<div class="gmail_quote"><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">could you perhaps paste relevant pieces of check_proc using -vvv? just out of interest, how did you install nagios-plugins on this host? package or compiled? </blockquote></div> Hmm, that's interesting - i didn't know about '-vvv' - lots of useful info there. Nagios-plugins on this host was compiled. By this time, our offending process (a jvm) is now behaving normally and has stopped consuming 100% constantly. But when we stress the server manually, if I watch 'top', i can see % CPU for that process go above 60% for several seconds (ie. if i hit 'space bar' for 2-3 seconds, it's pinned at 60% for a few seconds, then it goes back to normal {ie. < 1%}). But if i constantly run 'check_procs' (as a dummy test, i threw it in a loop and had it run 100 times with no sleep and log the results), it still doesn't view the java process a s above 0%... If i try exactly the same thing (running check_procs 100 times) with a non-jvm process (ie. i used '-C top' and then hit space bar to push top above 20% usage), check_procs reports properly. Any idea why I'd be getting this descrepancy?