I have the same problems with my 1.6 servers. However, I have not ran
the server without metamod.
Are we sure it is a memory leak? Is it possible that hlds does allocate
memory and uses it as some sort of cache?
I would suggest, running hlds without metamod until you run out of main
memory and see what happens.
IF the case is that the allocated memory is being used for caching
purposes (like map information), it shouldn't be a problem IF the coders
of hlds ensure that the game engine code stays in main memory. A way to
confirm this would be to allow hlds to use up all your memory and let it
start hitting swap. Once you know it is hitting swap, check to see what
the latency becomes for all players in the server, if the server keeps
the same latency, then it isn't a problem (until you run out of swap)
for game play. But, if everyone's ping skyrockets once your hitting the
swap partition, then that means the the net code/engine code and/or the
engine/net code data is being swapped, which is definitely a memory leak
and should be fixed.
Currently the machine running my servers have:
total used free shared
buffers cached
Mem: 514784 511640 3144 0 8256 95732
-/+ buffers/cache: 407652 107132
Swap: 987988 170836 817152
with an uptime of...
[EMAIL PROTECTED]:~$ uptime
12:33:34 up 4 days, 3:55, 1 user, load average: 0.68, 0.70, 0.64
I have a 26 and 21 player server. And the only users processes on the
machine are hlds.
I played last night on a full server with memory at that usage level and
experienced no problems.
I am using metamod, amx 0.9.9, hlguard, and headshot mod
m0gely wrote:
CS 1.6 server Iceworld 24/7 - 12 player
Mods: NONE. No metamod, NOTHING
Mem usage: 164M after *24 hours* (from 65M at startup)
Player connections: 1075
_______________________________________________
To unsubscribe, edit your list preferences, or view the list archives, please visit:
http://list.valvesoftware.com/mailman/listinfo/hlds_linux
