Monday 16 July 2012

Intel Integrated RAID Modules - POST Hang and/or Data Loss Issues

A few Intel Technical Advisory documents to take note of with the first being the most critical.

Description

The PCB’s (Print Circuit Boards) for the controllers listed above are being modified to resolve an AC power cycling cache backup issue with the Intel RAID Smart Battery AXXRS8BU9 and the Intel RAID Maintenance Free Backup Unit AXXRMFBU2 cache backup solutions. In addition to some minor component changes a jumper wire will be added to the PCB’s during manufacturing. Intel has completed product qualification for the changes to verity that there is no impact to product quality and reliability.

Root Cause

During AC power cycle testing Intel found a rare occurrence where data yet to be written to disk from the RAID modules DDR3 memory may be lost even though cache protection is enabled by the Battery Backup Unit (Intel® RAID Smart Battery AXXRSBBU9) or the RAID Maintenance Free Backup Unit (Intel® RAID Maintenance Free Backup Unit A)(XRMFBU2) that is installed. The issue has a very low occurrence rate and is observed on a small percentage of modules. The component changes and jumper wire solution provides a resolution to this issue.

The units we just brought in for a couple of 2 node clusters we are building out do indeed have the jumper wire in place.

With the factory default firmware in place the server will hang during POST. We update all firmware on the server builds and deploys that we do prior to introducing them into a production environment. Once updated we run a series of burn-in tests over a period of days to validate the build.

Firmware, drivers, and more can be had from Intel's Download Centre.

Philip Elder
MPECS Inc.
Microsoft Small Business Specialists
Co-Author: SBS 2008 Blueprint Book

*Our original iMac was stolen (previous blog post). We now have a new MacBook Pro courtesy of Vlad Mazek, owner of OWN.

Windows Live Writer

1 comment:

Philip Elder Cluster MVP said...

Rob Pelletier said...

Maybe not relevant to the topic, but I have a 25PB080 in a server with a S2600CP motherboard. RAID card has 4 drives - 2 RAID1 SAS drives and 2 RAID1 SATA drives.
Put a 5th drive in the system to receive backups, but it's attached to the embedded controller on the motherboard. Large data copies from that 5th drive seem to be bogging down the whole works - users complain about "everything being really slow" when I copy that drive to a portable for off-site storage.
Think attaching that 5th drive to the RAID card would make that much difference in performance?

Philip Responds:

Yes.

Use a USB3 PCIe card and tag those backup drives to a SuperSpeed drive dock or external enclosure.

Philip