Heavy Disk I/O crashing ML330 G3 w/IDE Raid

[Denny Wyss]

I have a ML330 loaded with 4 Maxtor 160gb drives configured Raid10 running Windows 2003 Server. It generally runs OK but when do a large file copy it moves 50-100mb and then locks up. It doesn't seem to matter if the data is moving from partition to partition or from a share. I haven't found anything in the logs to indicate where the error might be. Any thoughts would be appreciated

[Iynkaran Jeyachandran]

Hi Denny,

Seems like it could be very well be your harddrives that are causing this problem. These drive compaq provided drive? or did you purchased them 3rd party?

Try updating the system bios would help if thats not up to date. Other then that pretty sure it is the harddrives that are causing the problem. Also make sure to have the SPY utility installed to monitor the health of the harddrives.

Good luck

-Iynkaran J

[Tarek Kawach · Accepted solution]

Hi Denny:

Are you using the Integrated Dual Channel Ultra ATA 100 IDE Adapter that comes with the server. And are these HP 160 ATA drives.

Possible you could have a failing hard drive, but in this case could be software o/S related setting.

As far as I know the only supported drives on that server are 40 and 80 gig IDE drives.
because of the 48 bit LBA. Don't think that HP has added support for bigger drives; is the ATA controller seeing the drives as 160 or 131 in the F8.

A Direct Memory Access (DMA) disabled setting on both primary and secondary IDE channels causes the slower performance.

enable DMA for the Primary ATAPI device

I have found HP softpaq but it is not for this server not sure if you can use it. this softpaq can enable DMA for the primary controller it will edit your registry.

sp11740.exe http://h20000.www2.hp.com/bizsupport/TechSupport/Document.jsp?objectID=PSD_WH991019_CW01

Systems ROMPaq Firmware Upgrade Diskette for ProLiant ML330 G3 (D15) Servers
version 4.07 D15-09/15/2004 (11 Oct 04)
http://h18023.www1.hp.com/support/files/server/us/download/21992.html

http://h20000.www2.hp.com/bizsupport/TechSupport/Document.jsp?objectID=PSD_EL021007_CW01

http://h20000.www2.hp.com/bizsupport/TechSupport/Document.jsp?objectID=PSD_EM040326_CW01

Note: there is software and hardware to this issue that we need to consider and supported options under that server if you are using the ATA controller that came with server you can install the spy utility and will give you indications if there is failed hard drives.

HP ProLiant CSB-6 ATA-100 IDE RAID Management Utility for Windows 2000/Windows Server 2003
version 7.30.16.1125 (18 Jun 04)
http://h18023.www1.hp.com/support/files/server/us/download/19843.html


Best regards,

Tarek.

[Denny Wyss]

I am using the onboard Integrated Dual Channel Ultra ATA 100 IDE Raid controller (CSB-6) that comes with the server. These are brand new Maxtor drives purchased third party. Utilizing https://servername:2381/ I was able to determine that only 131071 MB of each drive were being recognized but no drive failures. DMA is enabled. It seems that short of an update from HP to support larger drives this is my problem.

I will start searching now but is there a workaround to use these drives with reduced capacity?

[Tarek Kawach]

Until HP updatest the bios for the server to allow for larger LBA support; then you are stuck at 131 gig showing for any ata drive you connect to that controller.

the F8 bios for the ATA controller when server boots will be able to see the drives and that they are online or failed. etc.

also the spy utility from HP ; but not sure if it will work for 3rd party drives. But check from maxtor you should be able to download utility that will test the drive integrity.

Best regards,

Tarek.

[Denny Wyss]

I would be OK with 131gb per drive if the speed were decent and the machine stable. I think that perhaps the raid1+0 may be more than the controller can handle efficiently on larger file moves. I think I may try rebuilding the machine with 4 single disk arrays and use software raid in the OS. I will check the drives to be sure they are good but I don't think this is the issue.