@alan It sounded like there were a couple of points on the db cleanup - 1) complexity and 2) postgres can handle the millions of rows it just swaps stuff out to disk.

I wasn't going to even start to argue that 1) Whatever goes in to the db should be removable, for the most part, complexity be damned (I can say that as a layman as I see others have figured it out, right?), and 2) Of course this answer doesn't address the notion of infinite expansion and suggests all sorts of other little issues that come up with an excessively large db for smaller systems (such as initial load time when you reboot, etc).

And what about just having a nicely resource-maintained system so you can actually have a few handfuls of active users without just having oodles of extra junk around on disk or in the db?

Again, I wasn't going to argue it - I felt I needed to just go find the better solution for what I wanted to run.

I'm kind of glad that Mastodon has advanced and is as mature & stable as it is.

@alan It became apparent recently that instance resource manageability in Sharkey is going to continue to be an issue for the forseeable future.

Aside from remote content media cache purging (which is by default just a local disk store), there were no database cleanup routines in Misskey or Sharkey when I started with Sharkey, but there were several filed issues with Misskey and they had it on their to-do list - which is why I figured it was not too bad an idea to go with Sharkey. Misskey added their first round of db cleanup functionality in their August '25 release (IIRC), so was expecting Sharkey to have it brought in whenever they merged in that release.

Unfortunately, Sharkey devs did not like how Misskey handled that feature set and I believe are hard forking (or considering it), leaving that functionality up in the air for however long.

I had a bit of a chat with them to find out more about plans, and they were really quite against the notion of managing db cleanup in many ways - some types of tasks they're open to, but not, in my opinion, in enough ways to suggest that an admin has real control over the resource management of their instance (and this is one of the biggest issues I've seen in many of the fedi platforms - admin tools can be and often are a total afterthought - which is a terrible path to take if you want adoption that has any kind of longevity).

So, I decided I would re-check-out the Mastodon server side a little more in depth... and with Mastodon having caught up a bit on some of it's lack of features in the past few years, and having much better admin tooling in place, it seemed like I'd better push to move now.

So here we are... 🤷