Multiple Component Failures - IBM DS4800 User, Installation And Maintenance Manual

Multiple component failures

When multiple components in the DS4800 have failed, the behavior of the Service
Action Allowed and Needs Attention LEDs can vary from situation to situation. In
some cases, multiple Service Action Allowed status LEDs light when their
corresponding Needs Attention LEDs are lit. In other cases, none of the Service
Action Allowed status LEDs light when their corresponding Needs Attention LEDs
are lit. This variance exists because the ability to remove a component depends on
the different data availability dependencies of the storage subsystem components.
If replacing any one of the multiple components that have their Needs Attention
LEDs lit will cause the DS4800 Storage Subsystem to shut down, the Service
Action Allowed status LEDs will not light on any of the components that need
attention. For example, if both the right power supply-fan (which is linked to
Controller A) and Controller B need attention, then neither of the Service Action
Allowed status LEDs on these components would light. Removing the Controller B
before replacing the failed right power supply-fan would cause Controller A to lose
power, resulting in a loss of data availability. (This linkage exists because the power
distribution from each power supply-fan runs through the controller physically
connected to that power supply-fan.)
However, if replacing any one of the multiple components that have their Needs
Attention LEDs lit will not cause the DS4800 Storage Subsystem to shut down, then
the Service Action Allowed status LEDs will light on all of the components that need
attention. For example, if the right power supply-fan (which is linked to Controller A)
and Controller A both need attention, then the Service Action Allowed status LEDs
on both of these components would light, indicating that either component can be
replaced first.
Important: In this situation (in which replacing any one of the multiple components
automatically turn on. Before the interconnect-battery unit Service Action Allowed
status LED can turn on, you must use the DS4000 Storage Manager client to
place Controller B offline. (Controller A is the controller designated to receive all
I/O activity when the interconnect-battery unit is removed from the DS4800.) By
placing Controller B offline, you route all controller and I/O activity through
Controller A. This maintains data availability while the interconnect-battery unit is
removed, preventing any problems that would normally arise from the loss of
communication between Controllers A and B.
Attention: Before you remove the interconnect-battery unit, you must verify
that Controller A is in an optimal state, and the power supply-fan CRU that is
connected to controller A (the right power supply) is optimal. Never remove the
interconnect-battery unit when Controller A or the power supply-fan CRUs are not
in an optimal state.
that have their Needs Attention LEDs lit will not cause the DS4800
Storage Subsystem to shut down), after you remove one failed
component, the Service Action Allowed status LEDs on the remaining
components that need attention might turn off. This change in the
Service Action Allowed status LEDs indicates that removing additional
components now might cause the storage subsystem to shut down.
For example, if both power supply-fans have failed, the Needs Attention
and the Service Action Allowed status LEDs would be lit for both
components. However, when you remove one power supply-fan from
the system, the Service Action Allowed status LED on the remaining
power supply-fan turns off, indicating that you now cannot remove the
Chapter 5. Replacing components
135