HP NetRAID-4M Configuration and Upgrade Guide (Release 5)

Chapter 9 Issues and Problem Resolution
56
Problem: Microsoft WHQL (Whistler HCT) is now testing the driver’s ability to
address the entire 32 bits of memory. It has been determined that our communication
region can only exist in the lower 2 GB of memory.
Impact to Customer: Minimal, there is an error message in the event log:
BAD_FW_WARNING
Frequency of Occurrence: Often. Every time the system is rebooted, the entry is
placed into the System Event Log.
Workaround: This issue has been addressed in the latest driver version 4599 and is
available as part of the Release 3 set.
43 Automatic Failover rebuilds onto failed drive
Problem: Automatic Failover allows a user to replace a drive to trigger automatic
rebuild process without any additional steps to bring the container into a redundant
state. This failure uncovered a problem in the design allowing a rebuild process to
start onto the same drive that caused the failure. Here is a sample error log:
15:31:55 CST - Device\HPN0, RAID5 Container 0 Drive 0:9:0
Failure
15:31:55 CST - Device\HPN0, Container 0 started REBUILD task
on drive 0:9:0
16:53:58 CST - Device\HPN0, Container 0 completed REBUILD
task:
Impact to Customer: Serious. The data integrity is protected, but you must replace
the marginal drive. Since the controller is trying to rebuild onto a failing drive, these
attempts may slow or halt the system, making it unusable.
Frequency of Occurrence: Occasionally. Any disk can fail.
Resolution: To prevent from rebuilding onto a bad drive our design creates a drive
bit map that keeps track of the BBR (Bad Block Replacement) failures and prevents
the NetRAID-4M from rebuilding onto the same drive. If there is a new drive
inserted it will trigger the Automatic Failover process, providing there is no spare
assigned. If there is no new drive inserted the rebuild process will not happen. The
problem has been addressed in Release 3. Another part of the fix limits the BBR
attempts to make sure that the NetRAID-4M will not dominate the SCSI bus to the
point that the other devices cannot get a share of it.
44 SNMP agent reports yellow warning if no controller installed