As always, I was joking. I laughed. Of course the decision isn't very beautiful, but it's working now. To the pointReason:The flow of awesomium_process (the flow of awesomium) consumes more and more memory (and the movement itself should not be blamed, some sites are all working well, while others eat a certain amount of memory (1.5gb in 10 minutes). The slide of the x86, when reaching the target of awesomium_process, is crumbling (but not staining the form of the application itself), followed by hanging without any apparent error.Decision(s):It's a simple program, a launcher, where the browser applications are launched. In the lamb, we need to put a flag of events from the process and sign into the Exit event:Process.EnableRaisingEvents = true;
Process.Exited += Process_Exited;
And if there's an event, reset the annex browser. In the very annex browser, I did something like watchdog-timer that checks the performance of the browser. And if, for example, 10 seconds the timer has not been abandoned, the annex will cut itself out.If someone encounters the same problem: of course the launcher works a little harder if, for whatever reason, the annex cannot simply be taken and cut (some atomary action takes place), then somehow he needs to report that it turns out as soon as it can be done without logical mistakes. I used a simple WCF interface:[ServiceContract]
public interface ICloser
{
[OperationContract]
void Shutdown();
}
The whole system now works 24/7 without error.