Clustering questions

[Daveyd123]

I posted this in the Exchange forums with no luck so I thought I'd try it in here...

I currently have a single 2003 Exchange server.

I just purchased 2, Dell 1850s and also a PowerVault 220S direct attach storage. I am going to Cluster the 2 servers and have the PowerVault act as the shared storage.

I need to also migrate all Exchange data from the single server to the new cluster.

It will not be a front end/back end setup. The new cluster will be taking the place of the currect, single Exchange server.

Im looking for some info on how to cluster the 2 servers with Exchange 2003 and Server 2003 Enterprise and also migrate all Exchange data.

What are the steps involved?

From what I have read I would..

1. Install 2003 Enterprise on both servers.
2. Install MS-DTC
3. Install MS Clustering Services.
4. Install Exchange on Node1
5. Install Exchange on Node2
6. Install Exchange SP2 on Node1
7. Install Exchange SP2 on Node2


I know I'm missing a few steps. Can anoyone offer and assistance?

 

[Daveyd123]

Yea...I read that. For testing, I am trying to establish the Cluster on 1 RAID5 set. I configured 3 logical drives in the PERC BIOS. The O/S sees the drives as separate disks. I formatted each of the disks NTFS and assigned the letters Q,R,S to them respectiviely. The cluster still fails when assigning Physical Disk resources.

On each of my 1850, I have a PERC4/DC RAID controller, each with 2 channels. As it stands now, the PV220s has 3 drives and 2 EMMS. If I would to purchase 4 additional drives (2, RAID1 for quorum and 2, RAID1 for logs)would I need to purchase 2 additional RAID controllers for the 1850?

 

[pgaliardo]

Daveyd123,

Sorry I haven't been on the board for a while. In answer to your question, I have 1 raid controller in each node and 1 raid controller in the Storage Device. We use an HP MSA 500, which allows for an additional controller. Dangermouse is correct in that clustering is for failover and we will be adding the second controller to the MSA 500. I also plan to mirror any single drives in my config.

I also agree with dangermouse with the partition for log files only needing to be mirrored. I am not going to change my config at this time, but after doing some further research, mirroring seems to be the preferred RAID config for log files in Exchange and SQL.

I am not as familiar with the PV220, but in answer to your question as far as additional RAID controllers, if you can eliminate any single point of failure, do it.

 

[nhbilly]

The Quorum needs to be on its own physical harddrive so set it up as RAID 1 - mirror. If you set it up as a partition it will FAIL.

Did you verify the disk?
Cluster Installation
Updated: November 11, 2003

http://technet2.microsoft.com/Windo...8d62-4a27-a865-c6e79a7245c61033.mspx?mfr=true

I setup a sql 2005 cluster with exactly the same setup you had. Getting all the hardware requirements was a pain.
My 220S had all the harddrive - FYI the # 15 disk will be disbaled by default in cluster mode.

I mirrored the Quorun - waste of space but oh well.
The rest was RAID 5 - Data

 

[Daveyd123]

If, by verify, you mean wirting data to each disk, then yes I did. I created/deleted text files on each disk

 

[Daveyd123]

OK, I think I found the issue.

I had to enable Cluster Mode on each of the PERC4/DC controllers and change their initilazion IDs...oops!!

After I did that, I booted the O/S and formatted 1 big drive. I re-ran the Cluster Admin wizard and configured a new cluster. This time everything went smoothly!!

I booted the 2nd node with no issues. I added the 2nd node in Cluster Admin.

As it sits now, the shared drive only consists of 1 RAID5 array (140GB) which the Quorum resides. I will be ordering more drives very soon...

As I look in cluster admin, I see my cluster name...under that I see Groups...under that the only thing that is listed is Cluster Group. Should Disk Group 1 be also listed? Or does the fact I only have 1 shared disk have something to do with it not being listed?

 

[nhbilly]

only when it fail over does the other node see the "shared drive"


go figure it was the SCSI card. LOL
I had the damn same problem. Enable cluster, disable write back, and change SCSI ID. That why you lose the #15 slot in the powervault 220S