Hi, I can see that the value of your REDIS_URL in your medusa-server on production doesn't match the value of the REDIS_URL of your medusa-redis service?

Can you delete and reference the variable again?

Hello, Thank you very much for the quick reply!

I double checked and they seem to be the same from my side. Just to be sure, I changed them to be references: REDIS_URL="${{medusa-redis.REDIS_PUBLIC_URL}}"

However, the problem persists.

Seems medusa `db:migrate` might not be calling any redis url at all. Please can you check your code. How are you handling/calling the redis url env in that command?

You pointed me in the right direction, there was an issue with the config file indeed, but it seems that the "full pool" issue isn't related and the code was working earlier even without redis. Now I get this error:

> medusa-starter-default@0.0.1 predeploy

> medusa db:migrate

Knex: Timeout acquiring a connection. The pool is probably full. Are you missing a .transacting(trx) call?

npm warn config production Use --omit=dev instead.

Here is the current medusa-config.ts file:

import { defineConfig, loadEnv } from "@medusajs/framework/utils"

loadEnv(process.env.NODE_ENV || "development", process.cwd())

module.exports = defineConfig({

  projectConfig: {

    databaseUrl: process.env.DATABASE_URL,

    redisUrl: process.env.REDIS_URL,

    databaseDriverOptions: {

      connection: {

        ssl: {

          rejectUnauthorized: false,

        },

      },

    },

    http: {

      storeCors: process.env.STORE_CORS!,

      adminCors: process.env.ADMIN_CORS!,

      authCors: process.env.AUTH_CORS!,

      jwtSecret: process.env.JWT_SECRET || "supersecret",

      cookieSecret: process.env.COOKIE_SECRET || "supersecret",

    },

    workerMode: process.env.MEDUSA_WORKER_MODE as

      | "shared"

      | "worker"

      | "server",

  },

  modules: [

    {

      resolve: "@medusajs/medusa/cache-redis",

      options: {

        redisUrl: process.env.REDIS_URL,

      },

    },

    {

      resolve: "@medusajs/medusa/event-bus-redis",

      options: {

        redisUrl: process.env.REDIS_URL,

      },

    },

    {

      resolve: "@medusajs/medusa/workflow-engine-redis",

      options: {

        redis: {

          url: process.env.REDIS_URL,

        },

      },

    },

  ],

})

I recently updated the dependencies of medusa to v2.5.0 and mikro-orm to v6.4.3. Here's the package.json/

{

  "name": "medusa-starter-default",

  "version": "0.0.1",

  "description": "A starter for Medusa projects.",

  "author": "Medusa (https://medusajs.com)",

  "license": "MIT",

  "keywords": [

    "sqlite",

    "postgres",

    "typescript",

    "ecommerce",

    "headless",

    "medusa"

  ],

  "scripts": {

    "seed": "medusa exec ./src/scripts/seed.ts",

    "build": "medusa build",

    "predeploy": "medusa db:migrate",

    "migrations": "medusa db:migrate",

    "start": "medusa start",

    "dev": "medusa develop",

    "test:integration:http": "TEST_TYPE=integration:http NODE_OPTIONS=--experimental-vm-modules jest --silent=false --runInBand --forceExit",

    "test:integration:modules": "TEST_TYPE=integration:modules NODE_OPTIONS=--experimental-vm-modules jest --silent --runInBand --forceExit",

    "test:unit": "TEST_TYPE=unit NODE_OPTIONS=--experimental-vm-modules jest --silent --runInBand --forceExit"

  },

  "dependencies": {

    "@medusajs/admin-sdk": "^2.5.0",

    "@medusajs/cli": "^2.5.0",

    "@medusajs/framework": "^2.5.0",

    "@medusajs/medusa": "^2.5.0",

    "@mikro-orm/core": "6.4.3",

    "@mikro-orm/knex": "6.4.3",

    "@mikro-orm/migrations": "6.4.3",

    "@mikro-orm/postgresql": "6.4.3",

    "awilix": "^8.0.1",

    "pg": "^8.13.0"

  },

  "devDependencies": {

    "@medusajs/test-utils": "latest",

    "@mikro-orm/cli": "6.4.3",

    "@swc/core": "1.5.7",

    "@swc/jest": "^0.2.36",

    "@types/jest": "^29.5.13",

    "@types/node": "^20.0.0",

    "@types/react": "^18.3.2",

    "@types/react-dom": "^18.2.25",

    "jest": "^29.7.0",

    "prop-types": "^15.8.1",

    "react": "^18.2.0",

    "react-dom": "^18.2.0",

    "ts-node": "^10.9.2",

    "typescript": "^5.6.2",

    "vite": "^5.2.11"

  },

  "engines": {

    "node": ">=20"

  },

  "packageManager": "yarn@4.5.3+sha512.3003a14012e2987072d244c720506549c1aab73ee728208f1b2580a9fd67b92d61ba6b08fe93f6dce68fd771e3af1e59a0afa28dd242dd0940d73b95fedd4e90"

}

Another information, everything starts correctly in my staging environment, which is connected to the same branch:

MODULE: file

Skipped. Database is up-to-date for module.

MODULE: file

Skipped. Database is up-to-date for module.

MODULE: file

Skipped. Database is up-to-date for module.

MODULE: fulfillment

Skipped. Database is up-to-date for module.

MODULE: fulfillment

Skipped. Database is up-to-date for module.

MODULE: fulfillment

Skipped. Database is up-to-date for module.

MODULE: notification

Skipped. Database is up-to-date for module.

MODULE: notification

Skipped. Database is up-to-date for module.

MODULE: notification

Skipped. Database is up-to-date for module.

Migrations completed

Syncing links...

Database already up-to-date

No link to load from /app/.medusa/server/src/links. skipped.

Connection to Redis in module 'cache-redis' established

Connection to Redis in module 'event-bus-redis' established

Connection to Redis in module 'workflow-engine-redis' established

Connection to Redis PubSub in module 'workflow-engine-redis' established

Locking module: Using "in-memory" as default.

No pending migration scripts to execute

npm warn config production Use --omit=dev instead.

> medusa-starter-default@0.0.1 start

> medusa start

Skipping instrumentation registration. No register function found.

No link to load from /app/.medusa/server/src/links. skipped.

Connection to Redis in module 'cache-redis' established

Connection to Redis in module 'event-bus-redis' established

Connection to Redis in module 'workflow-engine-redis' established

Connection to Redis PubSub in module 'workflow-engine-redis' established

Locking module: Using "in-memory" as default.

Creating server

Server is ready on http://localhost:9000

Can you duplicate it into another environment to see if it works properly as well?

Sure, I just tried and it doesn't

Please try this https://docs.railway.com/reference/errors/enotfound-redis-railway-internal#ioredis.

I'm not entirely sure how to implement your suggestion. Medusa’s built-in Redis integrations expect a Redis Url string, not an instance of ioredis. So I simply added "?family=0" at the end of the REDIS_PUBLIC_URL, but it didn't help unfortunately.

Maybe you can point me to a reproducible repo example where I can check this later.

There you go: https://github.com/tevarg/mekanika-medusa-railway

Thank you again for the help!

I can see your DATABASE_URL is set to the redis url. Why is that?

Oh wow, many apologies for this long conversation that ended up with such a stupid error... I guess a bad copy paste happened at some point and I really didn't see it! The db url was correct in staging, which explains why it was working with the same code but not in two different environments. Thank you for your patience and your help

You are welcome