YEOAK@NUSDISCS.BITNET (12/19/89)
Re: vmstat and DSP90
I've collected the following statistics from our DSP90 (3MB RAM)
running SR10.1 with default SYSTYPE=4.3:
===> /systest/ssr*/vmstat 2 5
procs memory page disk net faults cpu
r b w avm fre re pi po fr de sr dr dw nr nt da ex cs us sy id
------ -------- --------------------- ----- ----- ---------- --------
1 0 0 2222 0 0 6 0 0 0 0 0 0 0 0 0 2 0 95 0 5
1 0 0 2222 0 0 4 0 0 0 0 2 0 0 0 0 0 0100 0 0
1 0 0 2222 0 0 0 0 0 0 0 4 0 0 0 0 0 0 97 1 2
1 0 0 2222 0 0 0 0 0 0 0 0 0 0 0 0 0 0100 0 0
1 0 0 2222 0 0 0 0 0 0 0 0 0 0 0 0 0 0 97 0 3
Notice that the CPU user processes times (us under cpu) is strangely high,
I wonder anybody has tried using vmstat on DSP90 and got this kind of results.
I am puzzled by the high cpu-us time b'cos I was the only log-in user
thru' the terminal connected to DSP90. I didn't run any program except
/systest/ssr*/vmstat. The similar kind of statistics on DSP90
can be collected anytime using 'vmstat'.
I ran 'vmstat' on our DN580 series, the statistics showed a low cpu-us but
high cpu-id time which is very reasonable since nobody is doing anything
on those machines. Can anyone tell me wether 'vmstat' is reliable or
SR10.1 on DSP90 is doing something very different when it is on DN580?
The following ps might be helpful:
PID TTY STAT TIME COMMAND
1 ? S < 0:32 /etc/init
2 ? R 2:13 null
3 ? S 1:41 purifier
4 ? S 0:05 purifier
5 ? S 0:53 unwired_dxm
6 ? S 0:00 pinger
7 ? S 0:06 netreceive
8 ? S 11:15 netpaging
9 ? S 2:06 wired_dxm
10 ? S 6:22 netrequest
85 ? S 25:34 /etc/tcpd
90 ? S 1:59 /etc/routed -h -f
93 ? S 0:02 /etc/inetd
96 ? S 0:01 /etc/ncs/llbd
99 ? S < 7:44 /etc/rgyd
104 ? S 0:00 /sys/net/netman
106 sio1 R 3011:15 spm
108 sio1 S 0:00 /sys/mbx/mbx_helper
121 tty01 S 0:07 -ksh
272 tty01 R 0:00 ps ax
Thanks in advance.
--AnnKian Yeo
Email: YEOAK%NUSDISCS.BITNET@CUNYVM.CUNY.EDU
Department of Information Systems and Computer Science (DISCS)
National University of Singapore (NUS)