Get started with auto-instrumentation ¶

Auto-instrumentation injects an OpenTelemetry agent into your application at startup. The agent hooks into popular libraries and frameworks to collect traces and runtime metrics — without code changes. Log export via OpenTelemetry is available as an opt-in.

Supported runtimes: Java/Kotlin, Node.js, Python, and SDK-only mode for Go and other languages that provide their own OpenTelemetry setup.

What you get ¶

Once enabled, the agent automatically instruments:

Category	Examples (vary by runtime)
HTTP servers	Spring MVC, Ktor, Express, Flask, Koa
HTTP clients	OkHttp, Apache HttpClient, `node-fetch`, `requests`
Databases	PostgreSQL (JDBC, node-postgres), Valkey/Redis
Messaging	Kafka producers/consumers
gRPC	Client and server calls
Frameworks	Spring Boot, Ktor, Micrometer bridge

This data powers the Nais APM dashboards — service inventory, RED metrics (Rate/Errors/Duration), dependency maps, and runtime metrics — with no extra configuration.

Enable auto-instrumentation ¶

Add the following to your nais.yaml and deploy:

Java / Kotlin Node.js Python SDK only (Go, etc.)

yaml

spec:
  observability:
    autoInstrumentation:
      enabled: true
      runtime: java

yaml

spec:
  observability:
    autoInstrumentation:
      enabled: true
      runtime: nodejs

yaml

spec:
  observability:
    autoInstrumentation:
      enabled: true
      runtime: python

This sets up the OpenTelemetry environment variables but does not inject an agent. Use this when your application configures the OpenTelemetry SDK itself.

yaml

spec:
  observability:
    autoInstrumentation:
      enabled: true
      runtime: sdk

Verify it works ¶

After deploying, check that traces are flowing:

Open the Nais APM service inventory
Find your application in the list — it should appear within a few minutes
Click through to see RED dashboards, operations breakdown, and dependencies

If your app does not appear, check the auto-instrumentation configuration reference for troubleshooting environment variables, and make sure your application is receiving traffic.

You can also verify in Grafana Explore using the Tempo data source:

Plaintext

{resource.service.name="<your-app-name>"}

Next steps ¶

Add custom spans and metrics for business-specific tracing
Explore traces in Grafana Tempo
Avoid false errors from expected 4xx responses — fix inflated error rates in APM
📚 Auto-instrumentation configuration reference