OK the problem really appeared to be a scaling issue (sorry for blaming the American netflix watcher
) . I replaced a component of the HTTP server module that allows more API calls and suddenly all server responses go through as intended (its even more responsive as before).
All users have an efficiency of around 100% now!
Unfortunately I had to restart the server in order so all balances are reset to 0. I’ll grep the credits lost from the log files tomorrow when I got some sleep and will send them manually.
Of course I hope this will be the last interruption, but on the other hand I am doing this to figure those things out, so thank you for your patience.