sar -M & top command

[Mariko.N]

Hello!

I'd like to know difference between 'top' & 'sar -M' command.

When I execute 'top' & 'sar -M 1 1' simultaneously to check CPU status, the results seem something different.
(top: USER SYS IDLE, sar: %usr %sys %idle)

I'd like to know the reason why.
Thank you.

Mariko

[Mark Grant]

Possibly down to the sampling time of the two tools. If you run "sar -M 1 5", presuming "top" is sampling every five seconds (the default) then I would think the results would be more similar. Remember, the figures are an average over the time period.

[Simon Hargrave · Accepted solution]

try sar -M 5 100

top, by default, is taking a sample over 5 seconds, so your sar over 1 second is effectively looking at a smaller set of data. If you sar based on 5 seconds (or run top with -s to set a different sample rate), you'll probably have similar results.

You'll probably only get "very similar" values if you start sam at the same time as top, to ensure they're sampling the same 5 seconds, too.

[Sanjay Kumar Suri]

man top

Set the delay between screen updates to time seconds. The default delay between updates is 5 seconds

If you use following command:

sar -M 5 5

then the output from sar & top may match.

sks

[Mark Grant]

woops, I did get my "sar" arguments the wrong way around didn' I! I keep doing that. Shoud, of course, be "sar -M 5 1" though porbably best to change that "1" for a higher figure for more accurate results.