K8S Autoscaling

Kubernetes Autoscaling

Comprehensive autoscaling using HPA, VPA, and KEDA with kubectl-mcp-server tools.

Quick Reference

HPA (Horizontal Pod Autoscaler)

Basic CPU-based scaling:

apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
  name: my-app-hpa
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: my-app
  minReplicas: 2
  maxReplicas: 10
  metrics:
  - type: Resource
    resource:
      name: cpu
      target:
        type: Utilization
        averageUtilization: 70

Apply and verify:

apply_manifest(hpa_yaml, namespace)
get_hpa(namespace)

VPA (Vertical Pod Autoscaler)

Right-size resource requests:

apiVersion: autoscaling.k8s.io/v1
kind: VerticalPodAutoscaler
metadata:
  name: my-app-vpa
spec:
  targetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: my-app
  updatePolicy:
    updateMode: "Auto"

KEDA (Event-Driven Autoscaling)

Detect KEDA Installation

keda_detect_tool()

List ScaledObjects

keda_scaledobjects_list_tool(namespace)
keda_scaledobject_get_tool(name, namespace)

List ScaledJobs

keda_scaledjobs_list_tool(namespace)

Trigger Authentication

keda_triggerauths_list_tool(namespace)
keda_triggerauth_get_tool(name, namespace)

KEDA-Managed HPAs

keda_hpa_list_tool(namespace)

See KEDA-TRIGGERS.md for trigger configurations.

Common KEDA Triggers

Queue-Based Scaling (AWS SQS)

apiVersion: keda.sh/v1alpha1
kind: ScaledObject
metadata:
  name: sqs-scaler
spec:
  scaleTargetRef:
    name: queue-processor
  minReplicaCount: 0  # Scale to zero!
  maxReplicaCount: 100
  triggers:
  - type: aws-sqs-queue
    metadata:
      queueURL: https://sqs.region.amazonaws.com/...
      queueLength: "5"

Cron-Based Scaling

triggers:
- type: cron
  metadata:
    timezone: America/New_York
    start: 0 8 * * 1-5   # 8 AM weekdays
    end: 0 18 * * 1-5    # 6 PM weekdays
    desiredReplicas: "10"

Prometheus Metrics

triggers:
- type: prometheus
  metadata:
    serverAddress: http://prometheus:9090
    metricName: http_requests_total
    query: sum(rate(http_requests_total{app="myapp"}[2m]))
    threshold: "100"

Scaling Strategies

Cost-Optimized Autoscaling

Scale to Zero with KEDA

Reduce costs for idle workloads:

keda_scaledobjects_list_tool(namespace)
# ScaledObjects with minReplicaCount: 0 can scale to zero

Right-Size with VPA

Get recommendations and apply:

get_resource_recommendations(namespace)
# Apply VPA recommendations

Predictive Scaling

Use cron triggers for known patterns:

# Scale up before traffic spike
triggers:
- type: cron
  metadata:
    start: 0 7 * * *  # 7 AM
    end: 0 9 * * *    # 9 AM
    desiredReplicas: "20"

Multi-Cluster Autoscaling

Configure KEDA across clusters:

keda_scaledobjects_list_tool(namespace, context="production")
keda_scaledobjects_list_tool(namespace, context="staging")

Troubleshooting

HPA Not Scaling

get_hpa(namespace)
get_pod_metrics(name, namespace)  # Metrics available?
describe_pod(name, namespace)     # Resource requests set?

KEDA Not Triggering

keda_scaledobject_get_tool(name, namespace)  # Check status
get_events(namespace)                        # Check events

Common Issues

Best Practices

1. Always Set Resource Requests

   • HPA requires requests to calculate utilization

2. Use Multiple Metrics

   • Combine CPU + custom metrics for accuracy

3. Stabilization Windows

   • Prevent flapping with scaleDown stabilization

4. Scale to Zero Carefully

   • Consider cold start time
   • Use activation threshold

Related Skills

• k8s-cost - Cost optimization
• k8s-troubleshoot - Debug scaling issues