Connection lost: The server closed the connection with Bun/Drizzle/MySQL/DockerSwarm

[rnfrafael]

I keep getting this error often, sometimes occurs many times, sometimes not...
These are some logs

Happens in any query, so is not related to the query itself, I tried many config in the connection, this is the connection code.

import { drizzle } from "drizzle-orm/mysql2";
import mysql from "mysql2/promise";
import { env } from "../env";
import { schemas } from "./schemas";
const connection = mysql.createPool({
  uri: env.DATABASE_URL,
});

export const db = drizzle(connection, { schema: schemas, mode: "default", logger: env.DATABASE_LOG ?? false });

I always used pool connection, didn't try the direct connection.

So, is this a bug? I think probably is something about the Docker environment, maybe? This is my docker stack

version: '3.8'

volumes:
  db-api-db:
  mysqlconf:
networks:
  api-db:
  proxy:
    external: true

services:
  api-db:
    image: api-db:1.6.9
    # command: ["sleep", "infinity"]
    logging:
      driver: loki
      options:
        loki-url: "http://localhost:3100/loki/api/v1/push"
    environment:
      RUN_MIGRATIONS: 'true'
      DATABASE_URL: mysql://root:password@mysql:3306/api-db
      MP_CLIENT_ID: id
      MP_CLIENT_SECRET: secret
      MP_REDIRECT_URI: redirecturi
      DATABASE_LOG: "false"
      DATABASE_ROOT_PASSWORD: password
      DATABASE_NAME: api-db
      DATABASE_USER: root
      DATABASE_PASSWORD: password
      DATABASE_PORT: 3306
      DATABASE_HOST: mysql
    depends_on:
      - mysql
    networks:
      - api-db
      - proxy
    deploy:
      mode: replicated
      replicas: 1
      placement:
        constraints:
          - node.role == manager
      labels:
        - traefik.enable=true
        - traefik.http.routers.api-db.rule=Host(`dns`)
        - traefik.http.routers.api-db.entrypoints=websecure
        - traefik.http.routers.api-db.tls=true
        - traefik.http.routers.api-db.tls.certresolver=leresolver
        - traefik.http.services.api-db.loadbalancer.server.port=3000
        - traefik.http.routers.api-db.service=api-db
        # - traefik.http.services.api-db.loadbalancer.healthcheck.path=/health
        # - traefik.http.services.api-db.loadbalancer.healthcheck.interval=30s

  mysql:
    image: mysql
    volumes:
      - db-api-db:/var/lib/mysql
      - mysqlconf:/etc/mysql
    networks:
      - api-db
    environment:
      MYSQL_ROOT_PASSWORD: password
      MYSQL_DATABASE: api-db

  phpmyadmin:
    image: phpmyadmin
    networks:
      - api-db
    ports:
      - 9001:80
    depends_on:
      - mysql
    environment:
      PMA_HOST: mysql

Can this be related to Drizzle? Maybe Bun?

Thanks for any help.

Environment

bun: v1.1.27
mysql2: v3.11.3

Edit:
I realized that I only tested using Bun, I decided to test it with Node now

Edit2: Tested with Node, got "same error"

2024-10-15 13:53:54.250 | {"level":50,"time":1729011234249,"pid":89,"hostname":"d8316eea0257","err":{"type":"Error","message":"read ECONNRESET","stack":"Error: read ECONNRESET\n    at PromisePool.query (/app/node_modules/mysql2/promise.js:356:22)\n    at MySql2PreparedQuery.execute (/app/node_modules/src/mysql2/session.ts:121:31)\n    at MySqlSelectBase.execute (/app/node_modules/src/mysql-core/query-builders/select.ts:961:25)\n    at MySqlSelectBase.then (/app/node_modules/src/query-promise.ts:31:15)\n    at process.processTicksAndRejections (node:internal/process/task_queues:95:5)","code":"ECONNRESET","errno":-104},"msg":"read ECONNRESET"}

[wellwelwel]

Hey, @rnfrafael 🙋🏻‍♂️

Could you try to check it without Drizzle?

[rnfrafael]

Yes, I can try, but it will take a while, I need to update the code and leave it running in production for a "real" test. This problem takes some time to happen, I haven't been able to find a logic yet, but that's it, I'll try without Drizzle.

[rnfrafael]

@wellwelwel I didn't make the change, I found this #2599 and I did make this change #2599 (comment)
maxIdle: 0 and did work, for two days without any errors, It may not be the best approach but works :).