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

Chapter 9 Issues and Problem Resolution

Problem: The Automatic Failover option gets executed regardless if the drive has

been replaced or not. When a drive reports a failure, the rebuild process should

happen onto a spare drive. If there is no spare drive assigned and the Automatic

Failover option is enabled, the rebuild should happen onto the same SCSI ID,

providing that the drive has been replaced. The UNCONFIG bit gets set when a new

drive is reinserted.

The bug is that the rebuild gets kicked off regardless if the drive was or was not

replaced.

Impact to customer: Moderate. If the failed drive is marginal, the rebuild could

attempt over and over, dropping the performance, and hiding the more serious issue

of replacing the drive.

Resolution: Fixed the rescan-clearing the drive rebuild bitmap.

2 Problem creating containers >1 TB in size

Problem: Error messages when working with the containers larger than 1 TB.

Impact to customer: Error messages only. No loss of data or functionality.

Resolution: Block numbers were previously stored as signed integer types. Block

number ranges for 1 TB containers required numbers that used the upper bit and

were used as the sign bit in signed integers. The fixes here switched signed integers

used for block numbers and associated items to unsigned integers such that the upper

bit could be used for the higher block number.

3 SCSI bus reset hanging controller

Problem: Windows 2000 Domain controllers hang in an unspecified state with no

black screen, no blue screen, and no errors in the event log. The server appears to be

in a stable condition until you try to login and there is no mouse or keyboard control.

The system appears hung. The server will respond to PING requests but will not

allow any other form of communication, the response given for a terminal server

session is “network error 53 network not responding.”

Impact to customer: Serious. Controller can hang causing data unavailability. If

the controller is hosting the operating system, the server will crash.

Resolution: Correction to the SCSI pass-thru that was exposed by Toptools.

4 Data corruption under dual synthetic test loads