Injectable instances can have references to Request objects and can hold it forewer in request scope

[breeeew]

Is there an existing issue for this?

• I have searched the existing issues

Current behavior

When nest handle any request on first time - it gets metadata for controller's handler and stores it in cache forever.
When that controller utilizes the request object, that injected by REQUEST token - it will have reference to request object in scope and will hold it forewer, because of cache.

Minimum reproduction code

https://github.com/breeeew/nestjs-leak-reproduction

FIX

Steps to reproduce

1. npm i
2. npm run start:debug
3. attach debugger and dump heap - search for AppService in debuggers console - there will be just prototype object
4. collect garbage
5. navigate in browser to /
6. collect garbage
7. dump heap again
8. search for AppService again - there will be second entry and it will be real instance

Expected behavior

All unused objects must be collected by gc

Package

• I don't know. Or some 3rd-party package
• @nestjs/common
• @nestjs/core
• @nestjs/microservices
• @nestjs/platform-express
• @nestjs/platform-fastify
• @nestjs/platform-socket.io
• @nestjs/platform-ws
• @nestjs/testing
• @nestjs/websockets
• Other (see below)

Other package

No response

NestJS version

latest

Packages versions

[System Information]
OS Version     : Linux 6.5.0-26-generic
NodeJS Version : v18.19.1
NPM Version    : 10.2.4 

[Nest CLI]
Nest CLI Version : 10.3.2 

[Nest Platform Information]
platform-express version : 10.3.5
schematics version       : 10.1.1
testing version          : 10.3.5
common version           : 10.3.5
core version             : 10.3.5
cli version              : 10.3.2

Node.js version

18, 20

In which operating systems have you tested?

• macOS
• Windows
• Linux

Other

Our case: sometimes user requests some handler in request scope, that downloads a huge json - and that json have been parsed and holds a huge amount of memory. And it also slows down garbage collectiong process (while collecting it increases cpu usage and memory usage by nodejs)