Increased build times after update

[MarcJose]

To Reproduce

After upgrading our app dependencies for next.js from v15.3.5 to v15.4.2 our build times increased by a factor of 10 suddenly:

Packages:

next
@next/bundle-analyzer
eslint-config-next

Build Time Before (v15.3.5):

> next build

   ▲ Next.js 15.3.5
   - Environments: .env.local

   Creating an optimized production build ...
 ✓ Compiled successfully in 72s

Build Time After (v15.4.2):

> next build
   ▲ Next.js 15.4.2
   - Environments: .env.local
   - Experiments (use with caution):
     ✓ taint
     · serverActions
     · optimizePackageImports

   Creating an optimized production build ...
 ✓ Compiled successfully in 7.8min

Any idea what could cause this? Building went from 72 seconds to almost 8 minutes :(
Downgrading results in faster build times again.

Current vs. Expected behavior

Build times should be mostly similar

Provide environment information

Operating System:
  Platform: linux
  Arch: x64
  Version: #1 SMP PREEMPT_DYNAMIC Thu, 17 Jul 2025 21:05:29 +0000
  Available memory (MB): 31923
  Available CPU cores: 4
Binaries:
  Node: 22.17.1
  npm: 10.9.2
  Yarn: N/A
  pnpm: 8.15.3
Relevant Packages:
  next: 15.4.2 // Latest available version is detected (15.4.2).
  eslint-config-next: 15.4.2
  react: 19.1.0
  react-dom: 19.1.0
  typescript: 5.8.3
Next.js Config:
  output: N/A

Which area(s) are affected?

Turbopack

Which stage(s) are affected?

next build (local)

Additional context

We tried deleting the typical folders which can cause problems after an update: node_modules, .next and .swc but no luck so far.
Disabling the experimental options taint, serverActions and optimizePackageImports did not help either.

[timneutkens]

Can you share the .next/trace file after running each? Then we can see exactly where time was spent.

When you upgrade the webpack cache is invalidated, did it consistently reproduce or only the very first build after upgrading?

[MarcJose]

trace_v15.4.2.json
trace_v15.3.5.json

Here are both traces, one with the old version and one with the new version.
The increased build time remains even after running a second build without any changes unfortunately.

[travistruett]

I think I'm probably running into the same thing except I'm hitting OOM before a build is able to finish on a Github worker. As soon as I downgraded back to 15.3.4 everything worked again no problem.

#17 0.339 > next build
#17 0.339 
#17 1.193 [@sentry/nextjs] ACTION REQUIRED: To instrument navigations, the Sentry SDK requires you to export an `onRouterTransitionStart` hook from your `instrumentation-client.(js|ts)` file. You can do so by adding `export const onRouterTransitionStart = Sentry.captureRouterTransitionStart;` to the file.
#17 178.1 
#17 178.1 <--- Last few GCs --->
#17 178.1 
#17 178.1 [18:0xc68b000]   175651 ms: Scavenge 2021.6 (2031.1) -> 2013.9 (2031.1) MB, pooled: 0 MB, 0.70 / 0.00 ms  (average mu = 0.766, current mu = 0.615) allocation failure; 
#17 178.1 [18:0xc68b000]   177788 ms: Mark-Compact 2614.8 (2624.2) -> 2211.9 (2229.3) MB, pooled: 0 MB, 1780.11 / 0.00 ms  (average mu = 0.633, current mu = 0.449) allocation failure; scavenge might not succeed
#17 178.1 
#17 178.1 
#17 178.1 <--- JS stacktrace --->
#17 178.1 
#17 178.1 FATAL ERROR: Reached heap limit Allocation failed - JavaScript heap out of memory
#17 178.1 ----- Native stack trace -----
#17 178.1 
#17 178.1  1: 0xe98a91 node::OOMErrorHandler(char const*, v8::OOMDetails const&) [node]
#17 178.1  2: 0x12a7710 v8::Utils::ReportOOMFailure(v8::internal::Isolate*, char const*, v8::OOMDetails const&) [node]
#17 178.1  3: 0x12a7aac v8::internal::V8::FatalProcessOutOfMemory(v8::internal::Isolate*, char const*, v8::OOMDetails const&) [node]
#17 178.1  4: 0x150a9e5  [node]
#17 178.2  5: 0x151de12 v8::internal::Heap::CollectGarbage(v8::internal::AllocationSpace, v8::internal::GarbageCollectionReason, v8::GCCallbackFlags) [node]
#17 178.2  6: 0x14f6912 v8::internal::HeapAllocator::AllocateRawWithLightRetrySlowPath(int, v8::internal::AllocationType, v8::internal::AllocationOrigin, v8::internal::AllocationAlignment) [node]
#17 178.2  7: 0x14f6c48 v8::internal::HeapAllocator::AllocateRawWithRetryOrFailSlowPath(int, v8::internal::AllocationType, v8::internal::AllocationOrigin, v8::internal::AllocationAlignment) [node]
#17 178.2  8: 0x14cded0 v8::internal::Factory::NewFillerObject(int, v8::internal::AllocationAlignment, v8::internal::AllocationType, v8::internal::AllocationOrigin) [node]
#17 178.2  9: 0x194fefb v8::internal::Runtime_AllocateInYoungGeneration(int, unsigned long*, v8::internal::Isolate*) [node]
#17 178.2 10: 0x7f683de6f276 
#17 230.1 Aborted (core dumped)
#17 230.1  ELIFECYCLE  Command failed with exit code 134.
#17 ERROR: process "/bin/sh -c pnpm run build" did not complete successfully: exit code: 134

