Skip to content

First deployment

A first application online, over HTTPS, on a European cluster. The console's “First launch” flow, step by step.

Checked against the console on 23 Oct 2026

You need a Fransys account and credits: your balance shows in the left column of the flow. Credits pay for what your machines use, by the hour; there’s no subscription. To add some, see Credits and automatic payment.

The First launch flow opens right after you sign up. It has four steps: Your first service, Where to host it, Going live, What’s next. You can skip it with Skip to the dashboard.

  1. Choose your first service

    Three options:

    • An nginx page: the simplest way to see the whole chain working. You’ll replace it afterwards.
    • Your Docker image: published on Docker Hub, GitHub or another public registry. Enter the image (for example ghcr.io/your-team/app:latest) and, if needed, the port the app listens on.
    • From a Git repository: not in this flow yet. See Deploy your code.

    A private image? You’ll add registry access in the service settings, once the flow is done.

  2. Choose where to host it

    • On a cluster that’s already online: no extra machines, the service shares that cluster with your other projects.
    • On a new cluster: the smallest the hosting provider accepts. You can scale it up later.

    The console shows the monthly cost while the cluster runs, and how long your credits will cover it. Give the project a name.

  3. Go live

    Review the summary, then click Go live. Fransys works through:

    • reserving the machines from the hosting provider;
    • installing Kubernetes (up to 15 minutes);
    • creating the project and preparing the service;
    • deploying, then checking that the service responds.

    You can close the page: installation carries on, and the console lets you know when the service responds.

  4. Open your application

    The service’s address appears in its settings, General tab. It’s already on HTTPS.

The flow shows A step got stuck, with the reason. Fix what it points to, then click Resume: the flow picks up from the stuck step, nothing is created twice.

The most common reasons:

  • The service won’t start: its logs will say why. Open the service, then its status badge.
  • The cluster is taking more than 25 minutes to start: write to us from the help button.

Right after it’s created, the cluster may stay Degraded for a few minutes. That’s normal: you can already deploy.

On the Today page, the Getting started card offers the same path in five steps, about 20 minutes: connect your AI (optional), create your project, create your cluster, deploy, then make a first update. Each step also offers a sentence to give your agent. You’ll find it again in your account menu, Getting started.