Issue description

Using relationLoadStrategy set to "query" may result in a hang-up.

Expected Behavior

When calling the find method with relationLoadStrategy set to "query" on tables with 1:N relationships, the process completes successfully.

Actual Behavior

When multiple find method calls are made simultaneously with this setting, the system fails to complete the process and experiences a hang-up.

Steps to reproduce

Entities

We have tables with 1:N relationships.

import { Entity, PrimaryGeneratedColumn, Column, OneToMany } from "typeorm"
import { Photo } from "./Photo"

@Entity()
export class User {
    @PrimaryGeneratedColumn()
    id: number

    @Column()
    name: string

    @OneToMany(() => Photo, (photo) => photo.user, { eager: false })
    photos: Photo[]
}

import { Entity, PrimaryGeneratedColumn, Column, ManyToOne } from "typeorm"
import { User } from "./User"

@Entity()
export class Photo {
    @PrimaryGeneratedColumn()
    id: number

    @Column()
    url: string

    @ManyToOne(() => User, (user) => user.photos)
    user: User
}

Bussiness Code

In our business code, we make simultaneous calls using the find method that includes relationships, setting the relationLoadStrategy to 'query'.

import { AppDataSource } from "../typeorm/data-source"
import { User } from "../typeorm/entity/User"

const main = async () => {
    await AppDataSource.initialize()
    const findUserWithPhoto = async () => {
        const repository = AppDataSource.getRepository(User)
        return await repository.findOne({
            where: { id: 1 },
            relations: {
                photos: true
            },
            relationLoadStrategy: "query"
        })
    }

    const result = []
    for (let i = 0; i < 10; i++) {
        result.push(findUserWithPhoto())
    }
    await Promise.all(result)
    await AppDataSource.destroy()
}

main()

For reproduction purposes, we are making multiple calls using a for loop. However, the same issue arises when server-side endpoints are rapidly and simultaneously accessed by clients

My Environment

Additional Context

probably the cause

In TypeORM, the obtainQueryRunner() is used to acquire a connection. However, the issue here is that this.queryRunner is set with a value only in certain cases, which leads to the system consistently evaluating the right-hand expression of obtainQueryRunner.

In the connection pool, if no connections are available, the system waits until a new connection is released. If obtainQueryRunner is called multiple times within the queryRunner, there is a potential for a deadlock to occur between the release of a connection and the acquisition of a new one. This deadlock can cause the system to become unresponsive or hang.

reproduction code

https://github.com/saw81wt/typeorm-hung-up/tree/master

Relevant Database Driver(s)

• aurora-mysql
• aurora-postgres
• better-sqlite3
• cockroachdb
• cordova
• expo
• mongodb
• mysql
• nativescript
• oracle
• postgres
• react-native
• sap
• spanner
• sqlite
• sqlite-abstract
• sqljs
• sqlserver

Are you willing to resolve this issue by submitting a Pull Request?

No, I don’t have the time, but I can support (using donations) development.