1.21 Server Lockup Problem

[Drek]

Last night my Natural Selection server locked up, and version 1.21 of ServerLoader detected the lockup, but when it tried to shut HLDS down it failed. I found this a bit unusual because I was able to shut down HLDS by hitting the "X" with no problem. Maybe ServerLoader needs to change the way it tries to shut down unresponsive applications? I have a Dr. Watson log of the crash, regrettably I deleted the ServerLoader log entries without thinking (although all they said was basically what I said here). If you want a copy of the Dr. Watson log, email me at olstairsatnospamrogers.com (take out the nospam).

[botmeister]

Too bad about deleting the serverloader log, there probably was some useful information in it.

I sent you an email requesting the Dr Watson log.

I think you are right and I'll have to use alternate methods of shuting down the app. I figure I could simply have it try method 1 and if that fails, method 2, etc, in increasing desparation so to speak.

[Drek]

I think you are right and I'll have to use alternate methods of shuting down the app. I figure I could simply have it try method 1 and if that fails, method 2, etc, in increasing desparation so to speak.

This sounds like an excellent idea.

[Drek]

Just thought I'd add. My DOD server that runs SturmBot has a tendency to lock up once every six or seven hours. ServerLoader (1.21) detects the frozen server and restarts it every time. I did have some problems with lockups, or endless loops on my DOD server that ServerLoader wasn't able to detect or restart successfully (sorry, don't know which), but I've changed my configuration somewhat and now ServerLoader is working perfectly.

The configuration problem was based on some bad advice I recieved a long time ago. I was told that adding the "-zone 8192" switch to the server command line would improve performance, but just recently I discovered that several problems I have been putting up with for the last three or four years were caused by that switch... doh!

[botmeister]

Sounds like its working pretty good then. There's a plan to add a check for excessive CPU usage as a way of detecting a crashed program that is looping forever. I'll get to it once I have the time, however I don't know what's involved yet, or even if it can be done. I figure it can be done since the taskmanager is able to show a CPU usage.

[Drek]

So many ideas, so little time - the story of my life...