I'll be honest here, as a humble instance admin receiving all kinds of DMs and emails...

The abuse, excessive reports... folks our instance grew from 60 users to 22000 in the span of two weeks. I didn't expect to suddenly commit 6-8 hours of my day to this. Knowing I have other commitments like a day job and other things.

I'm burning the candle on both ends here. Look at this toot timestamp. Please save space for some patience.

Y'all clearly don't remember this image:

#mastoadmin #mstdnca

I do IT for a living. I know you're getting those 500 errors. I know that the instance is slow.

Do I want that for you? Abso-fucking-lutely not. I want mstdn.ca to load for you like the bird app does. Like how your gmail loads. Snippy snappy.

Unless you have explicit experience with Puma and Sidekiq and are willing to get on a call with me, save your abusive DMs and emails.

I'm scouring the internet, consulting with the Mastodon devs and other operators. This is a trial and error process.

@chad thanks for the good work!
I had good success switching a production ruby app to a ruby VM compiled with jemalloc (https://github.com/Binaergewitter/serious-bg/pull/390/commits/c877025f2eeeef36c77c371de96a64db1f754222), resulted in ?20%? memory savings, a bit less CPU and was basically a drop in. Not sure if you had looked into that, but if not, it’s a nice optimization without the need for code changes
Switch fly deployment to docker container with jemalloc ruby by rb2k · Pull Request #390 · Binaergewitter/serious-bg

We are running on the free tier of fly.io Running on the free tier of something is VERY much on brand for the podcast and handcrafted CMS with which we fully celebrate mediocrity in hobby projects ...

GitHub