Queue consumers, Celery or Temporal workers and self-hosted CI runners can now run as managed containers on Vercel without an HTTP server, configured with a daemons entry and per-region replica counts.
Key takeaways · AI-distilled
Supported consumers named in the post include Vercel Queues poll mode, Celery, BullMQ, RabbitMQ, and Kafka, plus Cadence workers and CI runners for GitHub Actions, GitLab, and CircleCI.
Daemons run as containers on Fluid compute and deploy alongside the app in the same Vercel project.
Setup needs a Dockerfile or Containerfile plus a daemons entry in one root config file (vercel.toml, vercel.ts, or vercel.json); the example sets replicas.iad1 = 1 for one Celery worker.
Workers read connection details from production environment variables and write output to the project's Logs, while CPU and memory metrics require Observability Plus.
