Cannot find module 'next/dist/server/next-server.js'

[realamirhe]

Hi I'm using vercel to for my mono repo (nx) next app. it is totally okay in my local machine.

Both dev and build scripts are okay.

Sometime the next-gh page pipeline fails with the following error after the hole build process finished

Done in 2.27s.
--
18:10:55.241 | Error! Cannot find module 'next/dist/server/next-server.js'
18:10:55.242 | Require stack:
18:10:55.242 | - /home/runner/work/ahimi/ahimi/apps/multiverse/noop.js

full log

There are two thing I can't understand:

• I don't have a noop.js file at all
• changing the next version will solve the issue and pass the pipeline again.

NOTE: The redeploy system on vercel with or without cache doesn't work in fixing this issue.
Isn't the .next directory regenerated when Redeploy with existing Build Cache. is off?

Then I come to second issue, the installation of new next version take so much time on M1 chips, it is quite fast in linux distributions but take nearly 20 min for me to update each new version.

[thlmenezes]

@realamirhe same error using Next.js v12.2.1, which version you reverted to in "changing the next version will solve the issue and pass the pipeline again." ?

Edit: upgrading to v12.2.5 fixed this for me

[realamirhe]

For me it happed in the following version, errors got fixed after upgrade

12.2.3 -> 12.2.4-canary.12

Are you (@thlmenezes) using anything beside next, I guess the problem might about the build cache artifacts?

First errors happened when I changed git history
And the second occurrence issue got solved when I removed nx cloud build cache layer.

[thlmenezes]

Well, I was deploying a monorepo as well, but using turbo.

The dependency list is quite large...

But bumping the version did the trick.

If you are curious, this was the tree at that time repo and the package I was trying to deploy was apps/client

[realamirhe]

I first assumed that altering any package in package.json would result in a change in the docker layer and address the issue, however it's something more internal and tied to the next version itself 😅.

Next relies on several os-dependent packages for their SWC compiler. I believe the issue arises because the internal caching service for these monorepo providers is not mature enough for these os-dependent npm packages.

@next/swc-android-arm64 @next/swc-darwin-arm64 @next/swc-darwin-x64 @next/swc-freebsd-x64 @next/swc-linux-arm-gnueabihf @next/swc-linux-arm64-gnu @next/swc-linux-arm64-musl @next/swc-linux-x64-gnu @next/swc-linux-x64-musl @next/swc-win32-arm64-msvc @next/swc-win32-ia32-msvc @next/swc-win32-x64-msvc

[andrewdoro]

I have the same error with next 12.3.0, deploying on Vercel with Turborepo.

[andrew-hu368]

Same error with Next 12.3.0 with Turborepo

[azzlack]

Same error when creating a dockerfile manually with Turbo and pnpm

[andrew-hu368]

We managed to solve this one by copying the node_modules in the monorepo and also the ones in the standalone output.

Something along the lines

COPY ./node_modules ./
COPY ./apps/my-next-app/.next/standalone ./apps/my-next-app/

[azzlack]

Yeah we are looking at a similar approach. It increases the size of the runtime image by a lot, but it can work as a temporary solution.
FYI: Upgrading to 12.3.1 does not give different results.

[alexbrazier]

For me this was caused by yarn workspaces hoisting the node_modules to the root folder, but the nextjs standalone option requires the next modules to be in the frontend workspace folder. The fix for yarn is to change your root package.json file to something like this (depending on your workspaces):

  "workspaces": {
    "nohoist": [
      "**/frontend",
      "**/frontend/**"
    ],
    "packages": [
      "api",
      "frontend"
    ]
  },

This tells yarn to install the frontend/next modules in the frontend folder directly.

[antspy]

Hi! Do you remember where should the node_modules folder be installed exactly? Inside the /frontend directory, /frontend/app, somwehre else?

[alfonsobravi]

I am experiencing the same with 12.3.1 and an NX 14.8.4 monorepo.
It's absolutely random and unpredictable, afaik.
Curious to hear if someone has found a fix for this combination too..

[TurtIeSocks]

Seeing this with Next.js 12.3.1 and Turbo 1.5.6.

