Deploy a new APIC Cluster HA Design + Multi-Pod on a Brownfield single POD without Ha on APIC Cluster

[WiliamRodrigues7077]

Hi Folks,

I am about to deploy a second POD on a production single POD ACI Infrastructure.

The current scenario is:

3 APIC Nodes (APIC1, APIC2 and APIC3) and a single pod

I have to move the current scenario to:

4 APIC Nodes (APIC1, APIC2, APIC3 and APIC4) and a multi-pod (2 Pods), being POD-1 (DC1 on City1) and POD-2 (DC2 on City2). The two cities are around 500km far from each other and due to logistics restrictions it is not possible to move one APIC node from DC1 to DC2 within few days.

From the best practices with 4 APIC nodes it is recommended to have 2 APIC nodes on POD-1 and 2 APIC nodes on POD-2

So I understand that the final scenario should be: Active APIC Nodes (APIC1 and APIC2 on POD-1). On POD-2 APIC3 as active and APIC4 as stand-by.

Do you have any recommendations on the following steps to reach the final scenario:

Step1 - Add APIC4 as stand-by node on POD-2

Step2 - Shutdown APIC3 on POD-1 and promote APIC4 from stand-by to active on APIC3 behalf

Step3 - Wipe-out all APIC3 node from its console and send it to DC-2 city

Step4 - Perform initial setup of APIC3 node as stand-by and connect it on POD-2, joining the APIC cluster

Step5 - Shutdown APIC4 on POD-4 and promote APIC3 from stand-by to active on APIC4 behalf

Step6 - Wipe-out all APIC4 node and perform initial setup as stand-by on POD-2, joining the APIC cluster

After theses steps I would have:

POD-1 - APIC1, APIC2 as active nodes

POD-2 - APIC3 as active node and APIC4 as stand-by node

Thanks in advance.

Looking forward your comments.

[RedNectar]

Hi @WiliamRodrigues7077 ,

I think you have small misconception when you say 

I understand that the final scenario should be: Active APIC Nodes (APIC1 and APIC2 on POD-1) and APIC3 and APIC4 as stand-by on POD-2

Now it could be a gramatical/parentheses misplacement error. The final scenario should actually be:

Active APIC Nodes (APIC 1 and APIC 2 on POD-1 and APIC3 on POD-2) and APIC4 as stand-by on POD-2 - which is pretty much how you explain it nearer the end of your question.

The key point being that you need three active nodes, and although the 4th APIC is classed as "stand-by", it is really a "potential replacement" for a failed node.

Now my first advice is that you raise a TAC case and make use of their expertise.

But in regard to your question, I think you logic is sound as far as Step4.  At which point I think you should stop. (I'm happy to be corrected on this point) where your new APIC has taken the role of APIC3, and the removed and cleaned APIC3 becomes the standby APIC4 at DC-2.

[WiliamRodrigues7077]

Hello Chris,

Thanks for the reply and you´re totally right about the final scenario. Definitely I haven´t used the right words (and already updated it). Thanks.

About the steps, steps 5 and 6 are just to leave the final scenario with APIC 1, 2 and 3 as active and 4 as stand-by, which is just a cosmectic change to leave APIC3 as active as it were in the beginning of the process.