Vercel Daemons is now in public beta
Vercel has launched Vercel Daemons in public beta, enabling long-running background processes without HTTP servers, integrating with queues and CI systems while running on Fluid compute.
Vercel Daemons, now in public beta, allows developers to run persistent background processes that interact with Vercel Queues, databases, or third-party systems without relying on HTTP servers. The feature supports polling modes for systems like Celery, BullMQ, RabbitMQ, Kafka, Temporal, or Cadence, as well as self-hosted CI runners for GitHub Actions, GitLab, or CircleCI. These daemons execute within containers on Vercel’s Fluid compute platform, deployed alongside applications in the same project. Vercel manages instance counts per region, automatically restarting failed or exited processes to maintain availability.
To use Vercel Daemons, developers package their application and dependencies in a Dockerfile or Containerfile, then add a daemons entry to a configuration file at the project root. The example provided configures one Celery worker in the iad1 region (Washington, D.C.) using worker/Dockerfile. This setup ensures the worker runs persistently without requiring inbound connections or manual intervention.
After configuring the Dockerfile, developers set connection details for the worker in the project’s production environment variables. These variables are then deployed to production, where Vercel automatically starts the configured instances. The output from these daemons is visible in the project’s Logs, providing visibility into their operation and any errors that may occur.
Vercel Daemons also integrates with Observability Plus, offering CPU and memory metrics for monitoring performance. This feature is designed to simplify the management of long-running tasks by offloading infrastructure concerns to Vercel, while providing developers with the tools needed to track and optimize their background processes.