Arbitration For Data Integrity in Serviceguard Clusters, July 2007

Arbitration for Data Integrity in Serviceguard Clusters

Use of a Lock LUN as the Cluster Lock

The lock LUN is similar to the HP-UX cluster lock disk but different in

certain respects. As with the lock disk, a lock LUN is marked when a

node obtains the cluster lock, so that other nodes will see the lock as

“taken.” This mark will survive an off-on power cycle of the disk device

unlike SCSI disk reservations. As with HP-UX, the lock LUN can be

used with clusters of up to four nodes. The lock LUN is not mirrored.

Here are the important differences between the lock disk in HP-UX and

the lock LUN in Linux or HP-UX:

• Only a single lock LUN can be configured. Dual cluster locking with

lock LUN is not supported. Therefore for extended-distance disaster

tolerant configurations a Quorum Server is required.

• The lock LUN is created on a Linux partition, or HP-UX partition or

disk, directly, not through LVM. The lock LUN is not part of a volume

group.

• The lock LUN partition is dedicated for cluster lock use; however, in

Linux clusters, and HP-UX Integrity clusters, other partitions on the

same storage unit can be used for shared storage. A lock LUN

requires about 100 KB.

In clusters consisting of HP-UX Integrity servers only, you can use

the idisk (1m) utility to create a partition for the lock LUN. In

clusters that include HP 9000 servers, you must use an entire disk or

LUN. On Linux systems, use the fdisk command to define the

partition as type Linux (83).

Serviceguard periodically checks the health of the lock LUN and writes

messages to the syslog file when a lock LUN fails the health check. This

file should be monitored for early detection of lock disk problems.