Making a BullMQ processor safe to run twice

[violetta98]

"My job ran twice" usually isn't a BullMQ bug. A job stalls and comes back, a worker dies mid-job, someone re-adds the same jobId. The fix is an idempotent processor, which is harder to get right than it looks.

import { InMemoryBackend } from "idemkit";
import { bullmqProcessor } from "idemkit/bullmq";

export const processCharge = bullmqProcessor(
  async (job) => ({ receipt: await chargeCustomer(job.data) }),
  { backend: new InMemoryBackend(), lockDurationSeconds: 30 },
);

Dedupes on job.id. The lease is derived from your lockDuration and kept shorter, so the claim expires before BullMQ decides the job stalled. Attempts come from job.attemptsMade, so the count survives a worker restart.

Same library covers HTTP endpoints and plain function calls, on Redis, Postgres, Mongo or DynamoDB. Apache-2.0: https://github.com/idemkit/idemkit

If you find it useful, a star would be much appreciated.