https://pixelfed.fediverse.observer/dailystats
Looks like Pixelfed’s growth spurt is slowing down. Absent any new external stimuli I’m guessing it’ll stabilize around 200K to 300K monthly active users – over a hundredfold order of magnitude from what it was just a month ago.
I’ve been trying to set it up for 2 weeks now, the docker containers. There are so many backwards decisions, it’s pretty clear he doesn’t understand docker unfortunately. The entire app could be a snap to set up, but it’s such a convoluted setup that it scares a lot of people away. (It pretty much is scripted assuming you will run it on a VM, and a high cost one at that).
The fact that it’s built on PHP and Laravel in 2025 says a lot, and the fact that he started Loops this year on the same architecture also says a lot. It just doesn’t scale, it’s locked to a single host, and he’s finding that as the servers are tipping over.
Check out YUNOhost. I had a Pixelfed server up and running in 20 mins.
I’m not a PHP fan but it scales better than Python or Ruby (Mastodon) does. I think Dan is a cowboy of an engineer, but blaming performance on his stack choice is a bad take.
Not php but Laravel. Like I said it forces workers to be on the same host with the same storage as the API, it doesn’t allow scaling of multiple API nodes or worker nodes, and his docker containers require me running special commands left and right. No other docker containers of mine do that
Haven’t tried self hosting Pixelfed. Just mostly been trying it out on the main instance. I’m really shocked that it’s a mess to deploy with docker if it’s on PHP/Laravel.
I hate PHP these days for dev purposes but I think laravel ought to be able to scale enough to run most Pixelfed instances. Facebook ran PHP when it was much larger than Pixelfed.
Facebook had to basically rewrite php so it would scale though, it’s called Hack now
I refuse to use snaps
Heh