[Invalid] machine sudden reboot when jme sdk has run my appplication some times?

[EDIT: I must say it was an unfortunate coincidence, as the machine sudden boot up in a session that I had not even opened JME SDK, neither ran anything created by it, sorry.]

I am trying to track where is the problem.
Could be my machine?

It have been happening lately only while jme sdk is running.
Seems to only happen also after I have run my application some times (and forcefully closed it most of these times).

I’ve researched, it could be my memory sticks, my gfx board, even my internet connection, and so on…;

I am on linux ubuntu 14.04, may be upgrading the system could help? I am postponing this as I dont like OS surprises and it is the LTS release…

I use Compiz/Unity with max effect woobbly and so on.

I try to run jme sdk on “X :1” also, so wont be under compiz power, I thought it was helping but it just rebooted.

My memory gets almost full when I run both chromium and jme sdk, it swaps a bit.

I wonder if I should try to exit my application more gracefully? so assumption: “it doesnt leave something on the gfx card that could be the problem”?

Is there some thing I could run after my app crashes or I forcefully exit to put the machine in a balanced state again (if that is the problem of course)?

May be some tweaks to the way I run the jme sdk could help on it?

What graphics card are you using and what driver? Do you have the “Use Canvas” option checked in the SDK options?

Also, is there anything in the syslog about the crash?

an old NVidia GeForce GTS 250 with 512MB, but I can run many games with it for hours

I am still trying to find, where is the “use canvas” option?

I just read the full syslog around the reboot time, I saw nothing special, but I am not sure what kind of words I should care; I tried look for “critical” “crash” but nothing special shown up, any tip on this? I just found this:
“sp5100_tco: Last reboot was not triggered by watchdog.”
but doesnt seem special, not sure tho.

btw, since that time til now, I had not crash “yet”.
and… I was exiting the application uncleany, I was not using jme Application.stop(true) (that I am using since then, just a bit after I created this thread); there were many application crashes by my fault (testing code), what could lead to unclean exit (I guess), but all other exits where clean since then.
I think I will trap the whole code and force it clean exit!

the “almost whole code” trap may help, lets see:


@Override
public void simpleInitApp() {
  try {
    initApp();
  }catch(Exception e){
    e.printStackTrace();
    stop(false);
  }
}

stop(true) would not work btw

I did this also for the simpleUpdate()

@teique said: the "almost whole code" trap may help, lets see:

@Override
public void simpleInitApp() {
  try {
    initApp();
  }catch(Exception e){
    e.printStackTrace();
    stop(false);
  }
}

stop(true) would not work btw

I did this also for the simpleUpdate()

stop(true) is not supposed to work, it will cause a deadlock by definition.

Is this the code that causes the freeze?

Which driver are you using? nouveau or nvidia proprietary? and what version?

@Momoko_Fan said: stop(true) is not supposed to work, it will cause a deadlock by definition.

Is this the code that causes the freeze?

Which driver are you using? nouveau or nvidia proprietary? and what version?

no no, that is the point; after I begin using the sdk, my machine reboots suddently and randomly, I am not being able to predict when it will happen, but it has been happening while the sdk IDE is opened, and I think it havent happened while my application was running; it happens usually when I am browsing the internet and leave the sdk IDE alone for some time like 30min to 1hour. I think it may be something the sdk IDE does that conflicts with my OS gfx (old?) drivers, as it has some 3D previews built-in (but I havent used them yet).

btw, the stop(true) is working only I exit normally by pressing a key, wont work on exceptions tho.

nvidia 331.38

ps.: I am still waiting the next reboot, as the logs are too hard to read, so I will shutdown the machine and wait like 30min so I can find the proper part of the logs more easily :frowning: