I joined a random server in the game and before I lost connection looked at the server’s memory. The client memory looked alright, but the server memory was…
Unfortunately I have no idea what’s causing the UntrackedMemory, but even so, the core memory is insane, even if it’s a 40 player server. Is that normal?
And that aside, what could have caused the Place memory to increase so much? When I joined a new server with ~20 players, the memory showed…
I mean, I have some GUIs stored in ServerStorage, and most of them are stored in ReplicatedStorage. And there are a few ScreenGUIs displaying only text information. But why would they increase over time? I’m not creating new GUIs in the server (or if I am, they’re Destroyed eventually).
Also, does anybody have an idea why the CoreMemory would jump from 308 to a whopping 1800? Like why is network/cache and network/raknet so high? Is there anything I can do about those?
By default, Any guis that you have in StarterGui (guis that all clients spawn with) also replicate to the server, and are therefore detected by/can affect the server.
Referring to this, I’ve seen a lot of games have the same issue, somewhat randomly. It seems to be an internal issue. There is nothing you can do about this area of memory as it is:
Arbitrary memory locations that are not tagged
(according to the wiki)
Place memory is pretty high. On the list, GUIs are reported second most expensive. Are more and more GUIs being created without being destroyed? As for signals, follow @LadyCelastia’s advice. Disconnect any events that could cause a leak (especially player events).
The network cache appears to be the highest in the core memory list. This is the cache of network objects which were loaded (e.g images or meshes). This is filled as the server caches in remote resources (such as those on the website). Try using fewer meshes or smaller image files, as most of them will end up cached into memory.
Because of the fact that you joined a new server and the memory was back to normal, it’s probably a problem building up over time. Try measuring server uptime and seeing how it is affecting the problem.
See if any of these modifications or checks solve the problem.