Deployment

Deployment is about provisioning: how you stand the server, workers, and database up in the first place. Once the processes are running, Operate takes over with the day-to-day concerns — auth, scheduling, observability, backups, capacity.

The pages here are ordered from “running on your laptop” to “running on someone else’s cloud account, codified.”

Top-level covers the patterns most teams need: Local development for the desktop loop, Docker for a portable single-host stack, Kubernetes for the cluster shape, High availability for running multiple server replicas against PostgreSQL, and the Production checklist for the boxes you must tick before exposing a server to real traffic.

Cloud deployments are starter recipes for the three public clouds: AWS, GCP, and Azure. They describe self-hosted Flux on managed infrastructure — not Flux Cloud, which is the separate managed offering. Each recipe is the smallest viable shape that respects Flux’s constraints (durable PostgreSQL, persistent bootstrap-token, sticky routing for worker connections when running more than one server replica). They are not vendor-comprehensive guides.

Infrastructure as code covers the Terraform and Pulumi shapes for codifying the above. Flux 0.56.0 does not ship official modules; the pages describe the module surface a production user would author.

Once the server and workers are running, head to Operate for everything that comes after.