Dell MD3000 Media Errors and Unreadable Sectors, Global Hot Spares, Hot Spare Operation, Rebuild

Global Hot Spares

The MD3000 supports global hot spares. A global hot spare can replace a failed physical disk in any virtual disk with a redundant RAID level as long as the capacity of the hot spare is equal to or larger than the size of the configured capacity on the physical disk it replaces, including its metadata.

Hot Spare Operation

When a physical disk fails, the virtual disk automatically rebuilds using an available hot spare. When a replacement physical disk is installed, data from the hot spare is copied back to the replacement physical disk. This function is called copy back. By default, the RAID controller module automatically configures the number and type of hot spares based on the number and capacity of physical disks in your system.

A hot spare may have the following states:

•A standby hot spare is a physical disk that has been assigned as a hot spare and is available to take over for any failed physical disk.

•An in-use hot spare is a physical disk that has been assigned as a hot spare and is currently replacing a failed physical disk.

Rebuild

If a disk fails in a fault-tolerant disk group (RAID 1, RAID 5, and RAID 10) and a hot spare is available, the RAID software automatically attempts to rebuild the data to restore redundancy. If no hot spares are available, an automatic rebuild occurs when a new physical disk is installed. You can use MD Storage Manager to specify a physical disk to rebuild.

The requirements for a replacement physical disk are the same as those for a hot spare: the capacity should be equal to or larger than the size of the configured capacity on the physical disk it replaces, including its metadata.

NOTE: For a stripe set of mirrors (RAID 10), it is possible for multiple disks to fail without a virtual disk failure.

Media Errors and Unreadable Sectors

If the RAID controller detects a media error while accessing data from a physical disk that is a member of a disk group with a redundant RAID level (RAID 1, RAID 5 or RAID 10), the controller will try to recover the data from peer disks in the disk group and will use recovered data to correct the error. If the controller encounters an error while accessing a peer disk, it is unable to recover the data and affected sectors are added to the unreadable sector log maintained by the controller.

Other conditions under which sectors are added to the unreadable sector log include:

•A media error is encountered when trying to access a physical disk that is a member of a nonredundant disk group (RAID 0 or degraded RAID 1, RAID 5 or RAID 10).

•An error is encountered on source disks during rebuild.

NOTE: Data on an unreadable sector is no longer accessible.

38

Using Your RAID Enclosure