21308 Segmentation Fault - TF2 Server

[Jcink]

Hi,

It seems I've managed to bork my server with the latest updates to TF2.. and I have no idea how to trace this one down.

Code:

[jcink@localhost orangebox]# ./srcds_run -console -game tf +ip 192.168.0.197 -port 27075 -debug +map pl_goldrush +maxplayers 32
Auto detecting CPU
Using SSE2 Optimised binary.
Enabling debug mode
Server will auto-restart if there is a crash.

Console initialized.
./srcds_run: line 352: 21308 Segmentation fault      (core dumped) $HL_CMD
cat: hlds.21294.pid: No such file or directory
email debug.log to linux@valvesoftware.com
Thu Jan 29 20:41:38 EST 2009: Server restart in 10 seconds
Thu Jan 29 20:41:39 EST 2009: Server Quit

debug.log doesn't contain anything interesting at all. There is a core dump but it's not like I understand it. This server was working just fine until yesterday when I ran the updater. Did it like this:

Code:

./steam -command update -game "tf" -dir .

After that, it downloaded it and all... did its thing. "HLDS installation up to date" ... and now this segmentation fault.

Any ideas? I know there's tons of reasons for segfaults, so... but is anyone else having problems with the latest updates?

Server data:

gcc --version
gcc (GCC) 4.1.2 20071124 (Red Hat 4.1.2-42)

ldd --version
ldd (GNU libc) 2.5
uname -a

Linux localhost.localdomain 2.6.18-92.1.22.el5.centos.plus #1 SMP Wed Dec 17 10:49:19 EST 2008 x86_64 x86_64 x86_64 GNU/Linux

Thanks for any advice. I have no idea what I'm doing wrong, this was working just fine for the past month and I've not really made any changes to my server itself...

[BehaartesEtwas]

try opening the core file with gdb:

Code:

gdb ./srcds_i686 core.XXXX

(XXXX is some number... look into the directory where srcds_i686 is located)
then run the command "bt" and post the output.

[kilorat]

I have the same thing happening, I wasted a lot of time checking metamod/sourcemod, but it crashes even without it!
And not everytime either, if I keep trying to restart it, it will start up every 5 times or so. Once its up, its good for half a day then usually dies.

Core was generated by `./srcds_i486 -game tf -port 27015 -debug'.
Program terminated with signal 11, Segmentation fault.
#0 0xf53269c4 in FactoryList_Store ()
from /steam/orangebox/tf/bin/server_i486.so
(gdb) bt
#0 0xf53269c4 in FactoryList_Store ()
from /steam/orangebox/tf/bin/server_i486.so
#1 0xf5623524 in CServerGameDLL:LLInit ()
from /steam/orangebox/tf/bin/server_i486.so
#2 0xf775836a in SV_InitGameDLL ()
from /steam/orangebox/bin/engine_i486.so
#3 0xf781fca5 in Host_Init ()
from /steam/orangebox/bin/engine_i486.so
#4 0xf78ad613 in Sys_InitGame ()
from /steam/orangebox/bin/engine_i486.so
#5 0xf78b0e85 in CEngine::Load ()
from /steam/orangebox/bin/engine_i486.so
#6 0xf78ae16e in CModAppSystemGroup::Main ()
from /steam/orangebox/bin/engine_i486.so
#7 0xf7914685 in CAppSystemGroup::Run ()
from /steam/orangebox/bin/engine_i486.so
#8 0xf78ae38f in CDedicatedServerAPI::ModInit ()
from /steam/orangebox/bin/engine_i486.so
#9 0xf7a3220a in CDedicatedAppSystemGroup::Main () from bin/dedicated_i486.so
#10 0xf7a91ba5 in CAppSystemGroup::Run () from bin/dedicated_i486.so
#11 0xf7a8c0db in CSteamApplication::Main () from bin/dedicated_i486.so
#12 0xf7a91ba5 in CAppSystemGroup::Run () from bin/dedicated_i486.so
#13 0xf7a3278b in main () from bin/dedicated_i486.so
---Type <return> to continue, or q <return> to quit---
#14 0xf7a2ab83 in DedicatedMain () from bin/dedicated_i486.so
#15 0x0804949d in ?? ()
#16 0x00000006 in ?? ()
#17 0xffe44674 in ?? ()
#18 0x0805396b in ?? ()
#19 0xffe425c8 in ?? ()
#20 0xffe45a07 in ?? ()
#21 0x00000000 in ?? ()

Its right when it is starting up, before it even loads the map I think.

$ ./srcds_run -game tf -port 27015
Auto detecting CPU
Using SSE2 Optimised binary.
Server will auto-restart if there is a crash.

Console initialized.
Game.dll loaded for "Team Fortress"
./srcds_run: line 352: 15449 Segmentation fault $HL_CMD


My guess is theres something about the OS that Valve doesnt know about. I'm using Centos 5.2 64-bit.

[Jcink]

I've been testing this and I have the same exact setup as kilorat, and problem, and I'm on CentOS 5.2 64 bit. And I can get it to start up, 1 in 10 times it will boot. That confused me a few times as well. And I did a complete fresh install as well to make sure metamod and sourcemod was not the case; it's still messed up, exact same problem. So I assume that since the defaults don't work us on CentOS 5 64 bit are screwed until valve fixes this. It's not as if CentOS is an obscure OS for servers or anything but I'm aware they still focus on Windows more...

What's the best way to go about emailing them? Just send them a copy of debug.log and one of the core dump files and all my server information? I think this is obviously a new problem they should be able to reproduce. It's annoying because it was working just fine before except there did seem to be an occasional crash (but no core dump/segmentation fault); whatever they did really messed it up.

The only solution I can think of for now would be, besides getting rid of CentOS 64 bit, is to set up some sort of OS virtualization program within CentOS and install Windows or something to host the server... agh...

UPDATE: It seems like a general issue with linux at the moment and not centos, and not a 32 bit or 64 bit problem. http://forums.steampowered.com/forums/showthread.php?t=794455

[ogre_x]

*sigh* Valve released a new patch today, but the problem is still there on my CentOS 4 based server

Quote:./srcds_run: line 352: 12721 Segmentation fault $HL_CMD
Add "-debug" to the ./srcds_run command line to generate a debug.log to help with solving this problem