Kubernetes containers exited with error code 139

What happened?

We've tried to deploy our APIP application based on the frankenPHP version 1.3.3 (with base image dunglas/frankenphp:1.3.3-builder-php8.2/dunglas/frankenphp:1.3.3-php8.2) in a AWS EKS cluster.

When the site is getting busy, the containers start to exit with code 139 very soon and we don't see any specific logs output.

However we still can't reproduce the same issue by load testing using tools like k6 in our testing environment.

Build Type

Docker (Debian Bookworm)

Worker Mode

Yes

Operating System

GNU/Linux

CPU Architecture

x86_64

PHP configuration

expose_php = 0
apc.enable_cli = 1
session.use_strict_mode = 1
zend.detect_unicode = 0

realpath_cache_size = 4096K
realpath_cache_ttl = 600
opcache.interned_strings_buffer = 16
opcache.max_accelerated_files = 20000
opcache.memory_consumption = 256
opcache.enable_file_override = 1

memory_limit = 1024M
post_max_size = 10M
upload_max_filesize = 10M

opcache.preload_user = root
opcache.preload = /app/config/preload.php

Relevant log output

No response

Start by reviewing the Kubernetes deployment and container logs for the FrankenPHP 1.3.3 images, focusing on the exit-139 events under production load. Compare the AWS EKS configuration with the load-testing environment and establish a reproducible failure; done means identifying the cause and documenting or validating a fix.