Self-hosting Triplit

To enable sync, you need to run a Triplit server. The server is a Node.js application that talks to various Triplit clients over WebSockets and HTTP.

You have several options for running the server:

• A local development server
• Use Docker and a cloud provider that supports container deployments
• Use a one-click deploy to a cloud provider that is integrated with Triplit.
• Build a custom server and use a cloud provider that supports Git-based deploys

Docker

Each release of the server is published as a Docker image (opens in a new tab). You can deploy the server to a cloud provider like fly.io (opens in a new tab), DigitalOcean (opens in a new tab), or AWS. You'll also want to setup a volume to persist the database.

The docker file starts a node server on port 8080, and you can pass in the following environment variables to configure the server:

• NODE_OPTIONS - Node.js options for the server (e.g. --max-old-space-size=4096)

One-click deploy

Triplit is integrated with Railway (opens in a new tab), a cloud provider that supports one-click deploys. Read about how to deploy a Triplit server using Railway in our Triplit Cloud guide.

We plan on adding support for more cloud providers in the future. If you have a favorite cloud provider that you'd like to see integrated with Triplit, let us know by joining our Discord (opens in a new tab).

Building a custom server

The server is published as an NPM package, and you can install it by running:

npm install @triplit/server

The server also contains the remote Triplit database, which will persist data synced from your clients. The server supports different storage adapters, such as SQLite. Using the createServer function, you can create and configure a new server instance:

import { createServer } from '@triplit/server';
 
const port = +(process.env.PORT || 8080);
 
const startServer = createServer({
  storage: 'sqlite',
  verboseLogs: true,
});
 
const dbServer = startServer(port);
 
console.log('running on port', port);
process.on('SIGINT', function () {
  dbServer.close(() => {
    console.log('Shutting down server... ');
    process.exit();
  });
});

You can now deploy the server to a cloud provider that supports Git deploys, like Vercel (opens in a new tab), Netlify (opens in a new tab), or Render (opens in a new tab).

Storage

Triplit is designed to support any storage that can implement a key value store. You may specify the storage adapter you want to use by setting the storage option. Triplit provides some default configurations of our storage adapters, which you can use by setting the storage option to the appropriate string value for the adapter. These include:

• memory (default) - See memory-btree
• sqlite - An SQLite storage adapter, which requires the installation of the better-sqlite3 package (opens in a new tab)
• lmdb - An LMDB storage adapter, which requires the installation of the lmdb package (opens in a new tab)

Typically this will use your LOCAL_DATABASE_URL environment variable so you'll want to make sure that's set.

You can also pass in an instance of an adapter or a function that returns an instance of an adapter.

function createAdapter() {
  return new MyCustomAdapter();
}
 
const startServer = createServer({
  storage: createAdapter,
});

Health checks

The server exposes a health check endpoint at /healthcheck. This endpoint will return a 200 status code if the server is running and healthy.

Secrets

There are a few secrets that you need to provide to the server to enable certain features. If you are planning on using the Triplit Dashboard, you will need to set JWT_SECRET to the global Triplit public RSA key associated with your project. Read the Triplit Cloud guide for more information.

JWT_SECRET

The server uses JWT tokens to authenticate clients, and you need to provide a symmetric secret or public key to verify these tokens that it receives. The JWT_SECRET environment variable should be assigned to this validation secret. Triplit supports both symmetric (HS256) and asymmetric (RS256) encryption algorithms for JWTs. You will need to generate client tokens signed with the appropriate algorithm.

You can generate tokens with the jsonwebtoken package (e.g. if you wanted to use asymmetric encryption) :

import jwt from 'jsonwebtoken';
 
const anonKey = jwt.sign(
  {
    'x-triplit-token-type': 'anon',
  },
  process.env.PUBLIC_KEY,
  { algorithm: 'RS256' }
);
 
const serviceKey = jwt.sign(
  {
    'x-triplit-token-type': 'secret',
  },
  process.env.PUBLIC_KEY,
  { algorithm: 'RS256' }
);

For more complicated authentication schemes, refer to our authentication guide.

LOCAL_DATABASE_URL (required for durable storage)

An absolute path on the server's file system to a directory where the server will store any database files. This is required for durable storage options: lmdb, and sqlite.

EXTERNAL_JWT_SECRET (optional)

If you want your server to support JWTs signed by a second issuer, you can also set EXTERNAL_JWT_SECRET to that signing secret (or public key). For the server to recognize a JWT as "external", it must not have the x-triplit-token-type claim or if that claim is set, it must not have the value of anon or secret. Those specific combinations of claims are reserved for "internal" JWTs, e.g. the special anon and secret tokens.

CLAIMS_PATH (optional)

If you are using custom JWTs with nested Triplit-related claims, you can set the CLAIMS_PATH environment variable. The server will read the claims at the path specified by CLAIMS_PATH. Read the authentication guide for more information.

SENTRY_DSN (optional)

If you want to log errors to Sentry, you can set the SENTRY_DSN environment variable. The server will automatically log errors to Sentry.

VERBOSE_LOGS (optional)

If you want to log all incoming and outgoing messages and requests, you can set the VERBOSE_LOGS environment variable. This can be useful for debugging.