Closed afoster closed 10 months ago
Does this happen on a fresh Laravel app with Pulse & the docker-compose project?
Hi Andy,
Not sure about that yet, but I've stripped my real codebase back and published it at https://github.com/afoster/pulse-128.
This more or less represents my app's platform and third party dependencies without most of the codebase.
Interestingly when testing it on an Intel Mac I can't immediately replicate the problem, but it can be replicated straight away on two different Apple Silicon Macs. Update: it does occur on the original app under Intel. Update: it occurs on both architectures in the demo app above
I will try a fresh Laravel app next.
Fresh Laravel install + similar Docker stack: https://github.com/afoster/pulse-128-bare
Can't replicate the problem here yet.
I've had some success disabling opcache, but I'm not sure whether this is down to my opcache settings or if there's a legitimate problem between Pulse and opcache.
I've re-enabled opcache with some saner config and it seems stable in my staging environment with no real users:
[opcache]
opcache.enable=1
opcache.jit_buffer_size=33554432
opcache.jit=tracing
opcache.revalidate_freq=5
opcache.validate_timestamps=1
opcache.memory_consumption=64
Interesting... what was the previous opcache config?
I can't think of anything directly in Pulse that would cause issues like this - the dashboard is pretty straightforward. I'm wondering whether it's something in Livewire (or another Pulse dependency) that's triggered the problem.
Hi Jess, so the previous opcache config was:
opcache.enable=1
opcache.enable_cli=1
opcache.jit_buffer_size=500000000
opcache.jit=1235
opcache.revalidate_freq=0
opcache.validate_timestamps=1
opcache.max_accelerated_files=10000
opcache.memory_consumption=192
opcache.max_wasted_percentage=10
opcache.interned_strings_buffer=16
opcache.fast_shutdown=1
Embarrassingly I suspect this was copypasta from some online tutorial so it wouldn't surprise me if it was rubbish. That said, it has been running OK up until now. I think like you say it'll be something to do with Pulse's dependencies or even something as simple as the fact that Pulse plus its deps adds more PHP files to the codebase and that's exceeded some watermark with opcache. Happy for you to close this if you think it's out of scope.
Thanks
@afoster Appreciate this. I think until we can reproduce it or nail it down to a specific thing in Pulse we will close this. Others can report if it is widespread and we can try and dig in further.
Pulse Version
1.0.0-beta2
Laravel Version
10.34.2
PHP Version
8.2.8
Database Driver & Version
MariaDB 10.11 under Docker on arm64 (Mac)
Description
My app is running as a docker-compose project. I have an "app" container based on php:8.2-fpm-bullseye and it sits behind a reverse proxy container based on nginx:alpine.
Shortly after running
composer require laravel/pulse
and browsing to the/pulse
route, PHP processes under php-fpm start crashing:Eventually all children die, and the app is inaccessible (nginx throws 502 instantly on all requests).
php artisan about
composer.json
I'm not sure how to go about debugging this.
Steps To Reproduce
I haven't tried reproducing in an open codebase yet, but happy to attempt it.