Node LRA

Example "Long Running Action" client implementation for Node.js. Inspired by Micropofile LRA specification.

This repository aims to implement a compatible Node LRA.js Client (@node-lra/core), while also providing integration libraries for various Node frameworks like express (@node-lra/express) and hopefully NestJS (@node/lra-nestjs).

This is done mainly as an excercise and personal interest in the technology, but it might actually become useful one day.

Install

The package manager of choice is pnpm, so if you don't have it, install it first

npm i -g pnpm

Install dependencies

pnpm install

The express example

Preparation

Execute the express example by runing

pnpm nx serve express-sample

For sucessful running of the examples, you need a running "LRA Coordinator" on port 8080. The coordinator must also see your running apps. You can use docker or podman in the network host mode to achieve that:

podman run -it --rm --name lra-coordinator --network host  docker.io/jbosstm/lra-coordinator

Verify that the coordinator container can see your running apps with:

podman exec lra-coordinator curl http://localhost:3333/ping

You shoud get pong back.

Running the example

The express example app consists of 2 services, running on ports 3333 and 3334 respectively.

To run the example, send a GET request to http://localhost:3333/start and watch the console where the sample app is running

curl -X GET http://localhost:3333/start

Service A starts a LRA, then calls Service B, which joins the LRA.

Now, 3 random events could happen:

1. Service B returns successfully, in which case Service A closes the LRA and the Coordinator calls the complete endpoint of both services

2. Service B throws error, and in turn cancels the LRA. The Coordinator calls the compensate endpoint of both services

3. Service B simulates "Bad Request" error and leaves the LRA. This causes Service A to cancel the LRA. The Coordinator only calls the compensate endpoint of Service A