HTTP server: Scaling across CPU cores

A single process handles requests on one core. The ende serve subcommand can run a whole pool of workers behind one port with the --parallel flag. The server exports its handler instead of calling Ende.serve.

Edit on GitHub

With ende serve, the module’s default export provides the fetch handler. The runtime owns the listening socket, which is what lets it share the port between workers.

export default {
  fetch(_req: Request): Response {
    return new Response("Hello, World!");
  }
} satisfies Ende.ServeDefaultExport;

Start one worker per CPU core: ende serve --parallel server.ts ende serve: Listening on http://0.0.0.0:8000/ with 8 threads Incoming connections are distributed across the workers. Set the ENDE_JOBS environment variable to control the worker count: ENDE_JOBS=4 ende serve --parallel server.ts

Each worker is a separate isolate: module-level state like counters or caches is per worker, not shared. Keep shared state in something external, like a database.

Run this example locally using the Ende CLI:

ende serve --parallel https://docs.ende.sh/examples/scripts/http_server_parallel.ts

HTTP server: Scaling across CPU cores

Additional resources