How to change my script to another so that all of it is done via awk?
 

This code works correctly to sum column $2 and $3 where $1=IP:
I want to improve it by deleting grep, and doing all of that via awk, is there anyway to do that!?how?

I have already told you: http://www.linuxquestions.org/questi...9/#post5163847.
Have you tried that solution?

yes, but it didn't work correctly! It shows the last line of IPvar repeatedly

This script is a test script, so I use awk in it;
It is an example of my DayReport-May-04

192.168.40.29 0% 1265.756
192.168.40.1 0% 693.506
192.168.40.4 10% 1112.107
192.168.40.1 10% 754.015
192.168.40.4 0% 635.516
192.168.40.29 0% 538.657
192.168.40.22 0% 640.583
192.168.40.4 0% 615.745
192.168.40.1 0% 507.429
...

--------------------------------------------------------------------------------
and I have some IP addresses in IPvar that is read line by line via while loop:
192.168.40.1
192.168.40.27
192.168.40.5
192.168.40.29
...
---------------------------------------------------------------------------------
I need to sum of amounts of column 2 and 3 for each IP, for example, the output for above files should be:

192.168.40.1 20% 1954.95
...
192.168.40.29 0% 1804.413
---------------------------------------------------------------------------------
I've tried this code via awk:
It just work for the last record of DayReport-May-04!!!!

/IP/ means awk will take IP as string, not as variable, therefore it will look for the text IP in your input file.
you can try the following instead:
but even better it can be solved without that while cycle ...

Simply read in the file with all of the addresses in it and then cycle through the second file creating your totals. You will then need to utilise the 'END' action within awk to display the data in the format you require

This is a case where I would replace some logic with, say, "a very small Perl script." Or, in this case, with "awk itself (you don't need a separate script at all)."

"You have some rather funky magic going on" in the present (convoluted ...) script, and, although it might (or might not) be argued that "well, it works ...", such things are so hard to maintain (and, as you now see, so hard to debug!) that I routinely replace them, and/or ask a member of my team to do so. (Yes, I take my turn at "sweeping out the garbage," when I'm not wearing my management-hat and thus sweeping in more garbage. ;) )

You see, what this script in-aggregate does is to read a file of "IP-addresses and other things," and to extract certain variables out of it to produce certain results. But does it, cleanly and with just one tool, do that? Uhh, no. It uses two tools clumsily piped together, and neither of the two bits of source-code provided to either of the two tools completely expresses the total of what you actually want to accomplish.

First of all, a short (but true) awk script could quite-effortlessly do this: it just requires, well, "a separate file instead of a one-liner." You simply invoke the awk command giving it the name of the IP-address file and the name of a short but separate awk script. All of the necessary logic now sits within that script; not in whatever invokes awk.

Perl is awk's big-brother, so to speak, in that it literally was conceived as "an extension of awk" in its early days and still bears noticeable resemblances to it. Aptly called "the Swiss Army® Knife of data processing," it's a tool that I use quite often. But once again, never as a one-liner in production settings.

The output of your code is like as the output of my script! showing the last IP in Day-Report-May-04 repeatedly!

I've done it via grep myself! I want to do all of that via awk!!!

to #10: yes, sure, did you put it into that while cycle from port #5? (you just need to replace the line containing the awk command).
to #11: probably you can simply drop that grep and use only the awk by ntubski

yes I've putted it there!