RHEL high percentage of processor time from system 5197

Ben Lots of possibilities.

The IO system on the Fedora machine might be much faster (multiple 15,000 rpm Ultra-320 SCSI hard drives vs. a single 5400 rpm ide hard drive) than that on the RHEL system. Not likely.

The memory of the RHEL might be too small, resulting in swapping (does not look it, obviously).

When I have trouble with throughput is when I run 4 instances of BOINC projects (all compute limited) at nice 19, so all they do is soak up spare CPU cycles. Then I run a big IO-intensive database suite that computes a good bit (uses up about one of the 4 CPUs). So the database fills all the cache RAM with its stuff, gets suspended for IO, the BOINC stuff runs, contagioning the cache memories, and when the database stuff completes its IO, the cache has none of its working set in there anymore, so the database runs relatively slowly. Killing the BOINC stuff speeds up the dbms stuff. The database stuff does get a processor when it wants it; i.e., the scheduler is working correctly. But basically, the 1 Megabyte L3 cache on each of my processors is way too small for what I am doing. But it was the largest cache I could get when I bought those things

What else is going on on your systems? The system time may have nothing to do with the four processes you show. Some slow IO limited background process could be doing that. Perhaps your Internet connection is being subjected to a DOS attack and the kernel firewall is very busy with that. (Grasping at straws here.)

Have you run iostat and vmstat to see what is going on? I know you can run vmstat as normal user. I do not remember about iostat.

-- .~. Jean-David Beyer Registered Linux User 85642. V PGP-Key: 9A2FC99A Registered Machine 241939. ^^-^^ 13:45:00 up 6 days, 12:07, 4 users, load average: 4.24, 4.23, 4.25