Disk Enclosures
1748282 Members
3987 Online
108761 Solutions
New Discussion юеВ

Arraydsp -m vs TFM description

 
Doug_3
Frequent Advisor

Arraydsp -m vs TFM description

Ooops! Sorry about that previous blank post. It's that nadvertant elbow...or perhaps too much coffee!

My question is about the manual and man pages which refer to column 5 on arraydsp -m as Disk Queue, yet the 12H and 11.x generate a fifth column of SCSI Q.

Is there a difference?

Also, the values I get are between 23 and 64, far higher than the manual's suggested typical amounts of 2-4. Yet nothing else in the other values (cache util, cache queue, etc) appears out of the unusual, and arraydsp -r has nothing to suggest. I do get disk alerts on one VG/LV which happens to be the database.

Any ideas on the above two questions?

TIA!!!!

Doug
4 REPLIES 4
Maarten van Maanen
Regular Advisor

Re: Arraydsp -m vs TFM description

Hi,

I seem to get the same resulsts with the 12H (and Oracle) as you. You don't give the configuration for your 12H, but with 6 disks my SCSI Queue is regularly up around a 100, it should not be higher than around 30. This queue-load listed is the internal dual SCSI-queue and not the wide-diff. queue running to the HPUX afaik.
I have switched the Resiliency Setting to High Performance and that changed things a lot but I switched it back to normal to be on the safe side. However, even with the high queue-loads, I don't get complaints from users so I don't worry too much.
CHRIS ANORUO
Honored Contributor

Re: Arraydsp -m vs TFM description

Check these links:

http://www.hp.com/essd/model12H_writews.html

http://www.enterprisestorage.hp.com/pdfs/idc_autoraid.pdf

When We Seek To Discover The Best In Others, We Somehow Bring Out The Best In Ourselves.

Re: Arraydsp -m vs TFM description

Doug:
The SCSI Queue is referring to the queue for the internal disks...it is the same.
Also the users guide is incorrect and the values for this parameter should be:
Array not in use = 0
Desired range = 0-50
Performance problems = >50
and the description of this metric is:
"Indicates the average number of transactions waiting in the internal disk SCSI queues. Too many commands in the queues decrease performance."
Doug_3
Frequent Advisor

Re: Arraydsp -m vs TFM description

thank you.