Retries and timeouts

Retries, timeouts, and terminal failures

Attempts use exponential backoff with equal jitter, stop after five attempts, and time out after five minutes by default. The retry delay function fully controls scheduling and can be overridden globally or per method on either helper:

import { exponential, jitter } from 'durability/utils';

const durability = new Durability({
  context: this.ctx,
  handlers,
  attemptTimeoutMs: 60_000,
  retries: {
    maxAttempts: 5,
    delay: (attempt) => jitter(exponential(attempt)),
  },
  methods: {
    resizeImage: {
      attemptTimeoutMs: 10 * 60_000,
      retries: {
        maxAttempts: 2,
        delay: (attempt) => exponential(attempt, 500, 30_000),
      },
    },
  },
});

Each attempt receives its own AbortSignal. A timed-out operation follows the normal retry policy; a timed-out named alarm is terminal unless retryTimeouts is set. If a handler ignores abort, its in-memory lock and concurrency permit remain held until it actually settles, so a same-isolate retry cannot overlap it.

Throw NonRetryableError to move work directly to failed without another attempt. Errors named NonRetryableError from other packages, such as cloudflare:workflows, are recognized too.