[bogdanned]

Same here with Next.js 13.0.7 and Turbo 1.6.3

[MarcinCebula]

Doing redeploy from the deploy page without the cache solves the issue for me. It works on the second deploy of the failing project, but it get's annoying having to manually redeploy after every git push. I think it has something to do with building the cache locally

[AryanJ-NYC]

Doing redeploy from the deploy page without the cache solves the issue for me. It works on the second deploy of the failing project, but it get's annoying having to manually redeploy after every git push. I think it has something to do with building the cache locally

I can confirm this also works for me.

[azzlack]

I solved it by doing some "fancy" copying and hoisting of the node_modules folder.
This is using pnpm and turbo

Here is my Dockerfile

# Builder image
FROM docker.io/node:16-alpine AS build

RUN apk add --no-cache curl bash git

ARG PROJECT_NAME

WORKDIR /app

# Set up pnpm
RUN curl -f https://get.pnpm.io/v6.16.js | node - add --global pnpm && \
    pnpm config set store-dir .pnpm-store
COPY pnpm-lock.yaml .npmrc* ./
RUN pnpm fetch

# Build
COPY . .
RUN pnpm install --frozen-lockfile --offline --ignore-scripts --workspace-root --filter ${PROJECT_NAME}...
RUN pnpm run build --filter=${PROJECT_NAME}...
# WORKAROUND FOR: https://github.com/vercel/next.js/discussions/39432
RUN pnpm install --prod --frozen-lockfile --offline --shamefully-hoist --ignore-scripts --workspace-root --filter ${PROJECT_NAME} && \
    cp -Lr ./node_modules ./node_modules_temp && \
    rm -rf ./node_modules_temp/.cache && \
    rm -rf ./node_modules_temp/.pnpm
# END WORKAROUND

# Runtime image
FROM docker.io/node:16-alpine AS release

ARG PROJECT_NAME

ENV NODE_ENV=production
ENV PORT=3000

WORKDIR /app
RUN addgroup --system --gid 1001 nodejs
RUN adduser --system --uid 1001 nextjs

COPY --from=build --chown=nextjs:nodejs /app/apps/${PROJECT_NAME}/.next/standalone/apps/${PROJECT_NAME} ./
COPY --from=build --chown=nextjs:nodejs /app/apps/${PROJECT_NAME}/.next/static* ./.next/static
COPY --from=build --chown=nextjs:nodejs /app/apps/${PROJECT_NAME}/public* ./public

# WORKAROUND FOR: https://github.com/vercel/next.js/discussions/39432
RUN rm -rf ./node_modules
COPY --from=build /app/node_modules_temp ./node_modules
# END WORKAROUND

USER nextjs

EXPOSE 3000

CMD ["node", "server.js"]

[lesderid]

I'm seeing this issue with next 12.3.1 using yarn berry (3.2.2) in PnP mode.

[rjzheng]

Using next@12.2.6 and turbo@1.6.1. Getting the same error when deploying app to Vercel, fixed by redeploying in Vercel. I think it may have something to do with caching.

[matthewdfleming]

Not sure if this is 100% the same issue, but I'm including this here cause it was pretty infuriating for me to sort this out.

So for my builds the issues was actually a docker COPY context problem. So what was happening is that (on my machine) I would run a next build inside my app (a yarn workspace) and it would do a full build and create the .next folder inside the workspace. Then I'd try to run a docker build at the root of the monorepo (which like the ones above essentially does a COPY . . kind of deal) and I'd get the error during the nextjs build:

[Error: ENOENT: no such file or directory, copyfile '/.yarn/berry/cache/next-npm-12.3.2-ac8bcf5092-9.zip/node_modules/next/dist/server/next-server.js' -> '/app/apps/web-gateway-application/.yarn/berry/cache/next-npm-12.3.2-ac8bcf5092-9.zip/node_modules/next/dist/server/next-server.js'] 

To fix this I needed to make sure that docker would not use that .next directory during its builds. To do this, at the same location as my Dockerfile I added a Dockerfile.dockerignore file that specified the .next folder within my 'app'. E.g. /apps/web-gateway-application/.next.

Then my builds started to get consistent.

