# Before production

> The checklist to go through before you open your application to the public: cluster, code, database, variables, domain, budget, access.

Source: https://docs.fransys.cloud/en/demarrer/avant-la-production/ Checked against the console on 2026-10-23.

## The cluster

- [ ] The criticality is **Stable** or **Critical**, not **Basic**. See [Criticality](https://docs.fransys.cloud/en/clusters/criticite/).
- [ ] The region is the one you want for your data.
- [ ] Machine management is set: AutoPilot with a cap that leaves some headroom, or well-sized fixed machines.
- [ ] The cluster is **Online**, with all its machines ready.

## The code

- [ ] The service is linked to your repository, the CI merge request is merged, and the pipeline passes.
- [ ] A push does deploy the new version.
- [ ] The startup and readiness probes give the application enough time to start.

## The database

- [ ] The database is managed by Fransys and linked to the application (no home-made database container).
- [ ] The backup schedule and retention suit you, and a **completed** backup exists.
- [ ] You've tried a restore, on a test database.
- [ ] Remote access is turned off.

## Variables

- [ ] No secrets in the code or in the image: everything goes through **protected** variables.
- [ ] Connection variables come from links, not from copied values.

## The domain

- [ ] Your domain is added, the DNS record is created, and the HTTPS padlock shows.

## The budget

- [ ] Automatic payment is on, with a monthly limit you've checked. See [Never get shut down](https://docs.fransys.cloud/en/budget/ne-jamais-etre-coupe/).
- [ ] Your billing details are filled in.

## Access

- [ ] Two-factor authentication is on for your team's accounts.
- [ ] Each agent has its own key, and unused keys are removed.

## Monitoring

- [ ] The **Monitoring** page is all green, with no repeated restarts.
- [ ] The service's logs show no errors at startup.