nextjs - 💡(How to fix) Fix [Turbopack] next-server process becomes zombie with 400-700% CPU, doesn't terminate [1 comments, 2 participants]

Link to the code that reproduces this issue

Private repository - cannot share, but reproduction steps below.

To Reproduce

1. Create a Next.js 16.0.7 app with Turbopack (default)
2. Have a large React component (~2400 lines) with complex context patterns
3. Run next dev (uses Turbopack by default)
4. Navigate to pages that use these components
5. Observe next-server process CPU spike to 400-700%

Current vs. Expected behavior

Current:

• next-server process spikes to 400-700% CPU
• Render times go from ~300ms to 50-100+ seconds
• Process won't terminate even with pkill -9 -f next
• .next directory cannot be deleted ("Directory not empty")
• Only fix is to kill -9 <specific PID> the zombie process

Expected:

• Normal CPU usage
• Process terminates cleanly when killed
• Render times stay consistent

Provide environment information

Operating System:
  Platform: darwin
  Arch: arm64
  Version: Darwin 25.1.0
Binaries:
  Node: 22.20.0
  pnpm: 9.14.4
Relevant Packages:
  next: 16.0.7
  react: 18.3.1
  react-dom: 18.3.1

Which area(s) are affected? (Select all that apply)

Turbopack

Which stage(s) are affected? (Select all that apply)

next dev (local)

Additional context

• Switching to next dev --webpack completely resolves the issue
• The Single item size exceeds maxSize warning appears in logs before the issue manifests
• Issue seems related to HMR on large files with complex React context provider patterns
• Logs show compile times are reasonable (~1s) but render times balloon to 50-100+ seconds
• Multiple zombie next-server processes can accumulate over time if not manually killed

Workaround

Use webpack instead of Turbopack:

next dev --webpack