Arch Linux

esh

Member

Registered: 2008-11-05

Posts: 28

Question on behaviour of 'nice' command

Hello, I just noticed something a little curious which I don't fully understand, and was wondering if it was 'expected' behaviour.

This machine has 2x Xeon E5420s, so Linux registers a total of 8 CPUs. It's mostly used for heavy job loads.

I have gkrellm which monitors things in the corner and 'ts' which just stacks up batches of jobs. We stick all of these with a 'nice' prefix. Today I ran a several minute job on it that was not 'nice'd while all 8 cores were working on 'nice'd jobs. The new job got 100% of one core and 3 cores became idle. When the job finished, all four cores on that CPU begin running again.

Any suggestions on this or am I being stupid?

Thanks.