Java app makes screen display unresponsive after 10 minutes of user idle time

问题

I've written a Java app that allows users to script mouse/keyboard input (JMacro, link not important, only for the curious). I personally use the application to automate character actions in an online game overnight while I sleep. Unfortunately, I keep coming back to the computer in the morning to find it unresponsive. Upon further testing, I'm finding that my application causes the computer to become unresponsive after about 10 minutes of user idle time (even if the application itself it simulating user activity). I can't seem to pin-point the issue, so I'm hoping somebody else might have a suggestion of where to look or what might be causing the issue.

The relevant symptoms and characteristics:

• Unresponsiveness occurs after user is idle for 10 minutes
• User can still move the mouse pointer around the screen
• Everything but the mouse appears frozen... mouse clicks have no effect and no applications update their displays, including the Windows 7 desktop
• I left the task manager up along the with the app overnight so I could see the last task manager image before the screen freezes... the Java app is at normal CPU/Memory usage and total CPU usage is only ~1%
• After moving the mouse (in other words, the user comes back from being idle), the screen image starts updating again within 30 minutes (this is very hit and miss... sometimes 10 minutes, sometimes no results after two hours)
• User can CTRL-ALT-DEL to get to Windows 7's CTRL-ALT-DEL screen (after a 30 second pause). User is still able to move mouse pointer, but clicking any of the button options causes the screen to appear to freeze again
• On some very rare occasions, the system never freezes, and I come back to it in the morning with full responsiveness
• The Java app automatically stops input scripting in the middle of the night, so Windows 7 detects "real" idleness and turns the monitors into Standby mode... which they successfully come out of upon manually moving the mouse in the morning when I wake up, even though the desktop display still appears frozen

Given the symptoms and characteristics of the issue, it's as if the Java app is causing the desktop display of the logged in user to stop updating, including any running applications.

Programming concepts and Java packages used:

• Multi-threading
• Standard out and err are rerouted to a javax.swing.JTextArea
• The application uses a Swing GUI
• awt.Robot (very heavily used)
• awt.PointerInfo
• awt.MouseInfo

System Specs:

• Windows 7 Professional
• Java 1.6.0 u17

In conclusion, I should stress that I'm not looking for any specific solutions, as I'm not asking a very specific question. I'm just wondering if anybody has run into a similar problem when using the Java libraries that I'm using. I would also gladly appreciate any suggestions for things to try to attempt to further pinpoint what is causing my problem.

Thanks!

Ross

PS, I'll post an update/answer if I manage to stumble across anything else while I continue to debug this.

Update: my app involved multi-threaded processes each initializing their own Robot objects and creating input events asynchronously. I refactored the app to only contain one Robot singleton object, but the different processes still asynchronously invoke input commands. As far as I can tell, this did not change the behavior of my app. My next step might be to created a synchronized wrapper around the Robot singleton to see if that helps, but given the symptoms, I don't know why it would.

回答1:

I've had problems using the Robot class before. I forget exactly what I've done, but it has caused the computer to lock up and I've been forced to reboot.

回答2:

I'm not familiar with the vagaries of Robot, but Uncaught exceptions in GUI applications can produce very odd results as the event dispatch thread dies and restarts. You might get some ideas from How uncaught exceptions are handled.

回答3:

This can happen if u activated any screen saver or some thing like that, then this robot actions will stop working

i got this problem in the following way

i am having some GUI based applications and i written some test code based on Robot class.

but if i activated screen saver in my system this test cases stopped working...

please check any such scenarios are there in your case

回答4:

for Ross, I am pretty sure that you have a policy problem which mean that you don't have permission and getting block by win7. To get this permission you must create a policy and that is a big of a story. Check what info you can get from sun website.

For the others if your program stops working after some period of time it might be that you are not handling all of the exception in your program right. Try to get the line when it stops: you can do it by just sending a line of text to the system console, and rewrite it to get the performance you want.

For the rest check your code and check it good very good to see that you are not getting into a dead lock which might stuck your pc if it has something with it.

Also check to see the CPU usage and see if your program is overloading it if it does your CPU temperature control will restart or turn off your PC automatically.

If i didn't hit the problem, let me know; if you solved the case please let me know what you did.

回答5:

We have two machines with almost exactly same behavior with Java applications.

One is with Windows 7 64 bit, where Eclipse 64bit and Java 6 (64bit) is causing exactly same freeze.

Other is with Windows 7 32 bit and Java application utilizing a lot of CPU and disk activity causing freeze.

Both machines are notebooks from Toshiba with modern CPU's (Core 2 Duo).

Also both machines have NOD32 anti virus installed.

Standard apps (Office, Skype, Firefox,...)

回答6:

We seem to tracked down problem to NOD32. When we disable NOD32 it seems that beahaviour does not occure anymore.

I want to point out that I don't think it's NOD32 itself, but some combination of Windows 7, Java JDK and NOD32.

回答7:

I ran into a similar issue on Mac OS X 10.6.7 but it was not freezing entire system but the entire java process where my application was running in, and interesting at random basis. Workaround for this was call:

Toolkit.getDefaultToolkit();

Before create any Robot, in example:

    public static void main(String[] args) {    
        Toolkit.getDefaultToolkit();
        //Blah blah
    }

Same call is made in init method at Robot class so seems where the problems are coming from, this is stupid and does not make a lot sense to me but works perfectly now :)

来源：https://stackoverflow.com/questions/2233529/java-app-makes-screen-display-unresponsive-after-10-minutes-of-user-idle-time