Jobs and shutdown

Operations · Jobs and shutdown

Operational objective — Make background work finish, retry, pause, or fail with a known outcome when the process changes state.

Idempotent jobs, signal handling, draining, retry policy, and bounded shutdown.

Key concepts

Retry

Transient only

Drain

Finite window

Abandon

Recorded

Operational surface

Evidence over assumption

Signal	Healthy interpretation	Action when it degrades
Retry	Transient only	Do not duplicate irreversible effects.
Drain	Finite window	Bound deploy and restart time.
Abandon	Recorded	Make recovery visible and actionable.

Operational context

A process exit is normal infrastructure behavior; work should not depend on a lucky timing window.

Operator’s principle Graceful shutdown is a correctness feature for any system that performs side effects.

Implementation notes

Engineering move	Guidance
Classify work	Separate safe retry, manual retry, and non-retryable failures.
Persist intent	Store enough state to resume or explain a job after restart.
Stop intake	Reject or defer new work when shutdown begins.
Drain with a deadline	Finish bounded work and record what was abandoned.

Decision lens

Mode	Practical emphasis
Short job	May complete inside the request if cancellation is safe.
Durable job	Needs queue, status, retry, and idempotency.
Shutdown	Needs signal, drain, deadline, and forced-stop behavior.

Reference queue

InMemoryJobQueue is a reference implementation for local development and adapter conformance. It supports bounded concurrency, idempotency keys, retry attempts, configurable retry delay, and an inspectable dead-letter snapshot. It is intentionally process-local; use a durable provider for production or multiple instances.

const queue = new InMemoryJobQueue({
  concurrency: 4,
  handler: async (job) => processPayment(job.payload),
});

await queue.enqueue('payment', payload, {
  idempotencyKey: `payment:${paymentId}`,
  maxAttempts: 5,
});

layer-group · Choose job storage — Read the focused guide for this boundary.
stream · Cancel request work — Read the focused guide for this boundary.
cloud-arrow-up · Deploy with drain — Read the focused guide for this boundary.

References

Last modified on September 10, 2026

Start

Core

Platform

Reference

Operations

Jobs and shutdown

Jobs and shutdown

Operations · Jobs and shutdown

Key concepts

Retry

Drain

Abandon

Operational surface

Evidence over assumption

Operational context

Implementation notes

Decision lens

Reference queue

References

​Jobs and shutdown

​Operations · Jobs and shutdown

​Key concepts

Retry

Drain

Abandon

​Operational surface

​Evidence over assumption

​Operational context

​Implementation notes

​Decision lens

​Reference queue

​Related topics

​References

Jobs and shutdown

Operations · Jobs and shutdown

Key concepts

Operational surface

Evidence over assumption

Operational context

Implementation notes

Decision lens

Reference queue

Related topics

References