it appears that red hat 5.0 uses 1 cpu per vnc user. 1 cpu or core will be at %100 while another is at %0
red hat 6.2 appears to try and spread out the load evenly among the cpus. I think this may be at the heart of random memory allocation errors.
Any thoughts on this or do you know how I can change it to work like red hat 5.0?
I had the same problem on fedora 16