sums averages in using expr, awk

[rmueller58]

all,

I am using vmstat and sar, piping output to flat files for some long term performance monitoring.
I want want to create a script that builds a running average of CPU idle, and memory use.
I will have columnar data, I want to summarize daily data files, with sums and averages using "expr", and "awk".

Basically from vmstat out put I will keep a timestamp, avg, free.

I will do something similar with "sar", capturing CPU1 idle, CPU2 idle, and timestamp.


Basically,

I want to create a running average, and a daily average

if Free is Column 2 I want to sum the column and divide by "wc -l"

If someone can show me a quicky on how to use an expr to sum my columns I can probably figure the rest out.

[RAC_1]

awk -F "sepeator" '{s+=$1/$whatever} END {print s/NR}'

Anil

[Rodney Hills]

A typical way I have summed a column in awk is-

awk 'BEGIN{s=0;n=0};{n++;s=s+$2};END{print $s/$n}' myinputfile

HTH

-- Rod Hills

[Mark Greene_1]

You can do this in ksh and use the built-in to do the math, as it is much more straight-forward I think:

(( TOTAL_CPU1 = $TOTAL_CPU1 + SAR_CPU1 ))

where you can use some looping mechanism of your choice (for, while, etc) to loop through the file or variable where you've captured the sar data to be totaled.

mark

[rmueller58]

Rod,

in the awk
'BEGIN{s=0;n=0};{n++;s=s+$2};END{print $s/$n}'

is the $2 the column of which I want to sum?

[Rodney Hills · Accepted solution]

Yes $2 is the second field.

Change $s/$n to s/n (sometimes I think perl :-)).

-- Rod Hills

[rmueller58]

Thanks All.

Did it!!