Embed Notice

>Before the server died, if I had an hour to kill before lunch and didn't wanna do anything too heavy, I could go play with the monitoring scripts or ask Postgres what the long-running queries were and see if I couldn't tweak it to make the DB run smoother

I did this except with forensics. One thing I think is (mostly) a flaw with Pleroma's implementation is that Lain used Postgres like a document store in the objects and activities tables. It's great for doing things like generating top ten lists from scrobbles but my gut tells me that actually processing the ActivityPub objects/activities into third normal form tables would have made the program might lighter on system resources.

>see if I couldn't get the 5xx rate down

I have some hilarious hacks going on on this instance to keep it from shitting the bed. Every few days, Pleroma used to peg the CPU at 100% and stay there in perpetuity until restarted, so I wrote a script that checks if the CPU is at 85% or higher, and if it is for five runs in a row, it restarts Pleroma. I ran this in a cron job every minute lol

>Well, to be fair, I can see url field looking way too much like the url field in AP objects, and the Audio type having something like that around, and externalLink fits a little better.

The url field in AP objects is not really defined at all. You can put basically anything you want in there and it's valid AP. Audio objects can be an MP3 file or a page that's about the track, as explicitly listed examples in the AP Vocabulary document lol. That's why I maintain that externalLink is stupid. It's pointless. url is for whatever you want it to be, as long as it's a URL that's somewhat on topic, and if it's not, this is user error. I did ultimately acquiesce to this externalLink desire. but I didn't like

>Bunch of stuff about Alex Gleason being a shitbird

I used to really like Alex, and in some ways I still do. He can be a lot of fun. He's a subscriber to the zine. That being said, he does have a habit of being a bit of a drama llama, although I can't really cast stones in this respect, and in a lot of ways I am full of regret for my contribution to e-drama in the last few years. I kind of want to dip my toes into technical FOSS discussions again but if I do I'll likely start my own shit from scratch specifically so that I don't have to clear every little thing with a PR process. I know you like this process and respect it, but if you disagree with the project maintainers about what the project should fundamentally be about in the first place? It's a recipe for failure.

In the context of fedi, I want to try new things, experiment, create new features, and treat AP like the free-for-all pseudo-protocol it is, which should explain why I did shit like rice out my instance's FE with randomly-generated /mu/ playlists with WebAmp and changed the background based on one of the 20 most recent @papes posts every few minutes. I get the impression that most other techies here, yourself included, want to spend their time doing efficiency tinkering and playing whack-a-mole with subtle bugs and security holes. There's plenty of merit in that sort of thing, but it's not really what I want to do, and I think it's self-explanatory why efficiency/security tinkering would come into conflict with experimentation and feature expansion, since doing the latter in a way that doesn't take months and months for every little thing entails quite a bit of risk.