Hi,

I can't really reproduce... https://stackblitz.com/edit/github-zj7avh?file=src%2Fapp%2Fpage.tsx

I wonder if the problem is within your OS or Node.js version?

If you run node -e "console.log(process.arch)" does that match your OS architecture?

It does match my OS

Is it possible that this is just a cache issue?

Alright, could you try with a Node version in the 18 range? Say 18.18.0?

Clean up all of the node_modules and lock file too.

Here is what I did just now.

I still use the node version v20.11.0

Step I did try

# Clean up the cache
npm cache clean --force

# Delete yarn.lock
rm yarn.lock

# Run dev again
yarn dev

I did not see the TypeError anymore. For some reason it fixes the issue. I cannot determine what caused the issue though.

Here is what I did just now.

I still use the node version v20.11.0

Step I did try

# Clean up the cache
npm cache clean --force

# Delete yarn.lock
rm yarn.lock

# Run dev again
yarn dev

I did not see the TypeError anymore. For some reason it fixes the issue. I cannot determine what caused the issue though.

Though this steps did not fix my other project with the same setup.

Here is the other project that I am talking about.
https://github.com/dashawk/swertres-app

I have tried

I have already tried clearing out the node_modules and the lock files and did a new yarn install using node v18.17.0 but it is still the same thing happening.

facing the same issue during HMR and navigation when using npm run dev
doing ctrl+shift+r (hard reload) removes the error and renders the page correctly but again when a file is changed and HMR is triggered resulting in the same error.
using node lts : 20.11.1

ok when I killed npm run dev and reloaded the localhost the page was still rendered. From the network tab, I got to know the page was being fetched from a service worker so I unregistered it and the problem is gone

ok when I killed npm run dev and reloaded the localhost the page was still rendered. From the network tab, I got to know the page was being fetched from a service worker so I unregistered it and the problem is gone

Is there a way that we can disable this feature? It is probably like a caching issue.

I think the service worker gets registered when you do npm run start and open localhost:3000. So, for previewing the prod build run on a different port or open in incognito, guessing.

I am getting TypeError: Cannot read properties of undefined (reading 'call') after introducing a very basic middleware.
It works well on my local but the error appears after deploying to a preview env in Vercel.

[GET] [middleware: "src/middleware"] /exhibitions/{***}/{***} reason=EDGE_FUNCTION_INVOCATION_FAILED, status=500, user_error=true

TypeError: Cannot read properties of undefined (reading 'call')

It happens on Nextj14.
Current version: 14.2.2

My middleware:

import {NextResponse} from 'next/server'

export function middleware() {
  return NextResponse.next()
}

export const config = {
  matcher: '/exhibitions/:path+',
}

I'm facing the same error with the app router.

I'm calling a function from a client component, then redirecting to a route which is intercepted by a middleware.ts.
I have used many server actions in my application that work fine, but this one doesn't, and idk why.

in development mode, I get:

⨯ TypeError: __webpack_modules__[moduleId] is not a function
    at Object.__webpack_require__ [as require] (/.next/server/webpack-runtime.js:33:42)

and in production mode, I get:

TypeError: Cannot read properties of undefined (reading 'call')
    at Object.t [as require] (/home/huilensolis/dev/projects/Pictura/.next/server/webpack-runtime.js:1:143)
    at /home/huilensolis/dev/projects/Pictura/node_modules/.pnpm/[email protected][email protected][email protected][email protected]/node_modules/next/dist/compiled/next-server/app-page.runtime.prod.js:15:473
    at process.processTicksAndRejections (node:internal/process/task_queues:95:5)
    at async rc (/home/huilensolis/dev/projects/Pictura/node_modules/.pnpm/[email protected][email protected][email protected][email protected]/node_modules/next/dist/compiled/next-server/app-page.runtime.prod.js:14:5243)
    at async rj (/home/huilensolis/dev/projects/Pictura/node_modules/.pnpm/[email protected][email protected][email protected][email protected]/node_modules/next/dist/compiled/next-server/app-page.runtime.prod.js:15:22333)
    at async doRender (/home/huilensolis/dev/projects/Pictura/node_modules/.pnpm/[email protected][email protected][email protected][email protected]/node_modules/next/dist/server/base-server.js:1406:30)
    at async cacheEntry.responseCache.get.routeKind (/home/huilensolis/dev/projects/Pictura/node_modules/.pnpm/[email protected][email protected][email protected][email protected]/node_modules/next/dist/server/base-server.js:1567:28)
    at async NextNodeServer.renderToResponseWithComponentsImpl (/home/huilensolis/dev/projects/Pictura/node_modules/.pnpm/[email protected][email protected][email protected][email protected]/node_modules/next/dist/server/base-server.js:1475:28)
    at async NextNodeServer.renderPageComponent (/home/huilensolis/dev/projects/Pictura/node_modules/.pnpm/[email protected][email protected][email protected][email protected]/node_modules/next/dist/server/base-server.js:1852:24)
    at async NextNodeServer.renderToResponseImpl (/home/huilensolis/dev/projects/Pictura/node_modules/.pnpm/[email protected][email protected][email protected][email protected]/node_modules/next/dist/server/base-server.js:1890:32)

my node version:

> node -v
v18.18.0

my node architecture:

> node -e "console.log(process.arch)"
x64

my os architecture:

Linux 6.8.2-arch2- x86_64 GNU/Linux

We're facing the same issue in production. We are hosted on vercel. I believe it's building with Node version 20.11.0 and Detected Next.js version: 14.0.3.

