# uname -a
HP-UX inap01 B.11.11 U 9000/800 1991039449 unlimited-user license
Please look at the rds process CPU values or weighted CPU values;
# top
----i ignored these lines -------------------
CPU TTY PID USERNAME PRI NI SIZE RES STATE TIME %WCPU %CPU COMMAND
1 ? 15059 root 152 20 52828K 33756K run 723:21 93.22 93.06 rds
1 ? 22171 root 180 20 798M 3892K sleep 11353:26 33.53 33.47 oninit
0 ? 22160 root 180 20 798M 6432K sleep 3603:08 11.48 11.46 oninit
0 ? 7161 nwauto 154 20 9996K 5916K sleep 13:03 0.80 0.80 perl
and when I run glance the value of CPU looks like:
Process Name PID PPID Pri Name ( CPU util 200% max) Cum CPU IO Rate RSS Cnt
rds 15059 1 168 root 0.0/ 1.3 43364.0 0.0/ 0.6 34.2mb 11
so CPU usage is totaly different;
top says:
%CPU : Raw CPU percentage. This field is used to
sort the top processes.
glance say about CPU usage is:
The total CPU time consumed by a process or kernel thread as a percentage of the total CPU time available during the interval.
OK then, is glance working wrong ? or top %CPU is a elapsed CPU time since process started in spite of TOP does not explain that ??
top has a very limited view of things compared to glance. I would tend to believe glance over top until proven otherwise. My guess is that rds is indeed running but it is doing something like a spinlock. glance would know this and would consider the spinlock to be a global cost. If it is something like this, then top and glance would closely agree on how much cpu is being used, and glance would list the missing 90% of a cpu to something else.