Job Queue

Vendure uses a background job queue to offload long-running or resource-intensive tasks from the main request-response cycle. Instead of making an API consumer wait while a heavy operation completes, the server adds a job to the queue and returns immediately. A separate worker process then picks up and executes the job asynchronously.

Architecture

The job queue system has two sides:

The Server — receives API requests and enqueues jobs when an operation requires background processing. For example, when a product is updated, the server enqueues a job to update the search index rather than doing it synchronously.
The Worker — a separate NestJS standalone application that listens to the queue and processes jobs. The worker runs the same Vendure plugins and services as the server, but it is optimized for background processing rather than handling HTTP requests.

This separation means the API remains responsive even when the system is processing heavy workloads like bulk imports or large-scale index rebuilds.

Built-in uses

Several core features rely on the job queue:

Search index updates — when products, variants, or collections change, the search index is updated via background jobs rather than inline with the API request.
Collection filter updates — recalculating which products belong to a collection can be expensive, so this work is queued.
Email sending — transactional emails (order confirmation, shipping notification, etc.) are dispatched as background jobs to avoid slowing down the triggering operation.

Plugins can also add their own job types to the queue for any custom background processing needs.

Job queue strategies

The underlying storage and processing mechanism for the job queue is controlled by a JobQueueStrategy. Vendure supports several strategies:

In-memory — jobs are held in process memory. Simple and requires no infrastructure, but jobs are lost if the process restarts. Suitable only for development.
SQL-based (DefaultJobQueuePlugin) — jobs are persisted to the database. Reliable and works with any supported database. A good default for small to medium deployments.
BullMQ / Redis — jobs are managed by Redis via the BullMQ library. Provides the best performance and is recommended for production workloads with high job volumes.

Running multiple workers

For larger deployments, you can run multiple worker instances to increase throughput. Workers can also be configured to only process specific queue types, allowing you to dedicate resources to high-priority or resource-intensive jobs.

For example, you might run one worker dedicated to search indexing and another for email delivery, ensuring that a flood of emails never delays search index updates.

Monitoring

Each job tracks its state through a lifecycle: PENDING, RUNNING, COMPLETED, FAILED, or CANCELLED. The Admin API provides queries to inspect job status, making it straightforward to build monitoring dashboards or alert on failed jobs.