This error is happening pretty often and slowing down our page load.

I attempted running the app in production with node version 18 and I'm still seeing the error.

Having this a lot in production as well, 14.2.3, node 20.

Cannot read properties of undefined (reading 'call')

Chrome 126.0.0
Windows >=10

Server action to login in a user, basic form that takes email and password

I'm having the same problem, nextjs 14.2.15, node 20.15.0

I'm having the same problem, nextjs 14.2.15 ( i tries all versions from 14.2.1 => 14.2.15), node v20.16 LTS IRON

facing the same issue during HMR and navigation when using npm run dev doing ctrl+shift+r (hard reload) removes the error and renders the page correctly but again when a file is changed and HMR is triggered resulting in the same error. using node lts : 20.11.1

this worked but i cant force my clients to do so , also NODE_ENV=production sort of helped (reduced frequency of occurrence smh)

Could this have something to do with version skew? I have a setup that allows me to deploy two separate versions at a time and pick which one I want based on a cookie to test what would happen for users after deploying a new version.

What I've noticed is if I am on one version of the app and then switch the cookie to the other version, then navigation seems to work fine, but server actions cause this error to occur.

I'm trying to track down if it's for all server actions or maybe specific to ones that redirect or revalidate or something.

For the record, this is on 14.2.4.

The component stack doesn't seem terribly useful as it's all minified code, but it looks something like:

at Lazy
at div
at div
at C (reference to line in minified chunk)
at c ... etc with a bunch of single letters, presumably custom components
...
at body
at html
at s ... more single letters
...

OK, I was able to work around this. When I detect that app versions are different and I get a request with accept: text/x-component I can force the response to be something else (e.g. text/plain). This tells the frontend that something is amiss and it should do an MPA navigation.

From what I understand, NextJS has built-in skew protection like this starting in 14.2.7: #67255

I'll have to play around with it because I don't really see it documented very well unless I'm missing something. But it's unclear to me if this would only work if the client is already on at least 14.2.7, or if the server responds with something that can trigger older versions to treat it as an MPA (like the text/plain mentioned above). It's also unclear if you need to set a NEXT_DEPLOYMENT_ID variable at runtime or build time, or if you need to set it as an option in your next.config.js (Based on the name I suspect a runtime variable would make the most sense). Most of the docs I'm finding are specific to Vercel's Skew Protection, which looks pretty similar.

Still getting this in production, assuming it's due to version skew? but as @redbmk suggested, nextjs has something built in? it definitely seems to be around when new versions get deployed, but im not sure how to fix this if next.js skew protection isnt working for me

using "next": "14.2.16" + react 18

I had a library in my package.json and was also dynamically loading the same library but with a different version, which was causing the issue. https://vercel.com/docs/deployments/skew-protection

The solution was to sync the versions of the npm package and the dynamically loaded script.

I'm on deno and I seem to reliably get this server-side when I update any module. Then doing a reset and a git clean persists the issue. No idea what causes it, and my solution have been going back and forth between versions for an hour or two until something works again.

I had a hard time getting the built-in skew protection to work. It seems like NEXT_DEPLOYMENT_ID needs to be set at build time, rather than deployment time, which made it harder for me to use because it acted as a cache bust. Maybe not a big deal when making changes to the app, but for any cicd changes or other unrelated changes in a monorepo, it meant the docker build could never be cached.

I tried making the deployment ID something like XXX_DEPLOYMENT_ID_XXX and replacing all instances of it in the code before starting the app but I must have missed something because that didn't seem to catch it everywhere. I ended up putting an nginx service in front of nextjs just to use the text/plain hack I mentioned above, which works pre-14.2.7 as well, but ideally I don't need another server in front of Next.

It seems like middleware would be a decent place to force MPA instead of RSC based on a cookie (e.g. app_version or deploy_id) but I wasn't able to get this to work. I might take another stab at it here soon, but if anyone has any ideas I'm all ears. Would be great to get rid of nginx if this was the only thing I needed it for

It happened a few times locally while recording network requests and starting/stopping the service multiple times for me

I am having the same error but instead with 'exec',

TypeError: Cannot read properties of undefined (reading 'exec')

does anybody know a fix for this?

Having same issue with NextJS `` "next": "15.2.4"```