Building a Kubernetes Operator for Gati

How Gati's Kubernetes operator automates deployment, scaling, and version management.

Why an Operator?

Manual Kubernetes management is error-prone. Gati's operator automates:

• Handler deployment
• Auto-scaling configuration
• Version lifecycle management
• Health monitoring

Custom Resource Definitions

GatiHandler CRD

apiVersion: gati.dev/v1
kind: GatiHandler
metadata:
  name: users-api
spec:
  path: /users/:id
  replicas: 3
  autoscaling:
    enabled: true
    minReplicas: 2
    maxReplicas: 10
    targetCPU: 70

GatiVersion CRD

apiVersion: gati.dev/v1
kind: GatiVersion
metadata:
  name: users-v2
spec:
  handler: users-api
  timestamp: "2025-11-22T10:00:00Z"
  active: true
  traffic: 100

Operator Logic

Reconciliation Loop

async reconcile(handler: GatiHandler) {
  // 1. Ensure deployment exists
  await ensureDeployment(handler);
  
  // 2. Configure HPA
  if (handler.spec.autoscaling.enabled) {
    await ensureHPA(handler);
  }
  
  // 3. Update service
  await ensureService(handler);
  
  // 4. Monitor health
  await checkHealth(handler);
}

Version Management

async manageVersions(handler: GatiHandler) {
  const versions = await getVersions(handler);
  
  // Deactivate cold versions
  for (const v of versions) {
    if (isCold(v)) {
      await deactivateVersion(v);
    }
  }
  
  // Update traffic split
  await updateTrafficSplit(versions);
}

Benefits

1. Declarative - Describe desired state, operator handles it
2. Self-healing - Automatic recovery from failures
3. Version-aware - Manages multiple versions automatically
4. Observable - Built-in metrics and events

Deployment

# Install operator
kubectl apply -f https://gati.dev/operator.yaml

# Deploy handler
kubectl apply -f handler.yaml

# Operator handles the rest

Related

• Kubernetes Guide
• HPA and Ingress
• Deployment Guide