Adding New Node Veritas Cluster Server with different hardware specification

[sriramrane]

Dear Experts,

I need your suggestion on the below:

Currently we have Two Node Veritas Cluster 6.2 running Windows 2008 R2 hosted on HPE DL380 G7 Servers. We are planning to refresh the hardware and want to move all workloads to new HPE DL380 G9/G10 Servers with Veritas Cluster 6.2 being deployed on Windows 2008 R2. It will only hardware refresh without any Application OR OS Upgrade. Currently Oracle 10gR2 is configured in Failover cluster mode. Application binaries are installed in C:\ drives on all cluster nodes.

Would like to know whether I can deploy New VCS 6.2 node on New HPE DL380 G9/G10 Server and Add to existing cluster? If possible, what is the way around.  OR this will not work? 

I tried to search articles, but no luck. Since the hardware architecture will be different, what will be the consequences when we do failover manually OR if we shutdown the Resource Group and start on Newly deployed server?

Appreciate you feedback, answers, and any ideas with new approach.

Thanks 

Rane 

[snockles]

The hardware platform isn't relevant. If you look through the Install and Upgrade guide there is a list of requirement, one of which states:

Each cluster node is in the same Windows Server domain, and uses the same operating system version

The hardware compatibility list is almost completely focussed on storage components.

Across all platforms that VCS supports, the same approach is taken. As long as the base architecture (32/64 bit) and OS are the same, then it's supported. The type of server and their configuration doesn't come in to the equation.

Think of it this way - if you can install the same application binaries on the servers, then you can install VCS and manage the environment as a single cluster.

[sriramrane]

Thanks Snokles for the details.

As I understood from your reply, there wont be any compatibility issue as far as there will be same hardware architecture whether it is x86 or x64 provided same level of OS being installed in new systems and same level of VCS components.

What will be the impact on Oracle database, if there is higher processing servers are being added to existing cluster? Will there be any issues, or we can seamlessly failover the Oracle resource group once the Oracle software installed and configured on new server. Below is the procedure I want to follow, just suggest below procedure will be fine or some additional task need to be accomplished.

- Deploy Windows 2008 R2 with required updates 

- Deploy Oracle 10G R2 on new Server (same version as, which is running on existing cluster)

- Map the Existing Disks from SAN Storage to new server node 

- Add New Node to existing cluster 

-  Shutdown resource group in existing cluster 

- Deport DiskGroup from existing Cluster 

- Import DiskGroup to new node 

- Initialize and define drive letter 

- Bring Cluster Resource Group in New Node 

[snockles]

The only impact on Oracle may be licensing costs!

For the procedure, I'd recommend some change to the steps after adding the node to the cluster.

Make sure all those one-time configuration tasks have been done on the new server. Configuration file changes etc.

Update any service groups you want to run on the new node. Certainly SystemList and AutoStartList. Check the resources for any localizations.

I'd recommend running a fire-drill against the new server at this point to verify the everything is set up.

If it's all good to go, switch the group to the new server.

[sriramrane]

Hi Snockles,

Thanks again for reply. Appreciate your given solution. 

Offsource will run Config to add node to existing cluster as well adding the node to intended resource group like System List, Autostart List etc. Will definitely reverify and compare main.cf file all all nodes once necessary configuration changes been made prior moving the resource to new node. 

If I am not wrong, Virtual Firedrill for High Availability on Windows Platform is not supported, VCS only support DR virtual firedrill for Windows platform. The Virtual firedrill on both HA and DR is supported only for Linux and Unix platforms. 

[snockles · Accepted solution]

As far as the Virtual Fire Drill is concerned, the v7.x documentation shows support for a number of the infrastructure agents. BUT I haven't tested this to confirm it the functionality is atually there. And doesn't help as you're running 6.2.

So a manual check it is then.