Communicator e3000 MPE/iX Release 7.5 PowerPatch 2 (Software Release C.75.02) (30216-90352)

Chapter 3

Technical Articles

Announcing High Availability FailOver/iX for FC Disk Arrays

by Jim Hawkins, MPE/iX Lab

General Information

At no charge, High Availability FailOver/iX (HAFO or HAFO/iX) is now available for HP e3000 systems

connected to many of HP’s VA and XP family Fibre Channel (FC) disk array products. Support is offered for

customers with HP e3000 A/N-Class systems running MPE/iX 7.5 using the A6795A 2Gbit FC-HBA and also

for customers with HP e3000 S9xx systems running at least MPE/iX 7.0 using the A5814A-003 SCSI-FC

Router. This greatly expands the coverage of the HAFO product which until recently supported only the SCSI

XP256 (SCSI XP256 remains supported with these changes).

Once the HAFO software is installed, the user configures each disk LDEV with a primary data path and an

alternate data path. In a normal state, data is routed through the primary path while HAFO software

continually monitors disk I/O reply messages for failed components. HAFO is designed to detect the following

event types:

•Hung I/O

• Failed disk controller

• Failed I/O host device adapter card

• Failed connection component (cables, switches, routers)

When these events are detected the HAFO software will “failover” to the alternate data path. That is, all

subsequent I/O is sent to the configured alternate data path. No application or higher level MPE/iX Operating

System component should experience an abnormal event. All I/Os complete as normal using the alternate

data path and alternate array controller.

In the event of a failover, the system operator is notified via an MPE/iX console message. This is a “repeater”

type message. Repeater messages are repeatedly sent to the console at approximately five minute intervals

until a REPLY command is used to acknowledge the message. In addition to the console message, the

failover event also causes I/O type log records to be written into the MPE/iX system log file.

Upon repair of the failed component, the system operator may switch the LDEV back to the primary path to

restore HAFO protection for that device.

NOTE To prevent an uncontrolled series of repeated failover events HAFO does not allow automatic

“fail back” from alternate to primary path

Who Should Consider HAFO

HAFO is for customers using disk arrays who want protection against failure for the following components of

the I/O subsystem:

• I/O Host Bus Adapter (HBA)

• Cabling

•FC Switches

• SCSI-FC Routers

• Disk Controllers