Smart Array P411 storage controller reports a "lockup" condition

[gtsg1]

Summary:

Smart Array P411 storage controller reports a "lockup" condition and the drives attached to the controller become inaccessible during Microsoft SQL backups.

Environment:

Prolaint DL380 G7

Windows 2008 R2 Enterprise

MS SQL Server 2005 SP4 (64bit)

Dual Hex Core Intel Xeon X5660 (multi threading enabled)

48 GB Memory

Smart Array  P410i, 1GB FBWC, firmware v5.14, connected to 12 internal SAS drives via SAS expander card (includes OS Partition)

Smart Array P411, 1GB FBWC, firmware v5.14, connected to Storage Works D2700 (firmware v 0131) containing 24 SAS drives configured in a single 1.6 TB, Raid 1+0, logical volume.

Detailed Description:

During Microsoft SQL backups that write to the array on the P411, the controller reports a "lockup" condition and the drives, enclosure and volume attached to the controller become inaccessible.  On a reboot the following lock up code is reported during POST: 0x13.

This condition has been observed on 4 separate systems with identical configuration.

The condition has been reproduced when the Array Accelerator (1 GB FBWC) is configured to the default read and write settings of 25% read/75% write and when the settings are 0% read and 100% write.   When the Array Accelerator is disabled or set to 100% read the problem subsides.

Accelerator W/R Setting                  Controller Condition

-------------------------------------------------------------------------------------------------------------------------------

25% read, 75% write                       lockup

0% read, 100% write                       lockup (lockup condition results more quickly in this configuration)

100% read,0% write                        no lockup

Array Accelerator disabled             no lockup

The issue seems to be a problem specifically with the way the HPCISSS.SYS driver, Smart Array v5.14 firmware or possibly the way the D2700 handles write functions in this hardware configuration.

I'm presently working with HP to resolve.

Has anyone else seen this issue?

[Michael A. McKenney]

Looks like a write cache issue.   24 SAS drives on an expander.   I have seen a LSI Logic issue with expander causing lockups.    The expander was getting confused during the write process according the engineers.  I would suggest doing a SAN with 24 drives over using an expander.   You might also want to consider getting a second RAID controlller and spliting the drives into two 12 drive arrays.