This is the spec I was using for the ignore that allows a separate dockerignore file in each workspace.

[damjtoh]

Downgrading tubro to 1.6.2 fixed it for me.

[sahil242927-github]

Downgrading tubro to 1.6.2 isn't wroking either!

Solution:
Update the following files with the code

apps\web\next.config.js
const path = require("path");
module.exports = {
reactStrictMode: true,
output: "standalone",
experimental: {
outputFileTracingRoot: path.join(__dirname, "../../"),
transpilePackages: ["ui"],
},
};

turbo.json
{
"$schema": "https://turbo.build/schema.json",
"pipeline": {
"build": {
"outputs": ["dist/", ".next/", "public/dist/"],
"dependsOn": ["^build"],
"env": ["NEXT_PUBLIC_API_HOST"]
},
"test": {
"outputs": ["coverage/"],
"dependsOn": []
},
"lint": {
"dependsOn": ["^build"],
"outputs": []
},
"dev": {
"cache": false
},
"clean": {
"cache": false
}
}
}

[agrippa1994]

Downgrading tubro to 1.6.2 isn't wroking either!

Solution: Update the following files with the code

apps\web\next.config.js const path = require("path"); module.exports = { reactStrictMode: true, output: "standalone", experimental: { outputFileTracingRoot: path.join(__dirname, "../../"), transpilePackages: ["ui"], }, };

turbo.json { "$schema": "https://turbo.build/schema.json", "pipeline": { "build": { "outputs": ["dist/", ".next/", "public/dist/"], "dependsOn": ["^build"], "env": ["NEXT_PUBLIC_API_HOST"] }, "test": { "outputs": ["coverage/"], "dependsOn": [] }, "lint": { "dependsOn": ["^build"], "outputs": [] }, "dev": { "cache": false }, "clean": { "cache": false } } }

Thanks, adding the outputFileTracingRoot property to my next.config.js configuration has worked for me!

module.exports = {
  // ...
  experimental: {
    outputFileTracingRoot: path.join(__dirname, "../../"),
  },
  // ...
};

[languageleague]

Same issue with next 13.0.3-canary.4, doing the redeployment without cache solves it.

[moodya]

For me this issue was because the .dockerignore file contained the following

**/.next

which resulted in the standalone/.next folder not being copied to my docker image
resulting in the following error when running the container

Error: Cannot find module 'next/dist/server/next-server' Require stack: /app/server.js at Function.Module._resolveFilename (node:internal/modules/cjs/loader:985:15) at Function.Module._load (node:internal/modules/cjs/loader:833:27) at Module.require (node:internal/modules/cjs/loader:1057:19) at require (node:internal/modules/cjs/helpers:103:18) at Object.<anonymous> (/app/server.js:4:20) at Module._compile (node:internal/modules/cjs/loader:1155:14) at Object.Module._extensions..js (node:internal/modules/cjs/loader:1209:10) at Module.load (node:internal/modules/cjs/loader:1033:32) at Function.Module._load (node:internal/modules/cjs/loader:868:12) at Function.executeUserEntryPoint [as runMain] (node:internal/modules/run_main:81:12) { code: 'MODULE_NOT_FOUND', requireStack: [ '/app/server.js' ]

[Howard86]

I am experiencing the same issue with the following when setting up lerna nx monorepo on vercel, caching build commands with nx-cloud

    "next": "^12.3.1",
    "lerna": "^6.1.0",
    "nx": "^15.2.4"
    "@nrwl/nx-cloud": "^15.0.2",

In the end I found this on next doc and applied following on next.config.js and it works!

  experimental: {
    outputFileTracingRoot: path.join(__dirname, '../..'),
  },

[shawnmclean]

well this works sometimes.

[lukaszkalemba]

having the same issue with turborepo 1.6.3 and next 12.1.5, locally build works fine but after deploying to AWS Amplify it throws this error when I enter the page

[Leejjon]

Hi all,

I hope to have some additional info for those that run into this error.

We talked to our aws enterprise support and they came up with this working example based on the "outputFileTracingRoot" solution:
https://github.com/waleedwasfy-96/my-turborepo

Which you can fork and import via the AWS Management console (Amplify -> Host new App -> Connect to GitHub -> Select Branch -> add "apps/docs" in the monorepo repo field).

