#Can't access admin after update to 2.3.1

34 messages · Page 1 of 1 (latest)

sudden inlet
#

Hi guys. After update to 2.3.1 I can't access to my admin. Error:

Blocked request. This host ("myulr.top") is not allowed. To allow this host, add "myulr.top" to server.allowedHosts in vite.config.js.

But my URL is added to AUTH_CORS and ADMIN_CORS in .env and it's used in medusa-config.ts

twilit summit
#

i got same issue

#

somehow 2.3.1 can not run medusa build too

sudden inlet
#

My app has been running for about a month with these settings and I haven't changed anything. Just performed an update via npm update.

Also, when starting the server, warnings appeared in the console:

The CJS build of Vite's Node API is deprecated. See https://vite.dev/guide/troubleshooting.html#vite-cjs-node-api-deprecated for more details. (node:78976) [DEP0040] DeprecationWarning: The punycodemodule is deprecated. Please use a userland alternative instead. (Usenode --trace-deprecation ... to show where the warning was created)

leaden dove
twilit summit
twilit summit
#

@sudden inlet @leaden dove i think i solved it, we must start medusa server in .medusa/server instead root folder

#

it's worked now

dawn vapor
#

same issue with my test deployment of version 2.3.1
also tried modifying medusa-config.ts to add some additional vite config to no luck

    vite: config => {
      if (isDevelopment) {
        return {
          ...config,
          server: {
            hmr: {
              overlay: true,
            },
            base: "/app/",
            allowedHosts: process.env.MEDUSA_BACKEND_URL,
          },
        }
      }
      return {
        ...config,
        server: {
          hmr: false,
          base: "/app/",
          allowedHosts: process.env.MEDUSA_BACKEND_URL,
        }
      }
    }

server is already starting from the .medusa/server build folder.

twilit summit
#

@dawn vapor somehow it's worked for me,

cp .env .medusa/server
cd .medusa/server && yarn && yarn start
#

NODE_ENV=production

dawn vapor
amber thorn
#

I'm also running into this on a new install. Does anyone have any advice on getting things working with Docker? I have set MEDUSA_BACKEND_URL in medusa-config.ts.

# Current LTS Node
FROM node:22
WORKDIR /app
# Node dependencies
COPY crm/package.json .
COPY crm/yarn.lock .
RUN yarn install
# Copy remaining content of crm to /app
COPY crm/ .
# Next two lines are attempt to workaround for "Blocked request" error
# https://github.com/medusajs/medusa/issues/11067
RUN yarn build
RUN cd .medusa/server
CMD ["yarn", "start"]
dawn vapor
fast summit
#

Hey guys I am trying to dig into this, but just need to point out that you always need to run medusa start from the .medusa/server folder in V2. It's explained here: https://docs.medusajs.com/resources/medusa-cli/commands/build#run-built-medusa-application

I don't think this is related to above issue, but the production build will never work unless you run the project from there.

build Command - Medusa CLI Reference - Medusa Development Resources

Explore Medusa's recipes, API references, configurations, storefront guides, and more.

dawn vapor
fast summit
dawn vapor
fast summit
amber thorn
#

I was able to fix this in my environment (Docker + Google Cloud Run). The final factor seemed to be setting NODE_ENV

# WORKING
# Current LTS Node
FROM node:22
WORKDIR /app
# Node dependencies
COPY crm/package.json .
COPY crm/yarn.lock .
RUN yarn install
# Copy remaining content of crm to /app
COPY crm/ .
# https://docs.medusajs.com/resources/medusa-cli/commands/build#run-built-medusa-application
RUN yarn build
WORKDIR /app/.medusa/server
RUN yarn install
ENV NODE_ENV=production
CMD ["yarn", "start"]

# BROKEN
# Current LTS Node
FROM node:22
WORKDIR /app
# Node dependencies
COPY crm/package.json .
COPY crm/yarn.lock .
RUN yarn install
# Copy remaining content of crm to /app
COPY crm/ .
# https://docs.medusajs.com/resources/medusa-cli/commands/build#run-built-medusa-application
RUN yarn build
WORKDIR /app/.medusa/server
RUN yarn
CMD ["yarn", "start"]
#

Oh, spoke too soon. The login page now loads but login requests go to localhost(!)

image.png
#

Makes me think backendUrl isn't being obeyed.

#

Yeh if I change localhost to my real URL using curl then I get a token.

fast summit
amber thorn
#

it's being set in the hosting environment via MEDUSA_BACKEND_URL

# 
module.exports = defineConfig({
  projectConfig: {
    databaseUrl: process.env.DATABASE_URL,
    databaseSchema: process.env.DATABASE_SCHEMA ||
      "medusa",
    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",
    }
  },
  admin: {
    backendUrl: process.env.MEDUSA_BACKEND_URL ||
      "http://localhost:9000"
  },
})
#

maybe my nesting is wrong?

#

hm no, seems to match docs

fast summit
#

Nesting looks right, could you try and console.log process.env and check that the env is actually set?

amber thorn
#

that would take a little setting up but it's almost certainly OK based on the GCP console

image.png
fast summit
amber thorn
#

That's very helpful, thank you. I've removed that section of config and it works.

fast summit
amber thorn
#

Yes it does seem odd. As far as I know, an env var is an env var, and Cloud Run doesn't do anything special. I use them in other projects without issue.