Summary

I built this set of programs since I run jmeter almost always in non-interactive mode and the measures computed by jmeter itself were insufficient for my tests.

• Completely dependent on jmeter.log
• Has several metrics to analyze the performance of jmeter itself and "client-end" measures:
  1. requests per minute
  2. average response time
  3. average response time across last X% of requests
  4. average response time across last X% of test duration
  5. mean time between consecutive requests (a nice way to know when things are slowing down)
  6. number of waiting requests (request sent but not received response yet)
  7. grouping response times based on URL regex matching
• XML of results can be seen at anytime using a simple command: telnet <host_running_jmserver> <port>
• Results aggregation across hosts running jmeter
• Dynamic/near-realt-time charting of results using a gnuplot script created dynamically
  

Dependencies

• XML::Simple
• Log::Log4perl
• Data::Dumper
• DateTime
• DateTime::Format::Strptime
• List::Util
• Gnuplot
  

Contents

jmeter.pm

Perl package containing the core class definition and algorithms for metrics computation

Gnuplot.pm

A perl wrapper for Gnuplot to automatically generate a gnuplot script from a perl hash.

jmserver.pl

A server program that invokes two threads: one that reads jmeter.log and computes all kinds of metrics; another that is listening to a socket and when a client attempts to read from the socket, spews out an XML containing the results.

jmclient.pl

A client program that reads XML containing measures from one or more servers, aggregates them, dynamically generates a gnuplot script, and runs gnuplot to generate charts.

Usage

Make sure you install all dependencies. I run jmeter on linux machines – so this package may not be platform independent. But its all written in perl, so porting to other OS's should be easy once you know perl.

There are some additional monitoring programs that I invoke as part of jmclient.pl. They should be commented out. Read the source code before using it.

On the host that is going to run jmeter:

1. cd <dir in which jmeter.log will be written>
2. rm jmeter.log; touch jmeter.log
3. tail -100f jmeter.log | perl /path/to/jmserver.pl
   

On the client that is going to aggregate and visualize charts (i strongly recommend this is a different machine)
1.#4 cd </path/to/results/directory>

1. mkdir -p <test_name>/figs
2. cd <test_name>
3. perl /path/to/jmclient.pl
   

There will be some files generated in the dir:

• stats.dat contains a space delimited text file containing all data which will be used by gnuplot
• figs/ contains all .png's containing metrics
• .jmclient.run which contains the time at which the test started. this is useful if you kill the test and restart, but want the original stats to be included while charting the new measures
• gnuplot is a dynamically generated gnuplot script. this script is generated once per execution of jmclient.pl. For every new stats XML sourced from the servers running jmserver.pl, this is run once.

Once again, make sure you read the source in all files before using. The program will not compile until you install all the dependencies and may have runtime failures too; but they should be easy to fix if you know a little perl or any other language.