They also said that the amplify team has acknowledged this problem and is looking into fixing it, but couldn't give an ETA.

We use the CDK, so I wrote a project with a CDK project that will create an infra as code pipeline that creates a cloudformation stack with the amplify app in it (configured to use Next.js 12+).
https://github.com/Leejjon/amplify-nextjs13

See the readme.md for instructions on how to run it.

[mwood23]

Here's a horrible hack, but it could help if you're desperate. The problem for me was node_modules was not copied to every .func folder when doing ISR on vercel. I'm in a NX monorepo. Added this to my next.config.js file:

output: 'standalone',
experimental: {
  outputFileTracingRoot: path.join(__dirname, '../../'),
},

Then I run vercel build and after it completes I recursively copy the standalone's node_modules output to all of the .func folders so they have everything they need to render.

// shimVercelNodeModules.js
const fs = require('fs')
const path = require('path')

function collectFuncDirectories(dir, funcDirectories = []) {
  fs.readdirSync(dir).forEach((file) => {
    const filePath = path.join(dir, file)
    if (fs.statSync(filePath).isDirectory() && file.endsWith('.func')) {
      funcDirectories.push(filePath)
    } else if (fs.statSync(filePath).isDirectory()) {
      collectFuncDirectories(filePath, funcDirectories)
    }
  })

  return funcDirectories
}

const run = () => {
  const foldersToAddNodeModulesTo = collectFuncDirectories(
    path.join(__dirname, '.vercel', 'output', 'functions'),
  )

  console.log(
    `Copying node modules for ${foldersToAddNodeModulesTo.length} folders...`,
  )

  foldersToAddNodeModulesTo.forEach((x) => {
    const writePath = path.join(x, 'node_modules')
    fs.cpSync(
      path.join(__dirname, '.next', 'standalone', 'node_modules'),
      writePath,
      {
        recursive: true,
      },
      (err) => {
        console.log(err)
      },
    )
  })

  console.log(`Completed.`)
}

run()

[matthewdfleming]

To extend on my previous comment. We use docker images for our production runtime environments. We're also using a monorepo based upon yarn workspaces. Yarn berry (> 2.x), uses a .yarnrc.yml file for its configuration. We ended up using slightly separate .yarnrc.yml files for the docker builds vs our local desktop development. For the docker builds we had this configuration:

nodeLinker: node-modules
nmMode: hardlinks-local
httpTimeout: 300000
enableInlineBuilds: true
enableGlobalCache: false

Then for our next config we use:

output: 'standalone',
....
  experimental: {
    // @link https://nextjs.org/docs/advanced-features/output-file-tracing#caveats
    // outputFileTracingRoot: undefined, // ,path.join(__dirname, '../../'),
    outputFileTracingRoot: path.join(__dirname, '../../'),

    // Experimental monorepo support
    // @link {https://github.com/vercel/next.js/pull/22867|Original PR}
    // @link {https://github.com/vercel/next.js/discussions/26420|Discussion}
    externalDir: true
  }

The nextjs compiler web pack then correctly outputs all of the dependencies correctly into the output build.. e.g.
.next/standalone/node_modules. So we were able to get the PnP linker to work on our local workstations but some of our dependencies don't work well in that mode (when running the standalone). We were able to use pnpify in addition to the PnP linker but even that was a bit of a pain (need to remember to source the pnp.cjs file for instance). So we ended up using the filesystem based linker and allow the standalone configuration compilation to really output a standalone running server.

