[RabbitMQ] Reduce number of connections from keda HPA

[MatheusGeiger]

Hello everyone, a time ago we change the default HPA for the cluster to KEDA hpa for our consumers who that use the RabbitMQ and it working perfectly.

Now looking the rabbitmq painel with most preciousness we have double connection "per consumer" one the consumer and other the hpa for this consumer.

This behavior and implementation makes the number of connections duplicates, and looking this the resource usage from the cluster increase too.

Is possible or exists one approach to reduce the number of connections?

HPA Example Receipt

apiVersion: keda.sh/v1alpha1
kind: ScaledObject
metadata:
  labels:
    deploymentName: monopoly-consumer-customer-configuration-deployment
  name: monopoly-consumer-customer-configuration-hpa
  namespace: production
spec:
  maxReplicaCount: 10
  minReplicaCount: 0
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: monopoly-consumer-customer-configuration-deployment
  triggers:
    - authenticationRef:
        name: keda-frizzy-cobalt-angelfish-trigger-auth-rabbitmq-conn
      metadata:
        mode: QueueLength
        protocol: amqp
        queueName: monopoly.customer-configuration.v1
        value: '100'
      type: rabbitmq

Number of connections from keda user

[zroubalik]

I mean one connection that will both serve the workload (client/consumer) and KEDA that monitors this workload?

This is not possible, KEDA has no means to reuse connection from your apps. It needs to get the data somehow