quartus_fit enters sleep state on Linux machines

[Altera_Forum]

I'm observing an issue on Linux machines where quartus_fit process goes back and forth between Running and Sleep states. When this happens, the build is not progressing, and eventually I need to kill the process.  

 This happens intermittently on different Linux distributions (CentOS 5 and custom-built), and at least two Quartus revisions.  There is sufficient amount of available memory, swap, and CPU in the system.  It happens when I run one or several builds concurrently.  I'm inclined to think that it's something to do with the server itself (but I've seen it on at least 3 very different servers). But I cannot rule out some obscure Quartus bug.   Any ideas why this happens are appreciated.  Thanks, Evgeni

[Altera_Forum]

that sounds really strange, i haven't seen or heard of that sort of behavior

[Altera_Forum]

Those are quite "heavy" designs that take 8-12 hours to build, require at least 12GB of memory and use 4 CPUs. I'd say that in 90% of builds this problem doesn't happen.

[Altera_Forum]

i'm not working on anything that big at the moment

[Altera_Forum]

A quick update.  

 Those builds that were entering sleep state have actually completed. I let them run, and it took several times longer than usual. Instead of 8-12 hours, it took between 3 to 4 days.   Opened an SR, and will see what Altera has to say about it.

[Altera_Forum]

unbelievable!

[Altera_Forum]

Another update.  

 After going back-and-forth with Altera for a couple of months, the conclusion is that the root cause of this problem is high utilization + high routing congestion of this design.  Hope Quartus 12.x will make things better.