System information

Advanced Functions
5-12
• Owner controller and preferred controller
The controller implements the ALUA (Asymmetric Logical Unit Access) algorithm to ensure that only one controller is
allowed to access a virtual disk. The controller that controls a virtual disk is the owner controller of the virtual disk. When
both controllers are healthy and all paths are online, the owner controller is the same as the preferred controller. There
are two possible cases that the preferred controller loses the ownership of a virtual disk to the peer controller:
1. If a controller fails, the survival controller will take over the ownership of all virtual disks of the faulty controller.
2. When all the paths to a controller are disconnected, the MPIO driver will force the virtual disks to transfer to the other
controller.
The preferred controller is specified when a virtual disk is created, and it can be changed later by users (see the
Preferred Controller option in Section 2.6.2, 2.6. 4, and 2.6.5 for modifying JBOD disk, logical disk, and volume,
respectively). But the owner controller is changed dynamically according to the current path and controller status.
• Single-controller mode in redundant-controller system
If a redundant-controller RAID system is powered up with only one controller installed, the system will be operating in the
single-controller mode, and there is no event or beeper alert because of the absence of the peer controller. But if the
other controller is installed, the system will be operating in redundant-controller mode.
While enabling the mirror write cache protects the data integrity, disabling this option may cause data loss when the
controller is failover but allows better write performance. For the purpose of data protection, it is suggested to turn this
option on.
5.3.2 Controller Data Synchronization
When running in a redundant environment, two controllers will automatically synchronize the configuration data, event
logs, task progress data, system time, and firmware. See further details as described below.
• Configuration data
The controller’s configurations and states are synchronized between two controllers. There are two copies of one
identical configuration data in the two controller. Updating the two copies of configuration data should be considered as
an atomic operation.
• Event logs
The event logs are mirrored between controllers. Users can view the event logs even one of controllers is failed. The
SMTP and SNMP configuration will be also the same across the master and slave controllers. Should the master
controller is failed to send the event notification, the system will try to send it through the slave controller, and vice versa.
• Task progress data
The task progress data of a controller’s background task will be synchronized to the peer controller. If one controller fails,
the other can resume the background task of the failed controller.
• Time
Two controllers will keep syncing the Real Time Clock (RTC) in a fixed period of time.
• Firmware
The redundant controller system must have the two controllers to be operated in the same firmware version. There are
four scenarios for firmware version update.
• Boot State Update
Upon system boot, if the firmware version and boot code of the two controllers are unmatched, the system will have
prompt text shown on the LCD. The user needs to select the target version or boot code through the LCD menu
interface.
Note
To get best performance, make all LDs in a DG have the same preferred controller, and evenly
distribute DGs between the two controllers.
Note
Once the a virtual disk has been exported to host as a LUN on a controller, the owner controller can
be changed only after restart the controller, and the Change owner controller immediately option is
not displayed.