@inngest/middleware-sentry

This package provides a middleware for Inngest to interface with Sentry.

Features

• Capture exceptions for reporting
• Add tracing to each function run
• Include useful context for each exception and trace like function ID and event names

Installation

npm install @inngest/middleware-sentry

[!NOTE] Requires inngest@>=4.0.0 and @sentry/*@>=8.0.0

Usage

To use the middleware, pass the SentryMiddleware class to your Inngest client. It assumes that Sentry has already been initialized elsewhere in your code using Sentry.init().

import * as Sentry from "@sentry/node";
import { Inngest } from "inngest";
import { SentryMiddleware } from "@inngest/middleware-sentry";

// Initialize Sentry as usual wherever is appropriate
Sentry.init(...);

const inngest = new Inngest({
  id: "my-app",
  middleware: [SentryMiddleware],
});

To customize options, use the sentryMiddleware() factory:

import { sentryMiddleware } from "@inngest/middleware-sentry";

const inngest = new Inngest({
  id: "my-app",
  middleware: [sentryMiddleware({ onlyCaptureFinalAttempt: false })],
});

Options

captureStepErrors

Default: false

When true, step-level errors are captured as separate Sentry events in addition to function-level errors. Each step error event is tagged with inngest.error.source: "step" and inngest.step.name so you can distinguish them from function-level events (inngest.error.source: "run").

When false, step errors are still visible as error spans and breadcrumbs in traces, but only function-level errors produce Sentry events.

onlyCaptureFinalAttempt

Default: true

When true, exceptions are only sent to Sentry on the final attempt of a step or function run (when retries are exhausted or the error is non-retriable). Intermediate retry attempts still set span error status and add breadcrumbs, but won't create Sentry events. Set to false to capture every attempt.

disableAutomaticFlush

Default: false

By default, the middleware will force Sentry to flush as part of the Inngest request, ensuring all events, exceptions, and traces are sent before a response is returned.

This is important for serverless environments where the runtime doesn't wait for the event loop to be empty and background tasks such as sending exception data may be lost.

If you're not in a serverless runtime or otherwise wish to have Sentry handle flushing itself, set this to true.