Chapter 9. Maintaining Units180 3ware Serial ATA RAID Controller User Guidedetected to have been shutdown uncleanly, the unit or sub-unit will change itsmode to either ‘Initializing’ or ‘Verifying’.3ware’s auto-initialization feature determines if this is required. During driverstartup, a flag is set that indicates that the driver loaded. Upon an orderlyshutdown, the flag is rewritten, indicating that there was a clean shutdown.During the next system power cycle the firmware queries the flag. If there wasa power failure, the flag will indicate that there was not a clean shutdown.This means that the data on the drives may not be correct.When the initialization is complete, the unit is guaranteed to be redundantagain. The initialization does not erase user data.About RebuildsFault tolerant RAID units provide data redundancy by duplicatinginformation on multiple drives.• RAID 1 and 10 units each use mirroring, where identical data is stored ontwo or more drives to protect against drive failure.• RAID 5 and 50 units achieve fault tolerance by using a simple (exclusiveOR) function to generate the parity data that is distributed on all drives.When one of the drives in a fault-tolerant unit is removed, unplugged or failson read or write requests, the unit is said to be “degraded.” Rebuilding is theprocess by which the RAID unit is made fault tolerant again.You can still read and write data from a degraded unit, but the unit will not befault tolerant until it is rebuilt using the Rebuild feature.In all cases, when a RAID unit becomes degraded, it is marked as such, andthe drive(s) that failed are marked as Not In Use in the 3BM screens andDegraded in the 3DM pages.If a hot spare is specified and a redundant unit degrades, an event notificationis generated and the hot spare automatically replaces the failed drive in theredundant unit without user intervention. The rebuild process willautomatically be launched as a background process at the next scheduledtime. If scheduling is turned off, the rebuild process will start almostimmediately (within a couple of minutes). If 3DM is running and E-mailnotification is enabled, an event notification will be sent to specified userswhen the rebuild process is complete.For 9550SX and 9590SE controllers, if Auto-Rebuild is enabled, other drivesin addition to spares may be used to rebuild degraded units. If no spare isavailable, the firmware will attempt to rebuild the unit using first anyavailable drives and then any failed drives. This allows you to insert a newdrive to use in the rebuild process.