keda-scaled-object.yaml

apiVersion: keda.sh/v1alpha1
kind: ScaledObject
metadata:
  name: keda-multi-metrics-scaler
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: go-rest-api-deployment
  pollingInterval: 20 # Must be seconds
  minReplicaCount: 2
  maxReplicaCount: 7
  cooldownPeriod:  30
  triggers:
  - type: prometheus
    metadata:
      serverAddress: http://prometheus-server.prometheus.svc.cluster.local:80
      threshold: '10' # one instance per 10 queue messages
      query: ((msg_queue_length)*(clamp(rate_429_errors < 5,1,1))) OR on() ((kube_deployment_status_replicas_ready{deployment='go-rest-api-deployment'}- 1) * 10)
      # query: (msg_queue_length)*(clamp_min(clamp_max((rate_429_errors < 5),1),1)) OR on() vector(0)
      # The above query returns the queue length if the rate of errors is less than the threshold
      # Otherwise it returns one fewer than then number of current replicas (multiplied by 10, the threshold value)
      # so that the number of replicas is reduced by one when we have too many errors