So our multi-stage docker build looks like this:

FROM node:16-alpine as builder
# Check https://github.com/nodejs/docker-node/tree/b4117f9333da4138b03a546ec926ef50a31506c3#nodealpine to understand why libc6-compat might be needed.
RUN sed -i 's/https/http/g' /etc/apk/repositories
RUN apk add --no-cache --allow-untrusted libc6-compat
RUN apk update

WORKDIR /app

ENV NEXT_TELEMETRY_DISABLED 1
ARG NPM_NETWORK_TIMEOUT=300000
ARG APP_TO_BUILD

# copy dependencies from the source
COPY package.json .
COPY yarn.lock .
COPY .yarnrc.docker.yml .yarnrc.yml
COPY .yarn/releases ./.yarn/releases
COPY .yarn/patches ./.yarn/patches
COPY .yarn/patches ./.yarn/plugins

# copy the app code over
COPY packages ./packages
COPY apps/$APP_TO_BUILD  ./apps/$APP_TO_BUILD

ENV NODE_ENV production
ENV NEXT_TELEMETRY_DISABLED 1
ENV NODE_TLS_REJECT_UNAUTHORIZED=0

# install dependencies
RUN yarn workspaces focus $APP_TO_BUILD

# build the web gateway application app
RUN yarn workspace $APP_TO_BUILD build

FROM node:16-alpine  AS runner
# Check https://github.com/nodejs/docker-node/tree/b4117f9333da4138b03a546ec926ef50a31506c3#nodealpine to understand why libc6-compat might be needed.
RUN sed -i 's/https/http/g' /etc/apk/repositories
RUN apk add --no-cache --allow-untrusted libc6-compat
RUN apk update

ARG NPM_NETWORK_TIMEOUT=300000
ARG APP_TO_BUILD
WORKDIR /app

ENV NODE_ENV production
ENV NEXT_TELEMETRY_DISABLED 1
ENV NODE_TLS_REJECT_UNAUTHORIZED=0

RUN addgroup --system --gid 1001 nodejs
RUN adduser --system --uid 1001 nextjs

COPY --from=builder --chown=nextjs:nodejs /app/apps/$APP_TO_BUILD/.next/standalone .
COPY --from=builder --chown=nextjs:nodejs /app/apps/$APP_TO_BUILD/.next/static ./apps/$APP_TO_BUILD/.next/static
COPY --from=builder --chown=nextjs:nodejs /app/apps/$APP_TO_BUILD/public ./apps/$APP_TO_BUILD/public
COPY --from=builder --chown=nextjs:nodejs /app/apps/$APP_TO_BUILD/env ./
COPY --from=builder --chown=nextjs:nodejs /app/apps/$APP_TO_BUILD/docker/healthCheck.js ./

USER nextjs

EXPOSE 3000

ENV PORT 3000

HEALTHCHECK --interval=45s --timeout=5s \
    CMD node healthCheck.js

COPY --from=builder --chown=nextjs:nodejs /app/apps/$APP_TO_BUILD/docker/docker-entrypoint.sh ./docker-entrypoint.sh
RUN chmod +x docker-entrypoint.sh
ENTRYPOINT [ "/app/docker-entrypoint.sh" ]

[ijjk]

Hi, we have identified an issue with turbo/nx cache and have deployed a fix for this, please redeploy your project and give it a try!

[ijjk]

@sethtjf if you're seeing this in the latest version it is unrelated to the initial issue, please open a new one with a reproduction to allow investigating further!

[rixrix]

is there any solution for this ? one of our project, after adding transpilePackages then this error started popping out

[savager]

Any fix for PNPM? using a pnpm workspace and getting this issue.

[MrDockal]

I'm still facing the same problem with turborepo + vercel CLI. An issue here #48949 @ijjk

[jakeleventhal]

#39432 (comment) this is the most succinct reproduction i've found for this issue. turbo is not needed/relevant

[Yagogc]

I'm having the same issue too. My setup is the following:

• Next 13 (SSR everything except two SSG pages)
• Yarn 1.22 w/ workspaces (it's a monorepo)
• AWS Amplify w/ NEXTJS - SSR framework.

