.tm

 on the first node and 

cat /etc/hosts > 

/dev/tmscsi#.im

 (enter twice!) on the second node where 

#

 is the 

appropriate tmscsi device number. Repeat the test in the other direction. 
Note that cluster services must be stopped on both nodes to perform this 
test.

  • To check a serial line between two nodes type 

stty < /dev/tty#

 on both 

nodes where 

#

 is the appropriate tty device number for the RS232 

heartbeat connection. Note that cluster services must be stopped on both 
nodes to perform this test.