SRCDS Steam group


Problem w/CS:S, but HL2 DM works fine...
#1
Sad 
Hello, I'm a fairly experienced Linux Admin (ran a CS 1.x Server on HLDS for a lot of years) and I'm finally able to try a CS:S server. One problem -- CS:S won't work for me, but HL2 DM works like a champ. Sad

1) Command Line:
Code:
[joe@server source]$ ./srcds_run -game cstrike +map de_dust -maxplayers 8 -debug
Auto detecting CPU
Using SSE2 Optimised binary.
Enabling debug mode
Auto-restarting the server on crash

Console initialized.
Attempted to create unknown entity type event_queue_saveload_proxy!
Game .dll loaded for "Counter-Strike: Source"
maxplayers set to 8
Network: IP (removed), mode MP, dedicated Yes, ports 27015 SV / 27005 CL
Executing dedicated server config file
*** glibc detected *** free(): invalid pointer: 0xb7fd7f00 ***
./srcds_run: line 344:  3807 Aborted                 (core dumped) $HL_CMD
cat: hlds.3793.pid: No such file or directory
email debug.log to linux@valvesoftware.com
Wed Apr 20 21:54:16 EDT 2005: Server restart in 10 seconds
Wed Apr 20 21:54:18 EDT 2005: Server Quit

2) Debug Log (seems useless in this case):

Code:
[joe@server source]$ cat debug.log
----------------------------------------------
CRASH: Wed Apr 20 21:54:16 EDT 2005
Start Line: ./srcds_i686 -game cstrike +map de_dust -maxplayers 8 -debug
End of Source crash report
----------------------------------------------

3) Distro: Fedora Core 3 w/all the latest updates.

4) gcc version: gcc (GCC) 3.4.3 20050227 (Red Hat 3.4.3-22.fc3)

5) glibc version: ldd (GNU libc) 2.3.5

6) Kernel version:

[b]Linux server 2.6.11-1.14_FC3 #1 Thu Apr 7 19:23:49 EDT 2005 i686 i686 i386 GNU/Linux[b]

7) Hardware is a P4 1.4GHz, 512MB RAM, 20GB IDE HD.

I'm not sure why, but HL2DM works absolutely fine without the "invalid pointer" glibc error, with no problems at all. I have tried re-downloading Counter-Strike Source in its entirety via Steam, but had the same results. I've also run srcds w/-autoupdate, same results. I've tried forcing it to use srcds_i486, same results (except the hex number in the error is different of course).

Any ideas? I have selinux disabled entirely, I have tried running srcds as root (temporarily), and I have verified all file ownership & permissions.
Reply
#2
try running export MALLOC_CHECK_=0

then start the server up.
[Image: userbar_wow.jpg]
starting 9/24/2006 if your problem has been solved please edit your first post and add [solved] to the begining of the title. Thanks.
Reply
#3
U DA MAN!

Awesome, that did the trick!

...Now why would I have to disable the memory check? I take it it's some sort of glibc environment var?? glibc bug?
Reply
#4
I believe its actually a cs:s buig with glibc.

although i thought that bug was fixed..

Are you sure your cs:s server is up to date? lol
[Image: userbar_wow.jpg]
starting 9/24/2006 if your problem has been solved please edit your first post and add [solved] to the begining of the title. Thanks.
Reply
#5
cryotek Wrote:try running export MALLOC_CHECK_=0

then start the server up.

I ran into the exact same error with hl2mp on a FC3 box (2.6.10-1.741_FC3), but I only saw the error only when trying to load a server.cfg file, or using the 'exec' command... The MALLOC_CHECK trick worked for server.cfg (that is, it loads server.cfg automatically w/o crashing now), however the exec command simplly spits back "cannot exec <filename>".

Since I can get everything working with the server.cfg file now, I'm more just curious what is going on, if anyone could offer some insight...
Reply


Forum Jump:


Users browsing this thread: 2 Guest(s)