High CPU usage when navigating directly to posts and profiles
Since about a week ago I've noticed very high CPU usage when navigating directly to /notice/*
and profiles. Sometimes it takes 2 or 3 tries because Pleroma can't return the request before 60 seconds and a 500 is served and some posts just won't load no matter how many times you try. Normal browsing of the site seems unaffected. I've helped take the strain off and keep my server relatively fast by using Nginx to cache the html responses of direct links posts so after the initial loading it loads subsequent requests without a problem and fast but this workaround isn't ideal. I'm not even sure what's causing it either. Logs don't display anything useful. I initially thought it might be the Pleroma.Web.Metadata
that was causing the issue because when that was added was around the time I really noticed this issue it but disabling it had no effect. Postgres isn't even under heavy load, it's all beam.
>nb4 reported beam CPU usage isn't accurate
I know reported beam CPU usage isn't always accurate but if that isn't pegging my CPU at 100% usage then at most there is only 25% CPU usage with the same issues.