SAS 9.4 M8 Installation

[mkiran]

I am performing new installation for M8 and completed below.

1. Metadata / Grid control - it has SASApp server and Object spawner to use as a compute for standalone tools. It is the LSF master too.
2. Grid Node 1 - Installed LSF and other SAS services like WIPDS, Stored process, Another application server for Grid ,Object spawner (second one - should I reuse the same one from Grid control server ? )

Both nodes are installed on a shared configuration directory.
And I see all services runnig properly on Grid node1 using sas.services script.

Do I need to perform any other configuration using SDW on Grid node 2 ?
How can I reuse the same sas.servers script on Grid 2 to start only the Object spawner with out starting other services like DIP, Locator, WIPDS, etc?.

[gwootton]

1. In this configuration you would not use the same object spawner between grid control and grid node 1.
2. No, you do not need to perform additional configuration on Grid node 2.
3. You would not use the sas.servers script to start the services on Grid node 2 for the reason you have identified, it would try to start services already running on node 1, you would just use the ObjectSpawner.sh script to start the Object Spawner on Node 2.

[mkiran]

@gwootton : Yes I am not able to use the Grid Node1 Object spawner and grid node1 Application server loadbalancing with Grid algorithm(Only COST algorithm is available here).

Do you recommend configuring the Grid Node1 and Node2 on the same Grid Control server configuration (shared)?. In this architecture, I will need to select this shared config directory from SDW and merge these additional Application server node products. 

then I need to control sas.servers to control services on each respective node. Please suggest the optimal method for this configuration, we are flexible to perform the configs.

[gwootton · Accepted solution]

Usually when I perform deployments I only perform the compute tier install and configuration once for the grid controller, and share that sas install and sas config to the grid nodes. When using a shared SAS Config you would not use sas.servers to start processes on any host but the initially configured node (e.g. grid controller). Then you'd use the individual service commands to start services on the node.
Your configuration of having one compute tier on the grid controller and another for grid nodes is also valid though.

[mkiran]

We have one compute on grid controller (metadata also on the same) and I am getting Objectspawner and SASApp context with grid algorighm enabled.

And when tried to install Other grid node(using the provided plan file) , it is prompting to create a new Object sapwner and server context - may be I should reuse the existing ones from Grid Controller server instead of creating news ones on Grid1 and add node1 and 2 in the Objectspawner and SASApp load balancer.

FYI : Below is the planfile snippet.

[gwootton]

Yes, that's what I do. I only install and configure the compute tier once, I do not use the plan file to configure grid nodes as well.