Technical Advisory 1029

This Technical Advisory describes an issue which may or may not affect the customer’s product
Intel Technical Advisory TA1029-01
5200 NE Elam Young Parkway
Hillsboro, OR 97124
June 6, 2013
Copyright © 2013 Intel Corporation. * Other names and brands may be claimed as the property of others.
Intel
®
RAID Controller Maintenance Free Backup Unit (RMFBU) may fail
to complete a learn cycle.
Products Affected
AXXRMFBU2 Used with RAID Modules RMS25CB080, RMS25CB040, RMS25PB080,
RMS25PB040, RMT3CB080, RMT3PB080
RS25AB080 Ships with an RMFBU in the product package
RS25SB008 Ships with an RMFBU in the product package
No other Intel
®
Server Systems or accessories are impacted by this issue.
Description
The RAID controller or RAID Module may not successfully complete a learn cycle and the RAID controller/module
firmware log will report that the RMFBU remains in a 0x804 (charging) status. The RAID volume will remain in write
back mode and the RMFBU should have enough energy to complete a cache offload. However, there is a possibility of
data loss during a power loss event if the charge level drops below the level required for the RMFBU to supply sufficient
energy to copy cached data into non-volatile memory on the RMFBU (a Cache Offload).
A learn cycle charges the RMFBU super capacitor pack above the typical voltage level and measures decay over a
short period of time in order to judge the health of the RMFBU. A learn cycle takes about 2 min to complete and cached
data is protected during that time because super capacitor energy never drops below the minimum threshold needed to
complete a cached data offload. Following a learn cycle, energy is maintained in the super capacitor via a small trickle
charge. Learn cycles are scheduled by default to occur once every 28 days, although the schedule can be adjusted by
the user.
Root Cause
The full charge indicator setting thresholds in the RMFBU firmware are set too tight for the normal variation of
components used in the charging circuit on the RMFBU, preventing the RAID controller firmware from fully assessing
the health of some RMFBU super capacitor packs. Environmental characteristics and circuit tolerance variability may
also impact whether the thresholds are met. The “failure to initiate a learn cycle” error is only visible in RAID controller
firmware logs.
Corrective Action / Resolution