Skip to content
Cloudflare Docs

Fly

Connect Hyperdrive to a Fly Postgres database instance.

This example shows you how to connect Hyperdrive to a Fly Postgres database instance.

1. Allow Hyperdrive access

You can connect Hyperdrive to any existing Fly database by:

  1. Allocating a public IP address to your Fly database instance
  2. Configuring an external service
  3. Deploying the configuration
  4. Obtain the connection string, which is used to connect the database to Hyperdrive.

  1. Run the following command to allocate a public IP address ↗.

    fly ips allocate-v6 --app <pg-app-name>

  2. Configure an external service ↗ by modifying the contents of your fly.toml file. Run the following command to download the fly.toml file.

    fly config save --app <pg-app-name>

    Then, replace the services and services.ports section of the file with the following toml snippet:

    [[services]]
      internal_port = 5432 # Postgres instance
      protocol = "tcp"
    

    [[services.ports]]
      handlers = ["pg_tls"]
      port = 5432

  3. Deploy the new configuration ↗.

  4. Obtain the connection string ↗, which is in the form of:

    postgres://{username}:{password}@{public-hostname}:{port}/{database}?options

2. Create a database configuration

To configure Hyperdrive, you will need:

  • The IP address (or hostname) and port of your database.
  • The database username (for example, hyperdrive-demo) you configured in a previous step.
  • The password associated with that username.
  • The name of the database you want Hyperdrive to connect to. For example, postgres.

Hyperdrive accepts the combination of these parameters in the common connection string format used by database drivers:

postgres://USERNAME:PASSWORD@HOSTNAME_OR_IP_ADDRESS:PORT/database_name

Most database providers will provide a connection string you can directly copy-and-paste directly into Hyperdrive.

To create a Hyperdrive configuration with the Wrangler CLI, open your terminal and run the following command. Replace <NAME_OF_HYPERDRIVE_CONFIG> with a name for your Hyperdrive configuration and paste the connection string provided from your database host, or replace user, password, HOSTNAME_OR_IP_ADDRESS, port, and database_name placeholders with those specific to your database:

Terminal window
npx wrangler hyperdrive create <NAME_OF_HYPERDRIVE_CONFIG> --connection-string="postgres://user:password@HOSTNAME_OR_IP_ADDRESS:PORT/database_name"

This command outputs a binding for the Wrangler configuration file:

{
  "name": "hyperdrive-example",
  "main": "src/index.ts",
  "compatibility_date": "2024-08-21",
  "compatibility_flags": [
    "nodejs_compat"
  ],
  "hyperdrive": [
    {
      "binding": "HYPERDRIVE",
      "id": "<ID OF THE CREATED HYPERDRIVE CONFIGURATION>"
    }
  ]
}

3. Use Hyperdrive from your Worker

Install Postgres.js ↗:

Terminal window
npm i postgres@>3.4.5

Add the required Node.js compatibility flags and Hyperdrive binding to your wrangler.jsonc file:

{
  "compatibility_flags": [
    "nodejs_compat"
  ],
  "compatibility_date": "2024-09-23",
  "hyperdrive": [
    {
      "binding": "HYPERDRIVE",
      "id": "<your-hyperdrive-id-here>"
    }
  ]
}

Create a Worker that connects to your PostgreSQL database via Hyperdrive:

// filepath: src/index.ts
import postgres from "postgres";


export default {
  async fetch(
    request: Request,
    env: Env,
    ctx: ExecutionContext,
  ): Promise<Response> {
    // Create a database client that connects to your database via Hyperdrive
    // using the Hyperdrive credentials
    const sql = postgres(env.HYPERDRIVE.connectionString, {
      // Limit the connections for the Worker request to 5 due to Workers' limits on concurrent external connections
      max: 5,
      // If you are not using array types in your Postgres schema, disable `fetch_types` to avoid an additional round-trip (unnecessary latency)
      fetch_types: false,
    });


    try {
      // A very simple test query
      const result = await sql`select * from pg_tables`;


      // Clean up the client, ensuring we don't kill the worker before that is
      // completed.
      ctx.waitUntil(sql.end());


      // Return result rows as JSON
      return Response.json({ success: true, result: result });
    } catch (e: any) {
      console.error("Database error:", e.message);


      return Response.error();
    }
  },
} satisfies ExportedHandler<Env>;

Next steps