Bug #81664 [Com]: php8.1 core dump

Edit report at https://bugs.php.net/bug.php?id=81664&edit=1 ID: 81664 Comment by: cbing at cybernetics dot com Reported by: yyb8 at vip dot qq dot com Summary: php8.1 core dump Status: Open Type: Bug Package: JIT Operating System: ubuntu 18.04 PHP Version: 8.1.0 Block user comment: N Private report: N New Comment: I feel that this is also related to, or a duplicate of, https://github.com/php/php-src/issues/8461 Previous Comments: ------------------------------------------------------------------------ [2022-01-10 07:48:44] yyb8 at vip dot qq dot com install laravel9: composer create-project laravel/laravel example-app "dev-master" then in the routes/web.php write: Route::get('/', function () { return 'hello world'; }); ------------------------------------------------------------------------ [2022-01-10 07:34:04] dmitry@php.net How to install that "laravel9 echo hello world"? ------------------------------------------------------------------------ [2021-12-17 01:14:45] yyb8 at vip dot qq dot com I run the ab -n 10000 -c 100 -k 'http://www.laravel9.com' then crash. If I open the http://www.laravel9.com first in browser, next step run the ab -n 10000 -c 100 -k 'http://www.laravel9.com', then will not crash. ------------------------------------------------------------------------ [2021-12-17 01:13:43] yyb8 at vip dot qq dot com I run the ab -n 10000 -c 100 -k 'http://www.laravel9.com' the crash. If I open the http://www.laravel9.com first in browser, nest step run the ab -n 10000 -c 100 -k 'http://www.laravel9.com', then will not crash. ------------------------------------------------------------------------ [2021-12-16 15:05:39] jfmeinesz at gmail dot com Hi, Not 100% sure if my comment is related, but I have reasons to think it is so I'll post it (basically, the logs says core dump as well). I am also using PHP8.1, Laravel 8 along with NGINX 1.21/1.19 (tested on both), and php-fpm using opcache and JIT tracing. I am using docker and docker swarm to serve this project. Also, my nginx and php containers have a health check that waits for a "service nginx status" (NGINX) and a custom laravel command for the php-fpm container. My problem is similar, but not quite the same : it does not happens every time I start the containers, sometimes it works fine, sometimes the bug occurs, I'd say it happens about 50% of the time... Seems like this bug is a tough nut When this bug happens, my php-fpm containers are turning crazy, returning HTTP 509 and printing the following output in loop : WARNING: [pool www] child 10366 exited on signal 11 (SIGSEGV - core dumped) after XXXXX seconds from start Some requests are served, some are HTTP 509, and the warning is returned every seconds in the logs. Strange thing that might helps : during this episode, I also had the following output printed twice in the logs : WARNING: [pool www] seems busy (you may need to increase pm.start_servers, or pm.min/max_spare_servers), spawning 8 children, there are 1 idle, and 22 total children Strange thing is that 22 children wasn't my max_children defined in the config, here is my config : clear_env = no pm = dynamic pm.max_children = 50 pm.start_servers = 10 pm.min_spare_servers = 10 pm.max_spare_servers = 15 pm.max_requests = 600 When this bug happens, I have to shutdown the containers, and start them up again until it disappears (sometimes I need to restart the containers twice in a row). When the bug is not there, everything works fine though, my whole project have no issues. Hope this helps you to find the origin of this bug. Regards, ------------------------------------------------------------------------ The remainder of the comments for this report are too long. To view the rest of the comments, please view the bug report online at https://bugs.php.net/bug.php?id=81664 -- Edit this bug report at https://bugs.php.net/bug.php?id=81664&edit=1