Segmentation fault with ruby 3.3-jemalloc

[diasjorge]

Describe the bug
While testing the upgrade for one of our applications to ruby 3.3 we are consistently getting segmentation fault errors when running on our CI systems (github actions self hosted k8s runners).
We tested changing to not use jemalloc and the problem disappeared.
Is this the right repository to report this bug?
I haven't been able to reproduce this outside that environment.
Any recommendations so we can find the root cause? The error said "Segmentation Fault (core dumped)" but I wasn't able to find the core dump anywhere on the system.

To Reproduce
Linux distribution: ubuntu 24.04 container inside github actions.
Package name: fullstaq-ruby-3.3-jemalloc

Steps to reproduce the behavior:

1. Run command 'bin/rails db:setup'

I'm sorry I can't give you more details, how could I gather more information? Thank you

[arashster]

I'm facing a similar issue when using ruby-3.4-jemalloc on Ubuntu 24.04
Switching to the non-jemalloc version works no problem.
It occurs in a similar command related to a database action (in our case a
rake db:migrate

[arashster]

This issue surfaced now on multiple different servers on AWS, all using Ubuntu 24.04 (x86_64) and Rails 8.0.2 , and mysql2 gem

This happened when doing rake db:migrate

I'm not sure if there is an issue with jemalloc on 24.04

[arashster]

As a workaround for now, I was able to use the regular "fullstaq-ruby-3.4" but load via jemalloc via environment variables:

LD_PRELOAD=/usr/lib/x86_64-linux-gnu/libjemalloc.so.2
MALLOC_CONF='dirty_decay_ms:1000,narenas:2,background_thread:true'

This drastically reduced our RAM usage (reduced it to a 1/3rd) and performed on par or better than than the jemalloc 3.6.0 version on older Ubuntu.

I believe Ubuntu 24.04 and jemalloc 3.6.0 are not fully compatible.