Say I have a service/deployment/pod with 5 replicas.
My goal is to have a dedicated pod for a specific customer and in case the customer pod is down, route traffic to the general pool. Regular traffic should not hit the dedicated customer pod.
Everything of that is possible and works, except defining the other pods as fallback, in case the customer specific one goes down.
Is there currently any way to accomplish that?
2 deployments of the same service (the customer one has an additional
customer: bob label)
1 DestinationRule with 2 subsets that target by labels (
version: stable &
version: stable, customer: bob)
1 VirtualService (match uri prefix => customer subset as destination)
Would be grateful for any ideas of how to accomplish that