Adding Segments To A Raid 4 Or 5; Adding Or Removing A Spare Disk; Do You Need A Spare Disk - Novell LINUX ENTERPRISE SERVER 10 SP3 - STORAGE ADMINISTRATION GUIDE 2-23-2010 Administration Manual

6.3.2 Adding Segments to a RAID 4 or 5

If the RAID region is clean and operating normally, the kernel driver adds the new object as a
regular spare, and it acts as a hot standby for future failures. If the RAID region is currently
degraded, the kernel driver immediately activates the new spare object and begins synchronizing the
data and parity information.

6.4 Adding or Removing a Spare Disk

The MD driver allows you to optionally designate a spare disk (device, segment, or region) for
RAID 1, 4, and 5 devices. You can assign a spare disk when you create the RAID or at any time
thereafter. The RAID can be active and in use when you add or remove the spare. The spare is
activated for the RAID only on disk failure.
Section 6.4.1, "Do You Need a Spare Disk?," on page 88
Section 6.4.2, "Adding a Spare Disk When You Create the RAID," on page 89
Section 6.4.3, "Adding a Spare Disk to an Existing RAID," on page 89
Section 6.4.4, "Removing a Spare Disk from a RAID," on page 89

6.4.1 Do You Need a Spare Disk?

The advantage of specifying a spare disk for a RAID is that the system monitors the failure and
begins recovery without human interaction. The disadvantage is that the space on the spare disk is
not available until it is activated by a failed RAID.
As noted in
failure. Any given RAID can have one spare disk designated for it, but the spare itself can serve as
the designated spare for one RAID, for multiple RAIDs, or for all arrays. The spare disk is a hot
standby until it is needed. It is not an active member of any RAIDs where it is assigned as the spare
disk until it is activated for that purpose.
If a spare disk is defined for the RAID, the RAID automatically deactivates the failed disk and
activates the spare disk on disk failure. The MD driver then begins synchronizing mirrored data for a
RAID 1 or reconstructing the missing data and parity information for RAIDs 4 and 5. The I/O
performance remains in a degraded state until the failed disk's data is fully remirrored or
reconstructed.
Creating a spare-group name allows a single hot spare to service multiple RAID arrays. The spare-
group name can be any character string, but must be uniquely named for the server. For
move spares from one array to another, the different arrays must be labelled with the same spare-
group name in the configuration file.
For example, when
if the array has a spare device. If no spare is available, mdadm looks in the array's assigned spare-
group for another array that has a full complement of working drives and a spare. It attempts to
remove the spare from the working array and add it to the degraded array. If the removal succeeds
but the adding fails, then the spare is added back to its source array.
88 SLES 10 SP3: Storage Administration Guide
"Overview of RAID Levels" on page
detects that an array is missing a component device, it first checks to see
mdadm
78, RAIDs 1, 4, and 5 can tolerate at least one disk
to
mdadm