MSA 1500 disks not available for Windows Cluster

[Jeff Riechers]

Sure I am missing something simple here.

Already have the MSA1500 setup with SSP for my 2 Enterprise servers for boot volumes and dedicated data arrays. Now looking to leverage cluster server with the remaining data space. Have both servers connected with Fibre, setup the disk volume as basic, formated to NTFS. Have SSP only to the first server in the cluster. Successfully create the cluster, but when I go to add the physical disk, it is not listed in the drop down box.

Am I missing something?

[James Muell]

Both HBA WWNs from ALL Cluster nodes need to be assigned (ownership) to all disk/LUNs the cluster needs to see.
If you clustered the 1st node already, doesn't that mean it recognized your external Quorum drive?

If you setup your data disks for the cluster after clustering, and have NTFS'd them and assigned a DrvLtrs, you should be able to add the disk resources any cluster group.

The GUI to do this is funny because at the point you expect to see the disks to add, you have to go to the next screen to get the drop down list of available disks. Select them one at a time and from the group window bring them online.

When adding the 2nd node (that also has SSP permissions to the cluster disks) you can bypasss part of the setup by opening Advanced and checking the "...dont look for all the disks" option. This just speeds up the 2nd nodes joining the cluster.

I am interested in your boot from SAN stuff. Do you use HP docs to set this up?

[Jeff Riechers]

And that is part of the problem. I am not sure how on the MSA to assign a LUN for clustering.

I have done the SSP and set it to use just the 1 node. Does enabling the disk for multiple nodes make it automatically a clusterable disk? Or is there something within the fibre channel itself it needs to setup.

On the clustering setup side, all I have done is run the initial configuration wizard, and have the cluster responding by names and pings. I can't actually select any of my mounted luns when I get to the screen.

[James Muell]

I am assuming you have redundant HBAs in both nodes. ALL CL Node HBAs listed in SSP need to be assigned to the LUNs that you want in the cluster.
U SAID ""Does enabling the disk for multiple nodes make it automatically a clusterable disk? ""
MY ANS : YES
With dual HBAs you would have to be running Multipathing software, and Node1_HBA1 going to sw1: and Node1_HBA2 going to sw2; and same for Node2.

As for not seeing the other disks you want to use for the cluster...Either
You can see them in Disk Management? For the cluster to see them they have discovered by the host. If so then they should be available to the 1st cl node.
Or
There is a switched fabric issue and how your FC cables are layed out. Shud B as mentioned above.

[Jeff Riechers]

Is it a requirement to have a redundant HBA, FC Switches, and MSA1000 controller in the MSA1500? Because those are not to be added to this client until next fiscal.

[James Muell]

Generally, Yes.
Your actual failover is much more limited if your not using redundant HBAs, switches, and MSA controllers, and your facing downtime for a controller failure.
A couple more questions.
Is this Windows2003?
SQL cluster?
What MSA firmware are you using?
Is your current hardware config:
1 Controller
1 switch module, or 1 HUB?
1 HBA in ea Node?
2 servers with exact same hardware, mem etc?

[Jeff Riechers]

2 identical 380 G5
1 MSA 1500 Starter kit with
1 controller shelf
1 controller
1 HBA per server
1 FC switch

Windows 2003, looking to just cluster file storage between the 2 servers.

[James Muell]

OK. I recommend the following.
Are your HBA drivers the same on both hosts?
Do you have ALL green lights on the switch ports you are using for the hosts and storage?
Am assuming you have 1 port connecting the MSA1500 to the switch.

If all your MSA disks are available (NTFS'd and DLtrs) in Disk Management on Node1 you should be able to add them to the cluster.
If not you could remove the cluster; discover all the clusterable disks your going to need in the cluster, and Create the cluster. This should automatically add the disks as cluster resources.

On Node 2 you should be able to open Disk Management and see all the MSA cluster disks. Don't do anything with them on Node2.
From Node 2 you should be able to run Cluster Admin and Join it to the cluster.

[Jeff Riechers]

Created the disk, with ssp assigned it to both servers. Created a basic volume, did not change it to dynamic. Set it to use mbr not gpt and formatted it with an NTFS permission. Not sure what you were referring to with DT. Started cluster admin, tried to add it as a resource and again, the disk drop down box was empty.

99.9 percent there, just missing 1 little thing I feel.

[CA881997]

You might have done this, but I always start with one server, make sure it sees all the drives in Disk Manager, shut it down do same for the other server. Then install cluster on the first node, 2nd server should be up so software can install there. If you already have a cluster established I think you should leave one node down , add the disk resource , then turn on the 2nd node.

[James Muell]

Honestly, I would remove the cluster and start over. Setup the drives you want to use for the cluster with drive letters, i.e, quorum (Q drive); the fileshare (D Drive). Test copy something to them. If that works your cluster should build correctly and find both drives. Then build your cluster.

Are you using Windows 2003 sp1 Enterprise Ed?

What Michael said is true to some extent but not really necessary with Win2003 ent.
There are some other issues with setting up SQP2005 on a cluster but your not doing that.

[Jeff Riechers]

Ok, here is the problem.

Since I am booting off of the same SAN that I want to make my disks from, and they are part of the same physical array, they appear to the system as being on the same bus.

Cluster server does not support the shared disk on the same bus as the boot disk.