[Solved...?] L4D2 - Server Crashes in Non-Coop Mode

[ThaiGrocer]

OS: Ubuntu Desktop 32-Bit 10.04
Processor: Pentium 4 2.66 ghz with hyperthread
RAM: 512 MB RDRAM
Game: Left 4 Dead 2
Start Up Command: ./srcds_run -console -game "left4dead2" -maxplayers 8 -autoupdate +ip 192.168.1.104 -port 27016 +map c1m1_hotel
Admin Mods: None

I can create and finish a coop game on my server via sv_search_key and lobby, but whenever I try to connect to a non-coop gamemode like survival, versus, mutation, etc., my server crashes and restarts accordingly. I don't think it's from complicated server.cfg settings, because I tried reducing down to something as simple as an rcon password, hostname, and sv_searchkey and still had non-coop lobbies crash the server. I don't have any server mods installed. All I have done so far is follow the srcds linux tutorial from srcds.com with help from the forums on how to install L4D2 specifically and server.cfg.

Crash info in debug mode:
-> Reservation cookie 1840725eaf53800: reason ReplyReservationRequest
Alarm clock
cat: hlds.22250.pid: No such file or directory
email debug.log to linux@valvesoftware.com
Sun May 16 05:21:11 CDT 2010: Server restart in 10 seconds
(debug.log only contains a note that the crash happened with no additional info)

Friends can connect and play in coop mode from or after lobby without any issues.

I just started playing around with srcds yesterday, so I apologize for my new-ness.

Thanks!

[ThaiGrocer]

I would like to thank Joonas from the srcds_linux mailing list for identifying what "Alarm clock" is. I went to go search for it on google (srcds SIGALRM) and I found this: http://noteand.blog.eonet.jp/weblog/2009/10/l4d2-demo--da30.html

Quote:
"* The linux dedicated server supports a watchdog timer functionality,
which is enabled by default. The intent of this timer is to make
anything which hangs the server, either due to unknown bugs or
misconfiguration, cause an abort, so that the server may restart or
be debugged. In the case of a forked server, this will cause a new
subprocess to start to replace the crashed one. On a non-forked
server, this can be used in conjunction with an auto-restart script
in order to increase server availability. If you see your server
dying with SIGALRM (signal 14), it means that this has triggered
because of either a server frame taking longer than 5 seconds of
wall time, or a map load taking too long. If this code causes
trouble for you, it can be disabled via giving the "-nowatchdog"
option on the command line."

I put the -nowatchdog switch in srcds_run and switching to something like realism, mutation, etc. from the lobby with sv_search_key WORKS now! I think reliability might be an issue, so now I have to test out how much I can push out of my old linux box.

Thanks a lot for all your help, guys!