2023-02-16T14:52:40.429+00:00 | Error: Cannot find module 'next/dist/server/next-server'
-- | --
  | 2023-02-16T14:52:40.429+00:00 | Require stack:
  | 2023-02-16T14:52:40.429+00:00 | - /tmp/app/server.js
  | 2023-02-16T14:52:40.429+00:00 | at Function.Module._resolveFilename (node:internal/modules/cjs/loader:933:15)
  | 2023-02-16T14:52:40.429+00:00 | at Function.Module._load (node:internal/modules/cjs/loader:778:27)
  | 2023-02-16T14:52:40.429+00:00 | at Module.require (node:internal/modules/cjs/loader:1005:19)
  | 2023-02-16T14:52:40.429+00:00 | at require (node:internal/modules/cjs/helpers:102:18)
  | 2023-02-16T14:52:40.429+00:00 | at Object.<anonymous> (/tmp/app/server.js:2:20)
  | 2023-02-16T14:52:40.429+00:00 | at Module._compile (node:internal/modules/cjs/loader:1105:14)
  | 2023-02-16T14:52:40.429+00:00 | at Object.Module._extensions..js (node:internal/modules/cjs/loader:1159:10)
  | 2023-02-16T14:52:40.429+00:00 | at Module.load (node:internal/modules/cjs/loader:981:32)
  | 2023-02-16T14:52:40.429+00:00 | at Function.Module._load (node:internal/modules/cjs/loader:822:12)
  | 2023-02-16T14:52:40.429+00:00 | at Function.executeUserEntryPoint [as runMain] (node:internal/modules/run_main:77:12) {
  | 2023-02-16T14:52:40.429+00:00 | code: 'MODULE_NOT_FOUND',
  | 2023-02-16T14:52:40.429+00:00 | requireStack: [ '/tmp/app/server.js' ]
  | 2023-02-16T14:52:40.429+00:00 | }

I've trying almost everything this thread has mentioned so far:

Tested different versions of Next13 included canary versions too.

nohoisting deps for the specific workspace

output: 'standalone;

outputTracingFile

Modifying the amplify.yml build file as mentioned here

NOTE: I also tried almost every combination of them.

If anyone can suggest a different fix I'll greatly appreciate it, I'm walking in circles for too long 😑

[iambryancs]

@Yagogc was having the same issue using nx monorepo with setting output: 'standalone'.

What I ended up doing was add the following to my next.config.js:

  output: 'standalone',
  // https://nextjs.org/docs/advanced-features/output-file-tracing#caveats
  experimental: {
    // this includes files from the monorepo base two directories up
    outputFileTracingRoot: path.join(__dirname, '../../'),
  },

Which is basically the same setup that @matthewdfleming described which I noticed too late already after going through a couple of tests and got it working.

[ijjk]

Hi, this should no longer be needed in the latest version of Next.js per #39432 (reply in thread) please update and give it a try!

[jakeleventhal]

experiencing this on 13.4.3. No errors during build or local dev. only errors for api functions when deployed to vercel

[azzlack]

I have the same issue still, running 13.2.4
This is when building and running locally, not via Vercel

node:internal/modules/cjs/loader:1024
  throw err;
  ^

Error: Cannot find module 'next/dist/server/next-server'
Require stack:
- /app/server.js
    at Function.Module._resolveFilename (node:internal/modules/cjs/loader:1021:15)
    at Function.Module._load (node:internal/modules/cjs/loader:866:27)
    at Module.require (node:internal/modules/cjs/loader:1093:19)
    at require (node:internal/modules/cjs/helpers:108:18)
    at Object.<anonymous> (/app/server.js:2:20)
    at Module._compile (node:internal/modules/cjs/loader:1191:14)
    at Object.Module._extensions..js (node:internal/modules/cjs/loader:1245:10)
    at Module.load (node:internal/modules/cjs/loader:1069:32)
    at Function.Module._load (node:internal/modules/cjs/loader:904:12)
    at Function.executeUserEntryPoint [as runMain] (node:internal/modules/run_main:81:12) {
  code: 'MODULE_NOT_FOUND',
  requireStack: [ '/app/server.js' ]
}

[trippygif]

@azzlack can you check your .dockerignore to see if you included **/node_modules? try removing that and see if that works for you :)

[azzlack]

Sure, I can try that when im back at work tomorrow. Im not sure if it would solve anything though, as the install and build happens inside the docker file in my case

[Zolomon]

I still have the same issue on 13.4.5.

[jakeleventhal]

same

[trappar]

I'm seeing this issue when using PNPM + a prebuilt deployment from a monorepo. Here's an extremely minimal reproduction repo for my case. See the readme for instructions on how to reproduce it: https://github.com/trappar/next-vercel-ssr-bug

Maybe I'm doing something wrong to cause the issue here?

