Google Cloud Trace

Google Cloud Trace collects and analyzes latency data from distributed applications. PURISTA sends traces via the Google Cloud Trace OTLP exporter — no global SDK initialization required.

Install

npm install @google-cloud/opentelemetry-cloud-trace-exporter @opentelemetry/sdk-trace-node

TypeScript setup

The Cloud Trace exporter implements the SpanExporter interface, so you wrap it in a SimpleSpanProcessor and pass it directly to PURISTA.

import { TraceExporter } from '@google-cloud/opentelemetry-cloud-trace-exporter'
import { SimpleSpanProcessor } from '@opentelemetry/sdk-trace-node'

export function getSpanProcessor() {
  // Uses Application Default Credentials automatically
  const exporter = new TraceExporter({
    projectId: process.env.GOOGLE_CLOUD_PROJECT,
  })

  return new SimpleSpanProcessor(exporter)
}

Wire it into your application:

import { getSpanProcessor } from './tracing.js'
import { AmqpBridge } from '@purista/amqpbridge'

const spanProcessor = getSpanProcessor()

const eventBridge = new AmqpBridge({ spanProcessor })
await eventBridge.start()

const myService = await myV1Service.getInstance(eventBridge, { spanProcessor })
await myService.start()

Authentication

The exporter uses Application Default Credentials:

• Local development: gcloud auth application-default login
• Cloud Run / GKE / Compute Engine: the default service account is used automatically
• CI/CD: set GOOGLE_APPLICATION_CREDENTIALS to the path of a service account JSON key

Required IAM permission

The service account needs the cloudtrace.traces.patch permission (included in the Cloud Trace Agent role).

Environment variables

Viewing traces

1. Open Cloud Trace in the GCP Console
2. Select your project
3. Use the trace list or waterfall view to navigate the PURISTA message flow across services

Resources

• Cloud Trace documentation
• OpenTelemetry Cloud Trace exporter