Discussion Locked

W2K Dynamic disk errors on SAN

0 helpful
4 replies
Closed to new repliesThis discussion is closed. It stays readable, but it takes no new replies.

Has anyone experienced DMIO errors and lost disk using the
Dynamic disks with Microsoft's Disk management Volume Manager feature
which was developed in conjunction with Veritas, it's a
hybrid of their Volume Manager product.

Our 300GB DFS boxes, about 12 of them have never had this error. The 300Gb Exchange 2000 (12 machines) and 60GB SMS with SQL (12 machines)have them frequently.

Here is the Environment:

Topology: Storage Area Network
Storage: Hitachi 9960 ( Code level 01-13-19-00/00 )
Switch: Brocade 6400 ( Firmware 2.2.2b)
Switch: Brocade 2800 ( Firmware 2.2.2b)
HBA: Troika Zentai ( Now JNI Zentai 2400 & 3000 series)
driver code level 1.0.8.15 and firmware 0x82000098
Windows 2000 SP2

Any assistance would be greatly appreciated!

4 replies

by jordana

I was just doing some reading on this last week, and remembered something that might help , (or not):

From the Windows Clustering chapter of Technet:

"Disk configuration
You cannot configure cluster storage devices as dynamic disks or spanned volumes (volume sets) if they will be used as cluster resources.
Disks to be used for cluster storage must be configured as basic disks and must be formatted using the NTFS file system. Windows 2000 does not support the use of dynamic disks or the features of dynamic disks, such as spanned volumes (volume sets), for cluster storage. You cannot change an existing cluster disk, such as the quorum disk, to a dynamic disk.
You cannot use the Encrypting File System, Remote Storage, or mountedvolumes on the cluster storage device.
"

Jordan

by storagegeek

Solution:

This problem is related to the queue depth setting on the HBA. Hitachi storage ports can have up to 255 queues. Each lun assigned to the port uses X number of queues depending on the queue depth set. Our HBA's came with a default of 32.That means that you can only have 255/32=7.9 (7) Luns on that port. Since this server is in a SAN it is not efficient to have so few luns per HDS port.

By dropping the queue depth to 2 the problem is solved and there can be 127 luns per HDS port. More than that and Windows may experience DMIO's and lost hdisk.

Finally, higher queue depths will eat up processor bandwidth with little I/O perofmance gain. A queue depth of 2 is plenty for I/O performance.

Note - Dynamic disks and Basis disks are both subject to this, however, Dynamic disks are hyper sensitive.

Thanks!

This discussion is closed to new replies.