Hardware Installation
PixelStor 5000 User Manual
Rev. 1, May 2013
5-83
5.2.3
RAID Controller Failure
When a RAID controller fails in the
PixelStor 5000
with redundant controllers, all drives
including those on interconnected PixelStor 100e enclosures will fail-over to the surviving
controller without interruption to data access. If the PixelStor 5000 is in a single controller
configuration the system will shut down and all data will not be available
Corrective action:
Dual Controllers - replace the failed controller and the system will automatically fail-back
to original state.
Single controller - replace the failed controller, reconfigure the host access and
rediscover the drives groups.
5.2.4
SAS IO Expander Failure (Lost Connection to Expansion Enclosure)
PixelStor 5000 capacity can be expanded by adding up to 5 additional capacity expansion
enclosures for a maximum capacity of 72 SATAII drives. These enclosures are daisy chained to
the PixelStor 5000 SAS IO expanders and 12Gbps SAS connections.
In the event a SAS IO expander fails, all expansion enclosures and corresponding drives
connected to the failed IO expander will become unavailable. Under this scenario, the system
will treat all drives associated with that SAS expander as failed. Since disk groups can span
across the PixelStor 5000 and interconnected expansion enclosures, a loss of one or more
expansion enclosures can affect multiple disk groups. If one drive for RAID 1/5/10 or two drives
for RAID 6 are missing from the associated Disk Groups, the system will rebuild them using any
available global spare. If multiple drives (2+ for RAID 1/5/10 or 3+ drives for RAID 6) are
missing then all writes to the DG will be suspended and the DG will be made available to the
system in a read-only state.
Corrective action:
Dual Controller - the system will fail over and the controller with healthy IO expanders
and cable links will take over all the disks without disruption to data access. The
administrator can replace the failed expander and the system will automatically fail-back
to regain redundancy.
Single Controller - all connected expansion enclosures (and drives) will become
unavailable. The administrator will need to replace the IO expander and do a Disk Group
Restore to reactivate all the disk groups brought off-line by the failure.
