Multiple Node-red instances in IBM cloud for redundancy not working for me

If you run multiple instances of Node-RED then you have multiple copies of the flows running at the same time.

That's fine for http triggered flows as the platform load balancer will spread the request across the instances.

The same cannot be said for the MQTT nodes. Each instance creates its own connection to the broker and subscribes to its topics. The broker will then send each message to all subscribed clients.

Some brokers do provide a means to load balance, using a particular topic/client id scheme. But it depends what you're using.