
Recently I was working on a nodejs/express application to troubleshoot a memory leak. Those were the steps and tools I followed to identify the root cause.

## 1. Reproduce the issue

We had it in two modules, one a http server and the other a sqs consumer.

#### For the http server

Created a [k6](https://k6.io/) script for a small performance test:

```javascript
// script.js
import http from 'k6/http';
import { check, sleep } from 'k6';

export default function () {
  const url = 'http://localhost:3000/test';
  const res = http.get(url);
  check(res, { success: (r) => r.status === 200 });
}
```

And ran it with:

```bash
# 100 users, 1 hour
k6 run -u 100 -d 1h script.js
```

#### For the SQS consumer

Built a send command (note: I am using [localstack](https://localstack.cloud/)):

```bash
# send.sh
AWS_REGION: "us-east-1" aws --endpoint-url=http://localhost:4566
    sqs send-message
    --region "us-east-1"
    --queue-url http://localhost:4566/000000000000/queue
    --message-body '{"my": "message"}'
```

and ran with with watch:

```bash
watch -n .1 ./send.sh
```

## 2. Next step, collect the data

Initially I was using node itself for generating it, e.g:

```typescript
import v8 from 'node:v8';
// ...
app.get('/heapdump', (req, res) => {
  const fileName = v8.writeHeapSnapshot();
  res.send({ fileName });
});
```

but we can also use chrome devtools for this (I am using [brave](https://brave.com/), but sames applies):

- run the app with the flag `--inspect`
- go to `chrome://inspect` / `brave://inspect`, connect
- go to `Memory` tab, click on `Take Heap Snapshot`

## 3. Analyze the data

That is the tricky part. Pick the initial snapshot, run the test, pick the final snapshot, and compare.

Try to correlate what you have there with the code. `Allocation on timeline` was also helpful for me.

I strongly recommend take a look on [this presentation](https://www.youtube.com/watch?v=hliOMEQRqf8) for detailed explanation and showcase:

<iframe width="560" height="315" src="https://www.youtube.com/embed/hliOMEQRqf8?si=iQ3kfqgyt6Aa_TZU" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen></iframe>

## Next?

Now is the time to apply a fix and re-run the tests.

We found a lot of [DerivedLogger](https://github.com/winstonjs/winston/blob/c69cdb0cec15a138e0b6e374501e027d1c39606c/lib/winston/create-logger.js#L36) in the memory. It was because the code was calling `winston.createLogger` on each call to enrich the log. We switched to [child](https://github.com/winstonjs/winston#creating-child-loggers) call and problem solved.

There was no need to use [clinic.js](https://clinicjs.org/) this time.



Some steps for finding/troubleshooting it

Memory Leak in nodejs/javascript


I just tried the new cloudflare worker. To create a sample and deploy there, you just need to run one single command:

```
pnpm create cloudflare@latest <project name>
```

... and of course, a lot of prompts.

I removed some dead code (editor config, prettier), added [biome](https://biomejs.dev/), added a [taskfile](https://taskfile.dev/) (I will forget the commands). Also played with [Hono](https://github.com/honojs/hono) - the web framework [suggested by Cloudflare](https://www.youtube.com/watch?v=H7Qe96fqg1M) and added an AI endpoint.

Pretty simple, reminds me the good old days with the free [Heroku](https://www.heroku.com/) tier.

For sure I will consider it for my pet projects.

My full code is available on [github](https://github.com/adamatti/learn-cf-worker).

Next, I will try it with [remix](https://remix.run/) for a full stack experience. If I got it right, it should be [super simple](https://hono.dev/examples/with-remix) (or [creating from scratch](https://developers.cloudflare.com/pages/framework-guides/deploy-a-remix-site/)).


Cloudflare workers


Hey, are you using [yarn](https://classic.yarnpkg.com)? Great tool from [Facebook](https://facebook.com), right?

But how are you installing it?

`npm install -g yarn`?

### I have a bad news to tell you...

> Yarn has not been distributed on npm since 2019 because of breaking changes that could impact deployments - Google AI

and

> Those versions entered maintenance mode in January 2020 - [yarn classic website](https://classic.yarnpkg.com/lang/en/docs/install)

### So, what should I do?

Need to consider a better/newer package manager.

It could be:

- [the new yarn version](https://yarnpkg.com/): as of today (09/06), it is on version 4.4.1
- [PNPM](https://pnpm.io/), the faster alternative
- [The NPM itself](https://www.npmjs.com/). The official tool. In 2024, it is faster than yarn classic

There are a lot of benchmarks out there, like:

- https://romanglushach.medium.com/comparing-npm-yarn-and-pnpm-package-managers-which-one-is-right-for-your-distributed-project-to-4d7de2f0db8e
- https://refine.dev/blog/pnpm-vs-npm-and-yarn
- https://www.dhiwise.com/post/pnpm-vs-npm-vs-yarn-which-javascript-package-manager

### How about me?

I am using [PNPM](https://pnpm.io/) in my latest projects, both pet and professional, but already had some issues with it.

It all case by case, but I strongly recommend to use [PNPM](https://pnpm.io/) for fast local builds and CI.


Memory Leak in nodejs/javascript

Cloudflare workers

Yarn Classic