Tek-Tips is the largest IT community on the Internet today!

Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!

  • Congratulations Mike Lewis on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!

Cluster Problems

Status
Not open for further replies.

jimthecanadian

IS-IT--Management
Jul 22, 2003
52
0
0
CA
I am trying to setup a cluster between 2 x250 servers with ServeRaid 4mx adapters in each. I have followed the instructions in the manual and am at the point where the shared bus can be seen by the controller on the first server but not the second. If I run the hostile take over utility on the second then the second can see the drives but not the first.

My questions are:

1) Is this normal or should both be able to see the cluster

2) If ths is normal what needs to be done to make the cards fail over

3) If this is not normal where should I look to try to remedy the problem.

Thanks for any help

Jim
 
Only one server can own the disk at any given time. MSCS uses SCSI reserve commands to control who owns the disk. This is the purpose for the clusdisk driver.

In a typical install situation, you have two nodes A and B. Power off node B, or halt it at the boot screen if the SCSI termination is provided by the server.

On node A, Use disk administrator to write a signature to the drive and assign a drive letter. While you're at it, this is a good opportunity to set the disk alignment. Windows uses the first 63 sectors on the first track for hidden volume data. The first partition will normally begin at the 64th sector. This can cause a percentage of IOs to cross track boundaries and result in two IOs on the back end. Disk alignment offsets the start of the first partition to compensate, and aligns the partiotion with a track boundary.

Anyway, after you have a signature on the drive, and a drive letter, You need to grate a cluster disk resource for the drive and bring it online. At this point, node A has a SCSI reservation on the drive. It's time to work on node B.

On node B, load MSCS. Choose join an existing cluster to join the cluster formed by node A. Once joined, you should be able to move the group containing the disk resource between nodes without issue. Either node A or node B will own the disk at any given time, never both.

John
MOSMWNMTK
 
Thanks for pointing me in the right direction. The problem I am having now is that when I try to add the resource it does not see any drives. Should I be using a physical disk or a ServeRaid logical drive resource?
 
On node A, I assume the SCSI ID of the controller is 7. Did you remember to change the SCSI ID of the controller on node B to something else?

You can use either a physical or ServeRAID logical. They must, however, be basic disks. MSCS will does not support dynamic disks, and they will not show as available when you go to create the disk resource.

John
MOSMWNMTK
 
I changed the ID of controller 1 to 6 and I left controller 2 as 7. I have also made sure there is no disk using ID 6.


Windows 2000 will not allow me to make either disk dynamic so that is not an option.
 
hi Jim,
I have written a FAQ ( faq491-3942 ) to try to
merge Microsoft directives and ServeRAID devices:

never IBM is Standard !

victor (an IBM-Business-Partner)

bye
 
Guys,

Thanks for all your help everything is up and running now. Victor...Excellent instructions. They were right on target.

Jim
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor

Back
Top