CloudWatch Container Insights

Container Insights collects metrics, logs, and health data from your cluster into CloudWatch.

CloudWatch Container Insights gives you observability for EKS: CPU, memory, disk, and network metrics for clusters, nodes, pods, and containers, plus centralized logs — all in CloudWatch dashboards.

What it collects

  • Metrics — resource usage at every level, ready for dashboards and alarms.
  • Logs — container stdout/stderr shipped to CloudWatch Logs via Fluent Bit.
  • Performance events — for spotting bottlenecks and failures.

How to enable it

Install the CloudWatch Observability add-on (or the CloudWatch agent and Fluent Bit as a DaemonSet). Pods need IRSA permission to write to CloudWatch.

Example

Example · bash
# Enable observability via the managed add-on
aws eks create-addon \
  --cluster-name demo \
  --addon-name amazon-cloudwatch-observability \
  --service-account-role-arn arn:aws:iam::111122223333:role/CloudWatchAgentRole

kubectl get pods -n amazon-cloudwatch

When to use it

  • An SRE team enables Container Insights to create CloudWatch dashboards showing pod CPU, memory, and network metrics without running their own Prometheus stack.
  • A platform team sets up Container Insights log groups for each namespace so engineers can query application logs and metrics in a single CloudWatch view.
  • An on-call engineer uses Container Insights during an incident to quickly identify which pod is consuming abnormal CPU across a 100-node cluster.

More examples

Enable Container Insights via eksctl

Enables CloudWatch Container Insights for logs and metrics on the cluster and verifies the DaemonSet agent is deployed on all nodes.

Example · bash
eksctl utils enable-observability \
  --cluster prod \
  --enable-logging \
  --enable-metrics
kubectl get daemonset amazon-cloudwatch-observability-agent \
  -n amazon-cloudwatch

Install CloudWatch agent via Helm

Installs the CloudWatch Observability agent via Helm, which collects Container Insights metrics and logs from every node.

Example · bash
helm repo add aws-observability https://aws-observability.github.io/helm-charts
helm install amazon-cloudwatch-observability \
  aws-observability/amazon-cloudwatch-observability \
  --namespace amazon-cloudwatch \
  --create-namespace \
  --set clusterName=prod \
  --set region=us-east-1

Query Container Insights metrics

Retrieves average pod CPU utilization from the ContainerInsights namespace for the past hour, useful for trending and alerting.

Example · bash
aws cloudwatch get-metric-statistics \
  --namespace ContainerInsights \
  --metric-name pod_cpu_utilization \
  --dimensions Name=ClusterName,Value=prod \
  --start-time $(date -u -d '1 hour ago' +%Y-%m-%dT%H:%M:%S) \
  --end-time $(date -u +%Y-%m-%dT%H:%M:%S) \
  --period 300 \
  --statistics Average

Related

eks best practiceseks addons

Discussion

  • Be the first to comment on this lesson.