reliability.md

Reliability: timeouts, retries, idempotency, and domains

Set poll, response, and execution timeouts on every task definition. Retry only idempotent or compensated work; use domain to route resource-bound work to isolated worker pools.

@worker({ taskDefName: "validate_order" })
async function validateOrder(task: Task) {
  const order = await getOrder(task.inputData.orderId);
  if (!order) {
    throw new NonRetryableException("Order not found"); // FAILED_WITH_TERMINAL_ERROR
  }
  return { status: "COMPLETED" as const, outputData: { validated: true } };
}

A worker may receive the same task more than once. Persist idempotency keys before external side effects. Long-running tasks should return IN_PROGRESS with setCallbackAfter/lease extension (see LEASE_EXTENSION.md) rather than blocking the poll loop. retryServerErrors/CONDUCTOR_RETRY_SERVER_ERRORS opts idempotent HTTP methods into retrying transient 502/503/504 responses from the server itself.

Verify behavior with failure-path tests (workflow testing) and inspect task reasonForIncompletion before retrying — see debugging.