Ubuntu system running under full load doesn't respond to SSH or keyboard input

[Chaotic42]

I've got a quad-core system running a very intensive POV-Ray render. It's been running for a couple of days. This afternoon I went to check on it and it does not respond to keyboard inputs when I try to wake up the displays. It also refuses to acknowledge SSH requests, but it does respond to pings.

Is there any chance this thing is still alive? Any hope of reaching it? I'd like to let it keep going if it's still rendering so that I don't lose the render. I'm anticipating the render will take about a week or so if it's really running.

Any thoughts or ideas? Thanks.

 

[KillerBee]

If it's running linux can you get a console alt-F1?

nm - reread no keyboard input

 

[thecoolnessrune]

When I see this happen on Linux servers we manage, it's usually because a customer's application is using all available memory on the host, and starts hanging key system tasks as a result. You might check on that once you get your system running again.

 

[you2]

There are a couple of possibilities; you could be swapping (is the disk light flashing). If it reached an out of memory condition it might have killed critical (sshd for example) processes. What happens if you telnet to port 22 - does it hang or does ssh respond ?
-
If the system is swapping heavilty it will eventually come back. Also i know that a few of the older kenerls have bugs that can cause them to get trapped in the kernel but from your above description i would suspect memory issues from the few details you provided.

 

[Red Squirrel]

Do you have any DNS issues, ex: is DNS running on that same machine and under too much load to respond to requests?

I found that in some odd situations, if DNS is not working, you can't ssh anywhere. It will just hang.

Does the render process generate any kind of live log file? Might be worth setting up some kind of process that will email you a report say every hour. Just a "tail" command of the log of something so you know it's still doing something.

 

[jhu]

I bet your computer is running out of memory and is swapping. What are you rendering? You may want to consider using instanced mesh objects since they use significantly less memory than regular CSG objects.