miwarnec / libuv2k_public Goto Github PK

from discord.
error: https://hatebin.com/zeqrqnpqzq

from discord:

imer
:
@vis2k you might want to link the msvc runtime statically (/MT as a compiler flag) for libuv or tell people they need to install/ship the correct redist to players
see https://discordapp.com/channels/343440455738064897/772106602652237834/780803293542481951

JamesFrowen
:
@vis2k server crashes at ~240 connections

OutOfMemoryException: Out of memory
at (wrapper managed-to-native) System.Object.__icall_wrapper_ves_icall_array_new_specific(intptr,int)
at libuv2k.WriteRequest..ctor () [0x00000] in :0
at libuv2k.TcpStream+<>c.<.cctor>b__72_0 () [0x00000] in :0
at libuv2k.Pool1[T].Take () [0x0000e] in <eb9ed698661e43a493a7da799bacb399>:0 at libuv2k.TcpStream.Send (System.ArraySegment1[T] segment) [0x00036] in :0
at Mirror.libuv2kTransport.ServerSend (System.Int32 connectionId, System.Int32 channelId, System.ArraySegment1[T] segment) [0x00018] in <3ec8af3d200d4011998ac0b02fe3b577>:0 at Mirror.NetworkConnectionToClient.Send (System.ArraySegment1[T] segment, System.Int32 channelId) [0x0005b] in <05786aaf81e545fabc4dd5d60afbc631>:0
at Mirror.NetworkServer.SendToReady[T] (Mirror.NetworkIdentity identity, T msg, System.Boolean includeOwner, System.Int32 channelId) [0x000ad] in <05786aaf81e545fabc4dd5d60afbc631>:0
at Mirror.NetworkIdentity.SendUpdateVarsMessage () [0x0007a] in <05786aaf81e545fabc4dd5d60afbc631>:0
at Mirror.NetworkIdentity.ServerUpdate () [0x00016] in <05786aaf81e545fabc4dd5d60afbc631>:0
at Mirror.NetworkServer.Update () [0x00033] in <05786aaf81e545fabc4dd5d60afbc631>:0
at Mirror.NetworkManager.LateUpdate () [0x00000] in <05786aaf81e545fabc4dd5d60afbc631>:0

from discord: JesusLuvsYooh
:
Probably not the best of times to mention this (sorry vis) but.
There was a crazy leak in libuv, one time i managed to close server just before i lost the vps, second time i did not.

What i think caused it, was i put another transport on network manager in unity editor, typed in the server ip and clicked connect.

(Client) Frame rate struggled to keep around 30fps, where as on the other transports it was 60fps (capped at 60, could have been higher).
Could be caused by the hosts possible memory leak, fixing that, may fix this?

Ram usage was 1gb more at the 151 client mark, compared to the other transports.
(Ram usage on other transports for 150 clients was under 100mb)
Did not leave on long enough to see if it kept increasing over time, or if its upon clients connecting.

• This test did not have a client connect with wrong transport like my theory on the other (insane) memory issue.