[MarcJose]

I think I'm probably running into the same thing except I'm hitting OOM before a build is able to finish on a Github worker. As soon as I downgraded back to 15.3.4 everything worked again no problem.

#17 0.339 > next build
#17 0.339 
#17 1.193 [@sentry/nextjs] ACTION REQUIRED: To instrument navigations, the Sentry SDK requires you to export an `onRouterTransitionStart` hook from your `instrumentation-client.(js|ts)` file. You can do so by adding `export const onRouterTransitionStart = Sentry.captureRouterTransitionStart;` to the file.
#17 178.1 
#17 178.1 <--- Last few GCs --->
#17 178.1 
#17 178.1 [18:0xc68b000]   175651 ms: Scavenge 2021.6 (2031.1) -> 2013.9 (2031.1) MB, pooled: 0 MB, 0.70 / 0.00 ms  (average mu = 0.766, current mu = 0.615) allocation failure; 
#17 178.1 [18:0xc68b000]   177788 ms: Mark-Compact 2614.8 (2624.2) -> 2211.9 (2229.3) MB, pooled: 0 MB, 1780.11 / 0.00 ms  (average mu = 0.633, current mu = 0.449) allocation failure; scavenge might not succeed
#17 178.1 
#17 178.1 
#17 178.1 <--- JS stacktrace --->
#17 178.1 
#17 178.1 FATAL ERROR: Reached heap limit Allocation failed - JavaScript heap out of memory
#17 178.1 ----- Native stack trace -----
#17 178.1 
#17 178.1  1: 0xe98a91 node::OOMErrorHandler(char const*, v8::OOMDetails const&) [node]
#17 178.1  2: 0x12a7710 v8::Utils::ReportOOMFailure(v8::internal::Isolate*, char const*, v8::OOMDetails const&) [node]
#17 178.1  3: 0x12a7aac v8::internal::V8::FatalProcessOutOfMemory(v8::internal::Isolate*, char const*, v8::OOMDetails const&) [node]
#17 178.1  4: 0x150a9e5  [node]
#17 178.2  5: 0x151de12 v8::internal::Heap::CollectGarbage(v8::internal::AllocationSpace, v8::internal::GarbageCollectionReason, v8::GCCallbackFlags) [node]
#17 178.2  6: 0x14f6912 v8::internal::HeapAllocator::AllocateRawWithLightRetrySlowPath(int, v8::internal::AllocationType, v8::internal::AllocationOrigin, v8::internal::AllocationAlignment) [node]
#17 178.2  7: 0x14f6c48 v8::internal::HeapAllocator::AllocateRawWithRetryOrFailSlowPath(int, v8::internal::AllocationType, v8::internal::AllocationOrigin, v8::internal::AllocationAlignment) [node]
#17 178.2  8: 0x14cded0 v8::internal::Factory::NewFillerObject(int, v8::internal::AllocationAlignment, v8::internal::AllocationType, v8::internal::AllocationOrigin) [node]
#17 178.2  9: 0x194fefb v8::internal::Runtime_AllocateInYoungGeneration(int, unsigned long*, v8::internal::Isolate*) [node]
#17 178.2 10: 0x7f683de6f276 
#17 230.1 Aborted (core dumped)
#17 230.1  ELIFECYCLE  Command failed with exit code 134.
#17 ERROR: process "/bin/sh -c pnpm run build" did not complete successfully: exit code: 134

Yeah actually this can be seen on our side too except we do not run into an OOM luckily, but building uses over 10GB of memory.

[johnforte]

It looks to be related to optimizePackageImports client was having the same issue, removed that and the build was actually faster.

[MarcJose]

I did try that as well but didnt change anything when removing all experimental options like taint, serverActions and optimizePackageImports.

Edit:
I just tried it with 15.4.4 again and all experimental functions disabled but same problem:

[dawi]

This might explain why our GitLab runners become unresponsive during Next.js builds.
We encountered a similar issue before: #77192 (comment)
It's happening again after updating to version 15.4.4.

[timneutkens]

Had a look at the shared .next/trace files for each build. Looks like file tracing is the issue. Will have a look if anything changed there.

15.4:

15.3:

@dawi the linked issue is unrelated (and that has been fixed). Can you share your .next/trace files too?

[timneutkens]

It seems to be related to this change: #78968. Will let the team know 🙏