A.12.2 When building a cluster system between guest OSes on two host OSes (PRIMEQUEST 1000 Series)

System configuration

Note

Set the host OS as an NTP server and the guest OS as an NTP client, and then perform time synchronization.
If you plan to operate guest OSes on multiple host OSes as clusters, use NTP for time synchronization on host OSes as well.
If you want to use the system as a test environment, you can configure it so that the redundant administrative LAN is created on the guest OS by GLS.
Assign more than two virtual CPUs to each guest OS which comprises the cluster system.

Used hardware

PRIMEQUEST 1800E x 2
FC-Switch x 2
LAN card x 14
Fibre channel card x 16
Disk array unit x 2

Used software

Red Hat Enterprise Linux 5.4
PRIMECLUSTER 4.3A00
Multipath Driver
Microsoft Internet Explorer 8 (installed on the operation management personal computer)
J2SE (TM) Runtime Environment Version 6 (installed on the operation management personal computer)

System disk settings

Host OS	Node name	Virtual disk name	Slice number	Slice name	Size (MB)
host01	primecl01	hda	1	/	15374
			2	/boot/efi	256
			3	swap	2048
	primecl11	hda	1	/	15374
			2	/boot/efi	256
			3	swap	2048
host02	primecl02	hda	1	/	15374
			2	/boot/efi	256
			3	swap	2048
	primecl12	hda	1	/	15374
			2	/boot/efi	256
			3	swap	2048

Shared disk array settings

Device name: ETERNUS	Number of LUN	Number of RAID groups	RAID level	Size
Device name: ETERNUS	6	6	5	248.2GB

Note

The local class disks or shared class disks used by GDS on the guest OS should be created as virtual SCSI devices if they are used in the virtual machine environment.

Note that you cannot change the virtual SCSI device configurations after creating the local class or shared class. To change the configurations, you need to first delete all the GDS settings on the guest OS, change the virtual SCSI device configurations, and then configure the GDS settings on the guest OS again.

Shown below is an example of the Cluster Configuration Worksheet.

When the cluster name is PRIMECLUSTER1