Chapter 2 — Architecture Overview High-Availability Configurations
December 2019 User Guide—Spectra T200, T380, and T680 Libraries
71
A failover event occurs when the secondary controller detects that the
primary controller is no longer communicating on the library’s internal
communication bus. After sending a query to the primary controller to
confirm that it is no longer communicating, the secondary controller takes
over the exact configuration of the primary controller and begins receiving
and processing the media changer commands sent from the host.
BlueScale Controller Failover allows backups to continue in the event that
library communication to the primary controller is disrupted. Because both
controllers are connected to the same Fibre Channel switch and maintain a
synchronized image of the library’s inventory database, the failover occurs
without manual intervention. You can then troubleshoot the failed
controller at your convenience.
After you resolve the problem, the controller that originally caused the
failover becomes the secondary controller in the failover pair. The failover
pair reverts to the original primary/secondary relationship the next time
the LCM is restarted.
The BlueScale Controller Failover feature only affects the operation of the
library’s controllers. Controller Failover does not detect if the controller is
accessible by the host. To take advantage of the feature, the host must be
configured to retry failed commands. This ensures that commands that fail
when the library connection to the primary controller is disrupted get re-
issued after the connection to the secondary controller is established. The
retry timing must be configured to allow for the time lag inherent in the
process of disabling the host’s connection to the primary controller and
enabling the connection to the secondary controller.
Exporting Drive Redundancy
As an alternative to controller failover or for drives, which cannot be
configured with controller failover, you can select multiple drives or RIMs
as controllers for a single partition, and export the same changer interface
over them to provide redundancy, as long as your data storage software
can support this.
Note: Assigning multiple drive exporters to one partition may reduce
the total number of partitions that can be configured in the
library.
Only the secondary controller (acting as primary) is functional after the failover. A
failure on this controller disrupts communications with the hosts. Troubleshoot the
cause of the failover and reconfigure Controller Failover as soon as possible.
To Next Page
