StorageWorks and Ultruim 448 Problems

[Paul Goldsmith]

The problem I am having is occurring across four HP DL380 G4 Servers, using two different operating systems and three separate tape drives.

The Servers in question are all under warranty with the following service tags

USE610N861 – Netware 6.5 SP5
USE609N406 – Netware 6.5 SP5
USE610N866 – Netware 6.5 SP5
GB8550SSTR – Windows 2003 Server SP1

The attached tape units are HP StorageWorks containing Ultrium448 Tape Drives.
Service tags. USE544003T USE544003P USE5440044

I am failing to successfully obtain a full backup from any of these devices.

The Netware Systems using Backup Exec 9.2 fail with the following information –

Vendor: HP Product : Ultrium448
Drv ID: Firmware: S53D
Function: Write(5)
Error: A Hardware Failure has Occurred on Drive HA:1 ID:3 LUN:0 HP Ultrium448 Drv

The Windows Server using NTBackup records the following –

Error: The Device reports an error on a request to write data to media Error Reported: Unknown Error

In both operating systems the backup start successfully and writes between 3 – 17GB of data and then errors out.

I have done the following.

Installed the HP Tape drive utilities on the Windows Server and tested all tape drives with successful results. The Netware Tape utilities cause the server to abend under 6.5 SP5.

Updated the HP Single/Dual Channel U320 Controller (374654-B21) Firmware to 5.05.21.00
Updated each tape drives firmware to S53D – 31/05/06
Updated the Netware HAM driver to LSIMPTNW Version 4.01.07 – 20 Sep 2006-12-15
Updated Backup Exec to 9.2 rev 1401
Downloaded the latest Tape driver for Windows 2003 from the HP website

We are using new HP Ultrium 200/400 Tapes

[Jaclyn Rothe]

Novell Netware version 6.5 SP5 is not supported because of buffering and cahce clearing isseues that affect tape drives. I would suggest downgrading to a OS version that does not have issues.

Regarding the backup failures. These are usually caused by drivers.
Depending ont he backup software being used to backup, sometimes the software manufacturer provides drivers.

If you need a WIN NT 4.0 driver... the HP version can be found here:
http://h20000.www2.hp.com/bizsupport/TechSupport/SoftwareDescription.jsp?lang=en&cc=ca&prodTypeId=12169&prodSeriesId=238671&prodNameId=238685&swEnvOID=24&swLang=8&mode=2&taskId=135&swItem=co-9175-2

Also check your SCSi card's compatibility. The wrong/corrupted drivers or incorrect card for the OS version can cause this type of communication failure.

Drivers are the most common problem for most communication problems.

[Jaclyn Rothe]

I am sorry I misread your post... I thought you were using NT 4.0 for the OS. My mistake.

If you are using Win 2003 there are problems with that, and you will likely not get reliable successful backups until you read the following microsoft doc and follow the steps to correct the problem withint the OS regarding RSM.

http://support.microsoft.com/kb/842411

Also, drivers should still all be updated :D

[Marek Nelec]

Paul

I had same problems once using Ultrium 448 and Netware (although it was 5.1 SP8) with LSI Logic SCSI HBA on ML350 G4p. The problem was Tagged Command Queueing capability enabled by default by the LSIMPTNW.HAM driver. Try to add option qtags=off to LSIMPTNW LOAD line in STARTUP.NCF
Worked for me.

[Son of Fabio]

On the Netware servers...I had applied all the fixes you mentioned to no avail. I just fixed the problem yesterday by unloading CPQSCSA.NLM from the server. Jobs ran after that. Also pound that .nlm's load statement out of the CPQSNMP.NCF file so it won't reload on reboot.

[Paul Goldsmith]

Thanks to all that have replied to this thread.

It turns out that the 1U storageworks racks has a known issue with these Ultrium448 drives. The PSUs are cutting the power to the drives for a split second which causes the drive to reset. This can happen anytime during a backup operation. The fault is better seen during the POST of the drive with the power cycling many times before the drive comes to the ready state.

Having only used these drives its hard to spot this POST issue as a problem when you don't know what normal should be. Found the soultion on another forum by chance. Even then I had 10 rounds with HP before they would send the replacement part which turns out to be the entire 1U storageworks enclosure. If I had not had three all doing the same thing I think it would have taken longer to convince them.

Once again thanks to all!!!