API Maker

The backend engine for the AI era

Chapters

What you will learn

  • Local and Cloud mode, over SSH with no agent
  • A server added, a project and its first environment
  • Four layouts, and what is still missing
  • API Maker v4 on the servers
  • Review and deploy, live
  • What lands on a server, the firewall, native or Full Docker
In the docsRun on your computer
Transcript949 words · click a time to play from there

Intro

The Cloud mode of API Maker Launchpad installs API Maker on servers you rent anywhere. It goes from one server to a load balanced cluster, and keeps them running. Everything happens over SSH, from your computer.

Servers, the usual way

Running a backend on your own servers usually means trouble. You keep a wiki of shell commands, one for each server. Servers drift apart, each set up a little differently. You install an agent or a control plane, and pay for it. And you hope the deployment will work. With Launchpad, every server is set up from one window. Every change is planned, and shown server by server before it runs.

It uses SSH only, with no agent. And the deployment runs in stages, with the output of every server, live. In this video: Cloud mode, a server added, a project and its layout, API Maker version four on the servers, the review, the deployment, and what lands on a server.

Cloud mode

At the top of the window are Local and Cloud. Cloud shows the projects, each with its environments. Here is production of Calm Home. It is deployed, with its version of API Maker, and its address. Below are the latest operations. Servers shows every server, its system, its size, and when it was last seen. Operations lists everything done to your servers. And there are the Cloud settings. Everything happens over SSH, from this computer.

Nothing is installed here, and there is no agent and no Launchpad server. Passwords and keys are encrypted with AES two five six G C M. The key is kept in the keychain of your system, never in the data of the projects.

A server added

Add a server. It runs Ubuntu twenty four point zero four, on x86 or ARM. Open the email your provider sent, and paste it. The form fills itself in: the address, the user, the password, and a name. A private key, the SSH agent, or a jump server work too. Test the connection. The system, the memory, the disk, and the rights are checked, and the host key of the server is shown.

Once you trust it, it is pinned. A server that shows another key is refused. It is saved.

A project and its layout

On a fresh computer, there is no project yet. Click New project, name it Calm Home, and its first environment, production, opens. Choose a layout: how many servers, and what runs on each. All in one: one server runs everything. It is the quickest way to start. There, API Maker can run in Docker too, or natively, as a systemd service. App plus data: API Maker and the gateway on one server, and the databases on another.

Dedicated databases: API Maker, its own MongoDB and Redis, and your databases, on three servers. Load balanced: a gateway, API Maker on as many servers as you need, and the data. Click Use this layout. The architecture is drawn. Launchpad lists what is still missing before a deployment: the servers to connect, and the license key of each API Maker server. The servers of the layout are planned. Connect each one, or use servers you already added.

API Maker v4 on the servers

Here is production, with its servers connected, but not deployed yet. In its settings, a new environment runs API Maker version four. That is one program on each server, the Linux build for its processor, with no Node.js on the server. And every API Maker server has its own license key. No key is shared by the servers of an environment.

The review

Click Review and deploy. Launchpad reads the servers first. Then it lists, server by server, what is installed, updated, or removed, what restarts, and whether API Maker stops in the meantime. Data is never deleted unless you tick it, and nothing changes before you click Deploy. In production, a change to what already runs asks for the name of the environment first.

The deployment

Click Deploy. The operation shows its stages and its steps, with the output of every server, live. Up to six servers work at the same time. API Maker downloads while the databases start, every server pulls its images at once, and a server with enough memory starts its databases side by side. You can stop a deployment, and deploying again goes on from where it stopped. It ends when API Maker answers through its entry, and each database gets its instance in API Maker.

What lands on a server

What lands on a server? Its configuration, in root config: the env file, a compose file for each service, and the reverse proxy. The data of every container, in docker data. The logs. And API Maker, in root projects. systemd runs the API Maker program. It starts it at boot, and again after a crash, and its log is rotated past fifty megabytes. The firewall is set up too. The web reaches only the reverse proxies.

Every other port is open only to the servers that use it. And SSH is never closed. Natively, API Maker is a systemd service, and the environment can spread over as many servers as you like. In Full Docker, API Maker runs in a container on the network of its server, with everything else, on one server. Switching is planned like any change. If the new one does not answer, the one before starts again.

Recap

To sum up. You run your servers from one window, over SSH, with no agent. You start from a layout, and Launchpad says what is missing. API Maker version four is one program on each server, with no Node.js. Every change is reviewed server by server before it runs. And the deployment runs in stages, live. Next: clusters, load balancing, and reverse proxies. API Maker. Build faster, spend smarter, deliver better.