Jump to content

[SOLVED] very hard to track down crash.


photo

Recommended Posts

Posted

This is for an onsite system. After a few hours of operation (varies between say 4 and 20) we get an application crash where non allocated memory is being accessed (c00...005 in windows).

We suspect something to do with syncker and the network code since we don't get this crash running in a non networked environment. The log file says nothing and it is windows that is killing off ungine - how do we even begin to debug this?

Posted

If not already done use main_x86/64.exe for running the application as this might give some additional error/debug information e.g. stack overflow/invalid array index etc. for tracking down the actual cause of the crash. Even better idea might be to run this version from within debugger so you can intercept the exception for getting a hint on error 'location' by inspecting the call stack.  

Posted

Yes, of course the debug build with the - in my post missing - little additional 'd'

Posted

BTW as you are expecting the problem within network/syncher code section: it might be worth to try to somehow dramatically increase the network communication frequency ("time-lapse") of your application, so you don't have to wait a whole day for the problem to show up, but just some minutes or hours.  

Posted

Thanks ulf... I will think about that, we are doing a pretty good job of saturating the network as it is.

The d version isn't giving any hints and the application is being killed from the windows side. 

 

Thanks binstream. I will see about getting my hands on a copy of Visual Studio...

Posted

I've got one of this kind of problem, and I successfully solved this problem, it's a problem of small class in your script.

 

Try to comment each class of your script to narrow down the problem. and I don't think the debug version of engine will do any help on this.

 

I've created a date class in script. and got exact same error like yours, so double check your class in script.

×
×
  • Create New...