Amazon EKS — Observability

You can’t fix what you can’t see. EKS observability covers metrics, logs, and traces — the three pillars that give you full visibility into cluster and application health.

---

Topics Covered

#	Topic	Description
1	CloudWatch Container Insights	AWS-native metrics and dashboards
2	Logging with Fluent Bit	Centralized log collection
3	Prometheus & Grafana	Open-source metrics and visualization
4	AWS Distro for OpenTelemetry	Unified telemetry collection
5	AMP & AMG	Managed Prometheus and Grafana
6	Hands-On Lab	Set up full observability stack
7	Practice Questions	Test your observability knowledge

---

Observability Tools Comparison

Tool	Type	Managed	Cost	Best For
CloudWatch Container Insights	Metrics + Logs	Fully	Per metric/log ingested	AWS-native, low effort
Fluent Bit	Log forwarder	Self (DaemonSet)	Free (agent) + destination costs	Flexible log routing
Prometheus + Grafana	Metrics + Dashboards	Self	Free (compute costs only)	K8s-native, custom metrics
AMP + AMG	Metrics + Dashboards	Fully	Per metric sample ingested	Managed Prometheus at scale
ADOT	Telemetry collector	Self (DaemonSet)	Free (agent) + destination costs	Unified metrics + traces + logs
X-Ray	Distributed tracing	Fully	Per trace recorded	Application tracing

---

EKS Control Plane Logging

Enable control plane logs before anything else:

aws eks update-cluster-config \
  --name my-cluster \
  --logging '{
    "clusterLogging": [{
      "types": ["api", "audit", "authenticator", "controllerManager", "scheduler"],
      "enabled": true
    }]
  }'

Log Type	What It Captures
api	API server requests
audit	Who did what, when (critical for security)
authenticator	IAM authentication events
controllerManager	Controller reconciliation events
scheduler	Pod scheduling decisions

Cost tip: audit and authenticator are the most useful. Enable those first; add others as needed.

---

Start Learning →