[dsebastien]

Same here with pnpm and Nrwl NX

[trappar]

Related issue: vercel/vercel#8794

[trippygif]

I removed **/node_modules/ from docker ignore and it worked for me! I would also recommend adding in

  // https://nextjs.org/docs/advanced-features/output-file-tracing#caveats
  experimental: {
    // this includes files from the monorepo base two directories up
    outputFileTracingRoot: path.join(__dirname, '../../'),
  },

to the next.config.js file.

I am using the turbo prune --docker build command in my docker file to containerize a turbo repo app. My assumption is that the next package is getting mounted inside of the monorepo package rather than the root so you would want to ensure your dockerfile doesn't ignore the relative node modules for your next app.

[jakeleventhal]

This issue is resolved for me after talking with @TooTallNate

Fix for me was the following:

1. Update vercel to 30.2.2
2. Set each client application rootDirectory to the real root directory (apps/appname/client)
3. Link repo in root via vercel link --repo
4. Build as normal (running build commands from the app folder (apps/appname/client)

[cherealnice]

Ran into this issue when dockerizing a nextjs app inside of a yarn workspace. The solution suggested above, copying the node_modules to the build fixed it for me.

[savager]

For the record, doesn't copying the node_modules to the build defeat the entire purpose of a standalone output? The goal was to reduce size by excluding node_modules. I understand this is just to avoid the error but the issue still exists and I look forward to the day when it's properly solved.

[cherealnice]

Not completely, but it does defeat the node_modules part of standalone. https://nextjs.org/docs/app/api-reference/next-config-js/output#automatically-copying-traced-files

[cherealnice]

I found a better fix above, using outputFileTracingRoot.

[savager]

Yes, agreed, besides the public and static directories. I did try the outputFileTracingRoot with no luck, but perhaps I'll give it another go. I an using a pnpm moonorepo so my mileage may vary. Appreciate your time.

[hadelyar-nex]

I've always been using the outputFileTracingRoot, still get the same issue.

Copying the node_modules from the root works for me, but ends up as a large docker image at 1.3gb compared to the standalone folder which is less than 300mb.

My setup is a monorepo with pnpm.

[alex-aircall]

vercel/vercel#8794 (comment)

We have been facing this issue recently. We found that with the following config it works.

In the root folder, under .vercel/project.json

{
  "projectId": "xyz",
  "orgId": "abc",
  "settings": {
    "createdAt": 1693232512727,
    "framework": "nextjs",
    "devCommand": null,
    "installCommand": null,
    "buildCommand": null,
    "outputDirectory": ".next",
    "rootDirectory": "apps/[app_name]",
    "directoryListing": false,
    "nodeVersion": "18.x"
  }
}

On the Vercel Settings, update:

• under root directory, set apps/[app_name]
• override .next for the output directory

Then, simply run:

• pnpm vercel build
• pnpm vercel deploy --prebuilt

[darcy-vitacca]

const path = require("path");

/** @type {import('next').NextConfig} */
const nextConfig = {
reactStrictMode: true,
// transpilePackages: ["ui"],
output: "standalone",
experimental: {
outputFileTracingRoot: path.join(__dirname, "../../"),
},
};

adding this to next config worked don't forget to bring in path

[sgup]

I'm seeing this issue with next 14.2.2. No turborepo. Downgrading to 14.1.0 fixed the issue for me.

Using pnpm to install with these deps:

    "@apollo/client": "3.9.11",
    "@apollo/experimental-nextjs-app-support": "0.10.0",
    "next": "14.1.0",
    "react": "18.2.0",
    "react-dom": "18.2.0",

[antspy]

Would it be possible to add some additional debugging, or explanation of what the vercel cli is looking for when building?

Otherwise debugging / fixing errors becomes impossible. I have a monorepo setup with bazel. Things run locally, and the .next folder is created, but the vercel build command still complains with Error: Cannot find module 'next/dist/server/next-server.js'.

I tried various solutions suggested in this thread, but nothing seems to work. The problem is that this is a black box, and it's not clear what is running, in which directory, where is it looking for files, etc.

Would love to have additional logging!