fix(dev): ensure socket path is accessable

[sbittmann]

🔗 Linked issue

nuxt/nuxt#15164

❓ Type of change

• 📖 Documentation (updates to the documentation or readme)
• 🐞 Bug fix (a non-breaking change that fixes an issue)
• 👌 Enhancement (improving an existing functionality like performance)
• ✨ New feature (a non-breaking change that adds functionality)
• ⚠️ Breaking change (fix or feature that would cause existing functionality to change)

📚 Description

On windows we are facing the issue, that the devServer returns status 503. This is caused by the existsSync() call, which returns false if the the api is under load. With fs.promises.access I didn't face this problem anymore.

📝 Checklist

• I have linked an issue or discussion.
• I have updated the documentation accordingly.

[codecov]

Codecov Report

Merging #1115 (e74885f) into main (20c841c) will decrease coverage by 0.04%.
The diff coverage is 65.21%.

@@            Coverage Diff             @@
##             main    #1115      +/-   ##
==========================================
- Coverage   76.58%   76.55%   -0.04%     
==========================================
  Files          70       70              
  Lines        7193     7212      +19     
  Branches      715      718       +3     
==========================================
+ Hits         5509     5521      +12     
- Misses       1683     1690       +7     
  Partials        1        1              

Impacted Files	Coverage Δ
src/dev/server.ts	83.81% <65.21%> (-1.52%)	⬇️

[danielroe]

In preliminary testing, this seems to decrease flakiness in nuxt CI dev server tests.

Is it also worth switching to fsp.access here too?
https://github.com/unjs/nitro/blob/34d9d1dc074f2c13f6d5c83a50d500d501a725ac/src/dev/server.ts#L97-L99

[pi0]

I am aware this is an important DX fix. However needs to be tested more if we can find a solution without per/request overhead of checking sockets and that this works universally. Delaying to next patch/minor release.

[csprance]

This issue still seems to be happening even with this fix.

[pi0]

Tracking via #1454