Jean Boussier

7.1 is in security only maintenance: https://rubyonrails.org/maintenance

Almost by definition, a SEGFAULT cannot be a Rails bug. This is either a Ruby bug or a bug in one C extension. Not to say we shouldn't investigate this,...

> How do I capture the C backtrace of the crash given that the process will still be alive and other code paths will be continuously executed. Ideally you configure...

> * ruby:alpine based Docker image Hum, I wonder if you aren't simply running out of stack space. `alpine` is infamous for having extremely small stack space for secondary threads...

@aliismayilov I hope these core dumps are from a test app? Because I didn't expect you to share them. Core dumps contains the entire program memory, including secrets and such....

No worries, I should have warned you about that before asking you. Looking at your stacktrace, the relevant part is: ``` 4 0x0000c074590782c0 [PAC] in sigsegv (sig=11, info=0xfa7ecc9a2da0, ctx=0xfa7ecc9a2e20) at...

> So, the 2 backtraces that I shared both failed on the following line: Oh, that's good. When I first looked at it, I couldn't see a common line. ```ruby...

@julik if you got some PR, or even a draft I'd be happy to look at it. But otherwise I'm quite busy on Ractors these days and this isn't very...

> PR removes all the thread pool executor code That entirely defeat the purpose of AC::Live though. If you don't spawn a new thread, the server thread won't be released.

> if its only about removing the thread pool introduced on [# 52731](https://github.com/rails/rails/pull/52731) and instead using simple Thread class I don't think that's it either. I'm not even convinced there...