Moving to serversideup/php:8.1-fpm-nginx increased bootstrap times to 180ms

[dejagersh]

Before I ran my Laravel app on alpine container on Fly.io.

Recently I moved to serversideup/php:8.1-fpm-nginx as it's much less code I have to manage. I am using the default Fly.io configuration that Fly provides for Laravel apps. However, when I run my app on serversideup/php:8.1-fpm-nginx, the bootstrapping times go up by A LOT. These are the p0.9 values:

and averages:

I deployed the container shortly after 19:00. This is what the trace is looking like:

As you can see, bootstrapping takes forever.

Anyone here have any idea why this could be? Or how I could debug it?

[jaydrogers]

Thanks for the report!

Out of curiosity, could you run these same tests with my beta image?

serversideup/php:beta-8.1-fpm-nginx

It’s a pretty major overhaul, but I have everything documented here: #59

[dejagersh]

Trying pm = dynamic (on serversideup/php:8.1-fpm-nginx). I looked at my old configuration for alpine and I was
using pm = dynamic.

Edit: didn't help anything.

[dejagersh]

The odd thing is that I can't reproduce this in a fresh Laravel project. There must be something in my project that doesn't work well on serversideup/php:8.1-fpm-nginx but works fine on alpine. The app is so simple though...

[dejagersh]

But looking at some of my other projects: they get similarly terrible response times... It's at least 130ms for pages that don't even do DB queries

[jaydrogers]

Is it possibly related to this openbase_dir stuff? https://community.serversideup.net/t/bad-performance-of-laravel-based-wintercms/684/8?u=jaydrogers

[jaydrogers]

@thejager: Just following up on this because I am interested in figuring out what the problem is. Let me know if it is related to that discussion above with openbase_dir 👍