Add fio-parse-jsons.py tool to parse a set of json FIO files.

[perezjosibm]

This is the separate commit only for the new standalone tool as described in #306.

New standalone script fio-parse-jsons.py

This script processes a list of FIO JSON output files and produces as output:

• a gnuplot script,
• a .dat file with the data to plot,
• a summary table of FIO results in wiki format, printed to stdout.

This is especially useful to produce a response graph from a set of executions ranging the number of FIO jobs and the iodepth values. The script was written before knowledge of CBT was gained, so in a way is independent of the script driving the tests. A future PR would integrate the functionality of this standalone script with that of CBT.

The following is an example of the execution of the script:

 # python3 /cbt/tools/fio-parse-jsons.py -c crimson200gb_1procs_randwrite_list -t 'Crimson 200GB RBD 4k rw' -a crimson4cores_200gb_1img_4k_1procs_randwrite_avg.json

the arguments are:

• -c config_file: a txt file containing the list of FIO output JSON file to process,
• -t title: the string ot use as title for the gnuplot chart,
• -a cpu_avg.json: a .json file containing the avg CPU utilisation, normally produced by the script parse-top.pl.

The following are the .dat and gnuplot files produced:

 crimson200gb_1procs_randwrite.dat
 crimson200gb_1procs_randwrite.plot

To produce the chart, simply execute

gnuplot classic200gb_1procs_randwrite.plot

the IOPs vs latency chart result is shown below:

[perezjosibm]

Hi @sseshasa, here is the separate commit as requested, thanks!

[sseshasa]

Just added a couple of comments related to the documentation. I just took a cursory look at the tool and in general it looks good to me.

[perezjosibm]

Hi @sseshasa, I've updated the PR as per your comments, would you mind to merge this PR (since this being my first one I don't have write permissions yet) Thanks!

[sseshasa]

Hi @sseshasa, I've updated the PR as per your comments, would you mind to merge this PR (since this being my first one I don't have write permissions yet) Thanks!

@perezjosibm Sure! Just FYI, when a change is requested on a commit, it is better to modify the same commit instead of adding a new one. You can make the requested changes and either rebase it on the existing commit itself or use "git commit --amend" to update the last commit like in this case. This helps in maintaining a good clean git history.

The above is something to keep in mind for next time. I will go ahead merge this.

[perezjosibm]

Many thanks @sseshasa, yes I was wondering about that, I'll amend each time I got a new commit then 👍 Cheers