Shell Programming and Scripting

Hi Guys,

I have this task to monitor a linux box. I found a program that displays the parameters that I want and I wrote a little .sh to run that program and record output into a file.

The findings look promising but I would like to graph them.

My output (for every iteration) looks like this:

Code:

blablabla (some guy bragging about his program and version)

05/16/16 16:50:58
Name            Param1  Param2 
one                     1.0        1.1
two                     2.0        2.1

I want, instead of my .sh making this program dump the output in the raw form, to take just the timestamp and the numbers, resulting in something like this:

Code:

16:50:58,1.0,1.1,2.0,2.1

(then, I could run this every minute for 2-3 hours and have a csv that I could graph)

Linux shell scripting is really out of my career scope at the moment.

What do I use here? gawk/sed? Maybe a clever grep?
If you could please just show me how to do it for one value, I will figure out the rest.

Moderator's Comments:

Please use code tags as required by forum rules!

That specification is a bit sparse. Does the file consist of a time stamp, a header line, and two parameter lines always? Does it contain these records only? Always the same structure?

To be exact, the output is such:

Code:

Linux 3.12.28-4-default (machine-name)     05/16/16     _x86_64_    (4 CPU)

05/16/16 16:50:58
Device:            tps    kB_read/s    kB_wrtn/s    kB_read    kB_wrtn
scd0              0.00         0.00         0.00       1598          0
sdb               0.46         0.31        11.56     468944   17228724
sda               0.82         0.25        27.99     379696   41709248

It's always the same structure and the Devices are always the same

Moderator's Comments:

Please use code tags as required by forum rules! Seriously!

In absence of some adequate logics, try

Code:

awk 'NF == 2 {if (L) printf RS; printf "%s", $2; L = 1; next} /Name/ {next}; {printf ",%s,%s", $2, $3} END {printf RS}' file
16:50:58,1.0,1.1,2.0,2.1