Adding SAN storage to MSA1000 WIndows 2000 Cluster

[Pat Winton]

Hi,

I have a two-node MS W2K Adv Server cluster with DL580 G2 servers, 2x2GB FC switches, that are dual-pathed to FCA2101 HBA Cards in each server. We are running SecurePath.
Servers are providing file services.

I am wanting to add an MSA30 SAN enclosure to increase shared drive capacity. The drives are racked and powered up.

Can I add disks without downtime? In other words, can I failover all cluster resources to node2 and then shutdown node1. Plug in the scsi cable from the msa1000 to the msa30, run Array Config Utility and then format logical drives in Disk Administrator?

[Steven Clementi]

Pat:

I would definitely plan for some downtime. You should not connect SCSI cables while the system is powered.


Since the MSA1000 does a bus scan during bootup, connecting more drives later on might hang up the bus and cause problems. At the very least, I would plan for 30 minutes down time to shutdown the servers, shutdown the msa, connect the new shelf, power up the MSA then powerup NODE 2. Leave Node 1 off for now.

After your back online with node 2, here are some steps you can follow to get your new storage configured...

1. Configure new array's and create new logical drives (if that is the plan. If you want to extend your existing array(s)/Logical drive(s), then you should probably boot up node 1 now as it will take some time to complete)

2. Present the new logical drives to node 2... Using Disk Admin, write signature, format and assign new drive letter(s).

3. Once done, disable access to those drives from ACU and boot up node 1.

4. Using ACU, present the new disks to node 1 and confirm accessability. Change drives letters if need be.

5. Once the drives are confirmed as having the same letters and such, go into Cluster Admin and start adding new resources.

At this point, I usually just re-present the new disks to both nodes as there should not be any read/write access to them from this point until you complete the resource wizard. Optionally, you can wait until after creating the resources on node 1, then re-present the drives to node 2 and confirm that the cluster failes them over.



If you want to expand your existing array and logical drives, there are some additional steps needed to be taken. Let us know your plans.


Steven

[Ian Selsby]

Steven

I've read your reply with interest as I'm looking to do the same thing as Pat.

Your comments fit very well with the documentation I've read, but I do want to expand existing arrays, so I would be most grateful to hear of the other steps you refer to.

Thanks in advance,

Russell

[Steven Clementi]

Ian:

I will gladly share some more, but it would be best to start a new topic to keep your config seperate. Also, add in some more details as to your current config and what you want for the future config, hardware details, etc.



Steven

[Pat Winton]

Steven, in your reply it seems you used SSP (selective Storage Presentation) within the ACU to deny access to one cluster node while defining logical units and formatting them to Windows on one node. What we ended up doing was disabling the cluster driver and the cluster services on both nodes before configuring them on the nodes one at a time with the other server off. Talked with HP Cluster support and followed items in Q175278 MS article. We left the node down instead of booting into F8 mode. All went well.

[Steven Clementi]

Pat:

There are many ways of doing things and yours is probably the better and more "best practice" way of doing it.

What happens when your not allowed to take the cluster offline and the system needs more space immediately? I had to modify the best practice on a few occasions because of situations similar to this.

I am just glad to hear you got it all worked out.


Steven