Hi,
We have a spring boot application, which runs as multiple Deployment pods in a Kubernetes container. This application use Kafka Java Streams to process its data.
I’ve observed that during a deployment of a new version, or if we do a kubectl rollout restart on the deployment, that it undergoes several rebalances as instances join and leave the group.
I understand the solution to this to not have dynamically allocated member id’s for the group members, and instead, to use static members configured with group.instance.id parameter.
But the question is, given I don’t have a per-instance static configuration (lol, not at all cloud native way of configuring this), and the application is running in a spring boot container, how would I go about configuring this parameter? I can’t just put group.instance.id: 1 in the spring application.yaml – every instance would have the same instance id. At the same time, anything available to me in the spring boot container like hostname or IP address would be unique every single time a container is rotated by the k8s controller.
Does anyone have any recommendations or suggestions?