The RA8000/ESA12000 is typically configured with two HSG80 controllers
in transparent failover mode, providing failover/redundancy to ensure high
availability. Each controller has two ports designated Port 1 (left) and Port 2
(right). In a typical redundant failover configuration, Port 1 on Controller A
(upper controller) and Port 2 (right port) on Controller B (lower controller)
are the only two active ports.
If all four ports of the HSG80 are connected to the Fibre Channel switch
or hub, then, if a controller module fails, the surviving controller assumes
control of both the Port 1 and Port 2 ports. Each port has its own worldwide
name so that during typical operation, only two ports would be logged into
the Fibre Channel switch (or hub) and visible when doing a
switchShow
command in a fabric configuration. If one or the other controller fails,
the surviving controller logs into the Fibre Channel switch or hub, and a
subsequent
switchShow
command appears as though the worldwide name
had moved from one Fibre Channel port to another. This, in itself, does not
affect the Tru64 UNIX software configuration.
However, to add a new RA8000 or ESA12000, you must follow these steps:
1.
Configure the disks on the new HSG80 controllers into units.
Performing this step first allows the various devices to be usable when
the member systems are rebooted.
2.
If this is an arbitrated loop configuration, set up the HSG80 for
arbitrated loop, as described in Section 4.3.1.
3.
Connect the new controllers to the switch.
____________________
Notes
____________________
Do not add more than one RA8000/ESA12000 at the same
time to an existing cluster.
The
emx
driver on each member system will recognize the
presence of the new HSG80 controller(s) and assign target
IDs. If the data bases are synchronized and have empty slots,
the target IDs assigned by each system should be the same.
4.
At the HSG80 controller, make sure that the connection for the new
adapter’s worldwide name is
TRU64_UNIX
.
5.
On each member system, run the
emxmgr -t
command to verify
that the worldwide name-to-target ID mappings are identical. If the
worldwide name-to-target ID mappings are not identical:
a.
Pick one system and designate that the mappings on that system
are valid.
5–10 Troubleshooting and Component Replacement Guidelines
