tiger-server

tiger-server

Tiger server is a very lightweight server for very simple process like webhooks.

Usage

npm install tiger-server --save

and create server.ts:

import { defineServer, http, cron, queue } from "tiger-server";
import type { QueueModule, CronModule, HttpModule } from "tiger-server";

export default defineServer(async (tiger) => {
  await tiger.use(http, cron, queue);

  await tiger.define<QueueModule<{message: string}>>({
    id: "hello",
    target: "queue:hello",
    async process(_state, message) {
      this.log(`Message received: ${JSON.stringify(message)}`);
    }
  });

  await tiger.define<CronModule<{count: number}>>({
    id: "cron",
    target: "cron:*/5 * * * * *",
    async process({ count = 0 }) {
      count++;
      await this.notify("queue:hello", { count });
      return { count }
    }
  });

  await tiger.define<HttpModule>({
    id: "request",
    target: "http:/hello",
    async process(_state, { req, res }) {
      await this.notify("queue:hello", { message: "request recieved" });
      res.send("success!")
    }
  });
});

Run it with Node.js 22.6+:

npx tiger-server run server.ts

This relies on Node’s native TypeScript loader, so there is no build step and the code runs directly.

The http plugin defaults to 0.0.0.0:9527. Override http.port/http.host (or TIGER_HTTP_PORT/TIGER_HTTP_HOST) to run multiple Tiger instances side-by-side while sharing the same cron queue.

The Monitor UI follows monitor.host:monitor.port (defaults 0.0.0.0:9753). Tweak monitor config or TIGER_MONITOR_* env vars—or disable it entirely with monitor.disabled/TIGER_MONITOR_DISABLED=1.

The queue plugin is now an in-memory message bus. It delivers queue: (or legacy zmq:) notifications to modules defined in the same process without any external socket or runtime-specific dependency.

The cron plugin persists its schedule either in LevelDB (cron.levelDbPath, default .tiger-cron) or, when you run distributed mode with a Postgres driver, inside the shared Postgres database. Multiple Tiger processes simply read from the same store and pop due runs cooperatively.

To build distributed modules, configure the distributed block. Set distributed.driver to postgres (and point DATABASE_URL to your Postgres instance) to enable a shared job/state store, or leave it as level for single-node experimentation. Every distributed module must declare an id and distributed: true; Tiger will enqueue work, persist state, and heartbeat the node registry through the configured persistence provider. Use distributed.maxQueueLength (default 100) to keep runaway producers from filling the queue indefinitely.

When running with the Postgres driver, apply the bundled Sequelize migrations (e.g. DATABASE_URL=postgres://... npx sequelize-cli db:migrate) before starting Tiger so the job, state, and registry tables exist.

When distributed mode is enabled, a management dashboard and API are available at /tiger/manage on the same port as the monitor. It lists every node’s last heartbeat, whether it is enabled/disabled, and lets you pause or resume queue consumption for any node while it continues to report heartbeats and finish in-flight work.

Logo is generated from Wikipedia, the original script is under GPL license.