[BUG] Hitting superfluous response.WriteHeader during stress test

roadrunner-server / roadrunner

🤯 High-performance PHP application server, process manager written in Go and powered with plugins

https://docs.roadrunner.dev

MIT License

7.92k stars 411 forks source link

[BUG] Hitting superfluous response.WriteHeader during stress test #333

Closed andrei-dascalu closed 4 years ago

andrei-dascalu commented 4 years ago

I tried this code:

Symfony 5 out of the box (then add roadrunner integration) or a simple php router (I can provide access to my repo). Running stress test with autocannon

autocannon -c 100 -d 40 -p 10 -t 120 http://127.0.0.1:8080

Seems to be in line with issue #301

I expected to see this happen: no logs

Instead, this happened:

2020/05/28 16:32:33 http: superfluous response.WriteHeader call from github.com/spiral/roadrunner/service/http.(*Handler).handleError (handler.go:135)
2020/05/28 16:32:33 http: superfluous response.WriteHeader call from github.com/spiral/roadrunner/service/http.(*Handler).handleError (handler.go:135)

The version of RR used: 1.8.1

rustatian commented 4 years ago

@andrei-dascalu Thanks for the report. Seems, that I need to deeper investigate this bug :) Linked: #301

andrei-dascalu commented 4 years ago

Hi,

I've made some other tests using RR and might be useful info:

tried a basic setup with Ubiquity framework (using their integration, though their README is outdated, their CLI tool can do biquity serve -t=roadrunner and will setup some sane defaults). They don't seem to setup a relay via socket and using the same stress test I got a much weaker general result, but no such errors across a dozen or so tests.
Symfony (and my setup) use sockets and got to a much higher request count, always getting some errors)

rustatian commented 4 years ago

@andrei-dascalu Thank you for the information. In high overview I know the root cause of such errors, just need to allocate some time to dig into :)