@sentry/opentelemetry

Official Sentry SDK for OpenTelemetry

This package allows you to send your OpenTelemetry trace data to Sentry via OpenTelemetry SpanProcessors.

If you are using @sentry/node, OpenTelemetry support is included out of the box. This package is only necessary if you are setting up OpenTelemetry support for Sentry yourself.

Installation

npm install @sentry/opentelemetry

# Or yarn
yarn add @sentry/opentelemetry

Note that @sentry/opentelemetry depends on the following peer dependencies:

• @opentelemetry/api version 1.0.0 or greater
• @opentelemetry/core version 1.0.0 or greater
• @opentelemetry/semantic-conventions version 1.0.0 or greater
• @opentelemetry/sdk-trace-base version 1.0.0 or greater, or a package that implements that, like @opentelemetry/sdk-node.

Usage

This package exposes a few building blocks you can add to your OpenTelemetry setup in order to capture OpenTelemetry traces to Sentry.

This is how you can use this in your app:

• Initialize Sentry, e.g. @sentry/node!
• Call setupEventContextTrace(client)
• Add SentrySampler as sampler
• Add SentrySpanProcessor as span processor
• Add a context manager wrapped via wrapContextManagerClass
• Add SentryPropagator as propagator
• Setup OTEL-powered async context strategy for Sentry via setOpenTelemetryContextAsyncContextStrategy()

For example, you could set this up as follows:

import * as Sentry from '@sentry/node';
import {
  SentryPropagator,
  SentrySampler,
  SentrySpanProcessor,
  setupEventContextTrace,
  wrapContextManagerClass,
  setOpenTelemetryContextAsyncContextStrategy,
} from '@sentry/opentelemetry';
import { AsyncLocalStorageContextManager } from '@opentelemetry/context-async-hooks';
import { context, propagation, trace } from '@opentelemetry/api';

function setupSentry() {
  Sentry.init({
    dsn: 'xxx',
  });

  const client = Sentry.getClient();
  setupEventContextTrace(client);

  const provider = new BasicTracerProvider({
    sampler: new SentrySampler(client),
  });
  provider.addSpanProcessor(new SentrySpanProcessor());

  const SentryContextManager = wrapContextManagerClass(AsyncLocalStorageContextManager);

  // Initialize the provider
  trace.setGlobalTracerProvider(provider);
  propagation.setGlobalPropagator(new SentryPropagator());
  context.setGlobalContextManager(new SentryContextManager());

  setOpenTelemetryContextAsyncContextStrategy();
}

A full setup example can be found in node-experimental.

Links

• Official SDK Docs

Changelog

9.33.0

Important Changes

• feat: Add opt-in vercelAiIntegration to cloudflare & vercel-edge (#16732)

The vercelAiIntegration is now available as opt-in for the Cloudflare and the Next.js SDK for Vercel Edge. To use it, add the integration in Sentry.init

Sentry.init({
  tracesSampleRate: 1.0,
  integrations: [Sentry.vercelAIIntegration()],
});

And enable telemetry for Vercel AI calls

const result = await generateText({
  model: openai('gpt-4o'),
  experimental_telemetry: {
    isEnabled: true,
  },
});

• feat(node): Add postgresjs instrumentation (#16665)

The Node.js SDK now includes instrumentation for Postgres.js.

• feat(node): Use diagnostics channel for Fastify v5 error handling (#16715)

If you're on Fastify v5, you no longer need to call setupFastifyErrorHandler. It is done automatically by the node SDK. Older versions still rely on calling setupFastifyErrorHandler.

Other Changes

• feat(cloudflare): Allow interop with OpenTelemetry emitted spans (#16714)
• feat(cloudflare): Flush after waitUntil (#16681)
• fix(nextjs): Remove ai from default server external packages (#16736)

Work in this release was contributed by @0xbad0c0d3. Thank you for your contribution!