Lemmy’s machine-generated ORM SQL and hand-made flawed PostgreSQL TRIGGER logic is so bad, bloated. The developers on GitHub brag about “high performance”. It’s unbeliable.
In reality, small instances work because it has so many SQL performance problems that it mostly only is stable with little posts and comments in the database. They dd everything they could to avoid using Lemmy itself to discuss !lemmyperformance@lemmy.ml topics and hang out on Matrix Chat to avoid using the constantly-crashing servers they created.
If you go to a server with no users creating comments and posts and only has a tiny amount of data, it does crash a lot less.
Lemmy’s machine-generated ORM SQL and hand-made flawed PostgreSQL TRIGGER logic is so bad, bloated. The developers on GitHub brag about “high performance”. It’s unbeliable.
In reality, small instances work because it has so many SQL performance problems that it mostly only is stable with little posts and comments in the database. They dd everything they could to avoid using Lemmy itself to discuss !lemmyperformance@lemmy.ml topics and hang out on Matrix Chat to avoid using the constantly-crashing servers they created.
If you go to a server with no users creating comments and posts and only has a tiny amount of data, it does crash a lot less.
That’s an interesting community.