Skip to content

Latest commit

 

History

History
103 lines (75 loc) · 3.57 KB

getting-started.md

File metadata and controls

103 lines (75 loc) · 3.57 KB
Raw

Get started

You can instrument a Go executable using OpenTelemetry without having to write additional code. All you need to do is configure a few environment variables and run the instrumentation with elevated privileges.

The following example shows how to instrument a Go application automatically on a Linux host, through Docker, and using Kubernetes.

Prerequisites

To instrument an application automatically, you need the following:

  • Linux with kernel version 4.19 or higher
  • x64 or ARM processor
  • Docker image or compiled binary of OpenTelemetry Go Automatic Instrumentation
  • Go 1.18 or higher

To compile the instrumentation binary, run make build.

Instrument an application on the same host

To instrument an application on the same host, follow these steps:

  1. Run the target application.

  2. Set the following environment variables before running the instrumentation:

  • OTEL_GO_AUTO_TARGET_EXE: Full path of the executable you want to instrument. For example, /home/bin/service_executable
  • OTEL_SERVICE_NAME: Name of your service or application
  • OTEL_EXPORTER_OTLP_ENDPOINT: Your observability backend. For example, http://localhost:4318.

For example:

sudo OTEL_GO_AUTO_TARGET_EXE=/home/bin/service_executable OTEL_SERVICE_NAME=my_service OTEL_EXPORTER_OTLP_ENDPOINT=http://localhost:4318 ./otel-go-instrumentation`
  1. Run the OpenTelemetry Go Automatic Instrumentation with root privileges.

Note If the target application isn't running yet, the instrumentation waits for the process to start.

Instrument an application in Docker Compose

To instrument a containerized application, follow these steps:

  1. Create or edit the docker-compose.yaml file. Make sure to add a Docker network, a shared volume, and a service for the application.

  2. Edit the docker-compose file to add a new service for the instrumentation:

  go-auto:
    image: otel/autoinstrumentation-go
    privileged: true
    pid: "host"
    environment:
      - OTEL_EXPORTER_OTLP_ENDPOINT=http://<address_in_docker_network>:4318
      - OTEL_GO_AUTO_TARGET_EXE=<location_of_target_application_binary>
      - OTEL_SERVICE_NAME=<name_of_your_application>
      - OTEL_PROPAGATORS=tracecontext,baggage
    volumes:
      - <shared_volume_of_application>
      - /proc:/host/proc
  1. Run docker compose up.

Instrument an application in Kubernetes

To instrument an application running in Kubernetes, follow these steps:

  1. Add the container image of the OpenTelemetry Go Automatic Instrumentation to your manifest. Make sure that runAsUser is set to 0, privileged is set to true:

    - name: autoinstrumentation-go
  image: otel/autoinstrumentation-go
  imagePullPolicy: IfNotPresent
  env:
    - name: OTEL_GO_AUTO_TARGET_EXE
      value: <location_of_target_application_binary>
    - name: OTEL_EXPORTER_OTLP_ENDPOINT
      value: "http://<address_in_network>:4318"
    - name: OTEL_SERVICE_NAME
      value: "<name_of_service>"
  securityContext:
    runAsUser: 0
    privileged: true

  2. Check if the configuration shareProcessNamespace is present in the pod spec, if not, please add it.

  3. Deploy the application and the instrumentation using the manifest.

Configuration

See the documentation for InstrumentationOption factory functions for information about how to configure the OpenTelemetry Go Automatic Instrumentation.