Technical data

11.2 Configuring an Eight-Node Cluster Using an UltraSCSI
BA356 and External Termination
Configuring an eight-node cluster is carried out in three distinct stages,
one stage for each shared SCSI bus:
1. Install member systems 1 and 2 and all associated cluster hardware
needed to place these two systems on a shared SCSI bus.
2. Install member systems 3, 4, and 5 and all associated cluster hardware
needed to place these two systems on a shared SCSI bus with member
system 2.
3. Install member systems 6, 7, and 8 and all associated cluster hardware
needed to place these two systems on a shared SCSI bus with member
system 1.
______________________ Note _______________________
You can switch steps 2 and 3 around and install member systems
6, 7, and 8 before member systems 3, 4, and 5.
11.2.1 Cabling the First Two Nodes on the First Externally Terminated
Shared SCSI Cluster
This section provides installation instructions for the cluster hardware for
the first two nodes of an eight-node shared SCSI bus cluster. Complete the
steps in order. When you are referred to another section or table, complete
those steps completely before returning to this section.
______________________ Notes ______________________
If you are upgrading from Tru64 UNIX Version 4.0F or V4.0G
and TruCluster Memory Channel Software Version 1.6 to Tru64
UNIX Version 5.1A and TruCluster Server Version 5.1A, you are
required to have shared storage, but you do not have to change
your Memory Channel or public network hardware. Therefore,
you can skip those steps except as follows.
If you are using the Memory Channel adapters in multiple-active
rail mode with the TruCluster Memory Channel Software
product, after you have installed the Tru64 UNIX and TruCluster
Server Version 5.1A software, you will have to reset the rm kernel
subsystem configuration rm_rail_style variable to zero. The
default for rm_rail_style for TruCluster Server Version 5.1A is
Configuring an Eight-Member Cluster Using Externally Terminated
Shared SCSI Buses 11–5