[vampire2k]

I recently installed a fresh HL1 dedicated server for a Natural-Selection public server.
Everything works smoothly except for one annoying problem: The server does not restart itself if it happens to crash

Auto-restarting the server on crash

Console initialized.
scandir failed:/home/ublood/hl1/./platform/SAVE
Protocol version 48
Exe version 1.1.2.1/Stdio (valve)
Exe build: 18:05:13 Oct 24 2008 (4383)
STEAM Auth Server
couldn't exec language.cfg
Server IP address 78.143.28.20:27015

   Metamod version 1.19p32 Copyright (c) 2001-2006 Will Day
     Patch: Metamod-P (mm-p) v32 Copyright (c) 2004-2007 Jussi Kivilinna
   Metamod comes with ABSOLUTELY NO WARRANTY; for details type `meta gpl'.
   This is free software, and you are welcome to redistribute it
   under certain conditions; type `meta gpl' for details.


   AMX Mod X version 1.8.1.3711 Copyright (c) 2004-2006 AMX Mod X Development Team
   AMX Mod X comes with ABSOLUTELY NO WARRANTY; for details type `amxx gpl'.
   This is free software and you are welcome to redistribute it under
   certain conditions; type 'amxx gpl' for details.

scandir failed:/home/ublood/hl1/./platform/SAVE
L 05/08/2009 - 04:45:55: -------- Mapchange to co_faceoff --------
L 05/08/2009 - 04:45:55: [GEOIP] Database info: GEO-106FREE 20060901 Build 1 Copyright (c) 2006 MaxMind LLC All Rights Reserved
[COMM ICON] Error: Can not find icon "gfx/vgui/640_comfg.tga" for precache.
[S_API FAIL] SteamAPI_Init() failed; unable to update local steamclient. Continuing with current version anyway.

Adding master server 68.142.72.250:27010
Adding master server 72.165.61.189:27010
Executing AMX Mod X Configuration File
[AMXX] Loaded 32 admins from database
Connection to Steam servers successful.
   VAC secure mode is activated.

(Notice the first line)
For testing purposes, i kill the hlds_i686 thread from another terminal; This is the output on the server console:

./hlds_run: line 321: 12630 Killed                 $HL_CMD

It just freezes there.. No restart after hours :/

My hlds_run parameters (the -binary parameter does not affect it, tested it.):

./hlds_run -binary ./hlds_i686 -game ns -ip 78.143.28.20 +ip 78.143.28.20 -maxplayers 32 -pingboost 1 +map co_faceoff -timeout 6

According to the steam update utility i have the latest "'Linux Server Engine' version 47".
I am running Debian 5.0.1 (Lenny)

Any help will be greatly appreciated

[alexinno]

what is this ?
... -ip 78.143.28.20 +ip 78.143.28.20 ...

[vampire2k]

The system has multiple interfaces and i need to specify the listening ip

[YamiKaitou]

You don't need to define both. Figure out which one is used and just define that one

[vampire2k]

Sadly that still doesn't solve my problem, with or without -/+ip..

Seems noone else has this kind of problem, im thinking about bypassing hlds_run completely and write my own bash script when i find some time.

[YamiKaitou]

Add -autoupdate

[vampire2k]

-autoupdate didn't work either but i got around to write that script, if anyones interested:

#! /bin/bash
export LD_LIBRARY_PATH=".:$LD_LIBRARY_PATH"

while [ true ]; do
        ./hlds_i686 -game ns -maxplayers 32 -pingboost 1 +map co_faceoff +ip 78.143.28.20
        echo "Server crashed at '`date`' - Restarting"
        echo "Server crashed at '`date`' - Restarting" >> crash.log
        sleep 5
done

This seems to work, thread can be closed

[fysiks]

Originally Posted by vampire2k For testing purposes, i kill the hlds_i686 thread from another terminal; This is the output on the server console: ./hlds_run: line 321: 12630 Killed $HL_CMD It just freezes there.. No restart after hours :/ My hlds_run parameters (the -binary parameter does not affect it, tested it.): ./hlds_run -binary ./hlds_i686 -game ns -ip 78.143.28.20 +ip 78.143.28.20 -maxplayers 32 -pingboost 1 +map co_faceoff -timeout 6 According to the steam update utility i have the latest "'Linux Server Engine' version 47". I am running Debian 5.0.1 (Lenny) Any help will be greatly appreciated

Killing a process from another terminal does not "simulate" a crash. The server should restart on a crash not on a forced termination otherwise you wouldn't ever be able to stop the service without getting in to the console .

I would say that what you describe is working correctly.

P.S. I didn't think any HL1 game was still on Protocol 47.

[8088]

Originally Posted by fysiks Killing a process from another terminal does not "simulate" a crash.

I think it does. hlds_i686 (or hlds_i486 or hlds_amd), which is the actual server, is 'monitored' by hlds_run. As long as you don't kill hlds_run, but the server that is being monitored by it, hlds_run should act as if the server crashed.

Originally Posted by fysiks P.S. I didn't think any HL1 game was still on Protocol 47.

It's the 'Linux Server Engine' version. The protocol is indeed version 48, as can be seen in the first post.

[fysiks]

Originally Posted by 8088 I think it does. hlds_i686 (or hlds_i486 or hlds_amd), which is the actual server, is 'monitored' by hlds_run. As long as you don't kill hlds_run, but the server that is being monitored by it, hlds_run should act as if the server crashed.

Interesting, I will have to try that. I didn't realize it was hlds_run that was doing the actual restarting of the server.