I have a small NUC-like machine in my home with an old external hdd connected to it. I use it to run my local gitlab, nextcloud and to test a few websites I build for the lolz.
If you too have a homelab, whether it's a single raspberry or an entire room full or racks, you know damn well that everything you have running locally as a web service keeps going until it doesn't, for whatever fucking reason. This time, it was the turn of my nextcloud.
The machine has arch linux running, I chose it since I already use it on my coding laptop and being a rolling release means I don't have to manually upgrade to a newer version, risking various fuck-ups and consequent screaming of profanity.
The downside is that arch is a bleeding-edge distro, so, despite being pretty good for what concerns security, as updates are pushed out some packages may still require legacy software to work as intended, since obviously not all developers for all packages can release simultaneously.
The problem was that php reached 8.2.x but nextcloud couldn't use anything beyond 8.1, so the highlighted solution was to download php-legacy, a package with a set of utilities which the cloud could use instead of mainline php.

Pretty easy, right? fuck my life, here we go.

I edited apache-httpd's configurations to link the new libraries, updated every reference in every virtual host that could possibly screw up the web server.

Done.

Then I went on and disabled the php-fpm mainline, creating a new systemd unit that would instead run the legacy executable and afterwards I edited nextcloud's additional configs so they use that instead.

Done, getting a bit dizzy, but I reboot everything and breathe.

At this point the migration should be complete, but wait, the server returns an error saying that the application is still trying to use php 8.2+...wait, what in the sysadmin Christ?

Back to nextcloud config, everything is set, everything else in every other fucking php-legacy and web server is fine, the old fpm service is disabled, I am confused, and why in the FUCKING FUCK is the new php-fpm unit failing to start at boot with "error 78/config - directory not found"? Hello? Am I being trolled by a shitty dual-core amazon fake NUC?
Maybe yes, cause it turns out that the unit was referencing a directory in the external hdd, which gets mounted at boot time after the unit itself starts, so nothing much, just a matter of tinkering with cron jobs, a reboot and at least this one is off my balls.

But why still isn't the server responding correctly? why? WHY?

After slamming my cock on the keyboard here and there scrolling back through all the config files I think to myself, hmmm, my gitlab is working flawlessly, well yeah, I didn't need to install the whole web stack, everything was nice and easy wrapped in a docker container...so why am I even here, why the fuck am I bothering with all this layered web-app bullshit, why don't I just run the up-to-date docker image that someone else has already set up for me, back up all the data and reupload them on the application?

Oh joy, you can't imagine, after 3...almost 4 hours of pure computer-touching the relief I had from seeing the blue web page with the "welcome to nextcloud" title.

Right now it's copying back all the files, and the external hdd is now linked to include the data folder.
Like really, everything was solved in two lines of bash.

I am still fuming, but at least I learned a valuable lesson, if you want a service up for yourself, implement it and deploy it as fucking easy straight-forward as you can, giving MAXIMUM priority to already fully-working options that are out there just waiting to be downloaded and used. I swing my scrotal sack on web-apps elegance as long as it's MY homelab in MY place.

Eat a fat dick php.

sudo pacman -Rns nextcloud
sudo systemctl disable --now php-fpm-legacy
sudo pacman -Rns php-legacy
sudo pacman -Rns $(sudo pacman -Qdtq)