Skip to main content

kubeswarm Autoscaling - KEDA-Based Agent Autoscaling on Kubernetes

kubeswarm scales agent pods on Kubernetes based on pending task queue depth using KEDA. Configure min/max replicas, target pending tasks and scale-to-zero.

Prerequisites​

Install KEDA v2 in your cluster:

helm repo add kedacore https://kedacore.github.io/charts
helm install keda kedacore/keda --namespace keda --create-namespace

The kubeswarm operator detects KEDA at runtime. No Helm changes needed.

Configuration​

spec:
  runtime:
    autoscaling:
      minReplicas: 1            # idle floor
      maxReplicas: 10           # scale ceiling
      targetPendingTasks: 5     # tasks per replica before scale-up

When autoscaling is set, runtime.replicas is ignored.

How it works​

  1. The operator creates a KEDA ScaledObject targeting the agent's Deployment
  2. KEDA monitors the Redis Stream length (pending tasks)
  3. When pending tasks exceed targetPendingTasks * currentReplicas, KEDA scales up
  4. When the queue drains, KEDA scales down to minReplicas

Scale to zero​

Set minReplicas: 0 to allow the agent to scale to zero when idle:

spec:
  runtime:
    autoscaling:
      minReplicas: 0
      maxReplicas: 5
      targetPendingTasks: 3

The first task in the queue triggers scale-up from 0. There is a cold-start delay while the pod starts.

Without KEDA​

If KEDA is not installed, use fixed replicas:

spec:
  runtime:
    replicas: 3

The operator gracefully skips ScaledObject creation when KEDA CRDs are not present.