SQL/MX 2.x Installation and Management Guide (H06.10+, J06.03+)
Performing Recovery Operations
HP NonStop SQL/MX Installation and Management Guide—544536-007
12-40
START DISK and STOP DISK
After the disk drive is stopped, you can bring the disk up as a phantom drive (without a
name), label the disk with another volume name, or reuse the disk in any other way.
This operation is often done on nodes where nonmirrored disk space is needed for a
short time.
Later, you can return the previously stopped disk drive to its original mirrored state by
performing a START DISK.
You should not use the stopped drive to store production SQL/MX database files. You
typically use the drive for a test database or for temporary space for sort files. You
must ensure that you can make the volume inactive and revive the drive back to its
original mirrored configuration with no effect on the original database.
Use the STOP DISK command on a volume to put a particular volume out of operation.
You can later use the START DISK command to put the same disk back into operation.
Inconsistency is unlikely if the disk brought up is identical to the disk brought down.
Always perform a STOPOPENS DISK on the volume and a CONTROL DISK,
REFRESH on the volume to ensure valid file labels before you make the volume
inactive.
The only exception to the preceding rule is if the entire database has been consistently
brought down as a unit. For example, suppose that you use STOP DISK to bring down
all the backup volumes of the mirrored pairs containing SQL/MX objects in a consistent
state. The other mirrored set continues the active database, but the inactive mirrors
also contain a set of consistent SQL/MX objects.
You can also use STOP DISK to bring down the active database and START DISK to
bring up the saved database in a database swapping technique. This technique might
be useful for testing scenarios. As long as you bring each set of mirrors down and then
up together, each copy of the database continues to be consistent.
Caution. You cannot use STOP DISK on a volume and replace the volume with an earlier
version of that same volume without causing inconsistencies in the database.










