Browse Prior Art Database

Automatic Swap of the Primary and Secondary Mirrors in a Disk Mirror System

IP.com Disclosure Number: IPCOM000106649D
Original Publication Date: 1993-Dec-01
Included in the Prior Art Database: 2005-Mar-21
Document File: 2 page(s) / 85K

Publishing Venue

IBM

Related People

Craft, JL: AUTHOR [+2]

Abstract

Disclosed is a method of having the Logical Volume Manager in AIX* swap the definitions of the primary and secondary partitions in a mirrored filesystem. The criteria for performing this swap will depend on user configurable parameters.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 52% of the total text.

Automatic Swap of the Primary and Secondary Mirrors in a Disk Mirror System

      Disclosed is a method of having the Logical Volume Manager in
AIX* swap the definitions of the primary and secondary partitions in
a mirrored filesystem.  The criteria for performing this swap will
depend on user configurable parameters.

      Disk mirroring is a method of managing a filesystem where there
exists an identical copy of one physical partition on another
physical partition.  During writes to the filesystem, every write to
partition A has a duplicate write to partition B.  There are two
types of mirroring, parallel mirroring and sequential mirroring.
Parallel mirroring writes information to both partition A and
partition B, but reads data from only one of the two disks.  The disk
it reads from depends on which disk head is considered physically
closest to the address location of the requested data block.  In
sequential mirroring, there exists partitions (or disks) called the
Primary and the Secondary.  For the following references, the use of
the descriptive word Secondary represents one or more disks.  With
reads and writes in the sequentially mirrored system, the read or
write is first issued to the Primary.  Write commands are then
mirrored onto the Secondary.  If the Primary has a problem with the
disk read request, then the request is attempted on the Secondary.

      An important situation arises if there is a problem with the
disk that makes up the Primary mirror.  The Primary mirror may have a
physical problem that requires retries by the disk device driver or
bad block relocation by the Logical Volume Manger device driver
(lvmdd).  Or, the Primary disk may be disabled.  Regardless, there is
a delay that is encountered by the lvmdd as it waits for the disk
device driver's response to its request.  After this delay, lvmdd
will attempt the same action on the Secondary if the attempt on the
Primary is considered a failure.

      This disclosure proposes having the lvmdd work in conjunction
with the LVM high level commands to switch the definition of the
Primary and Secondary disk once it detects that some consistent,
serious error causing a delay or failure in the retrieval of the data
from the Primary mirror.  What variable constitutes unacceptable
delay or error by the Primary disk can be managed by the user.  As an
example, one could list three values that could be kept by lvmdd in
its d...