GenLog version 0.0
(this document taken from http://www.organix.tv/research/genlog/)

Realistic* automated generation of extended web log entries
* Almost

genlog is a program for generating somewhat realistic (and tunable) log entries. I use it for testing a web usage
mining tool ... You can do whatever you like with it.

genlog is GPL'd, for your pleasure and mine.

With Apache you can create real extended log entries with the following directives in httpd.conf:

LogFormat "%h %l %u %t \"%r\" %>s %b \"%{Referer}i\" \"%{User-Agent}i\"" extended
CustomLog logs/extended_log extended
	

With genlog you can simulate a webserver responding to requests from an average load of 100 users at a time making
requests at 3000 requests per minute for half an hour to the website at http://www.cs.rpi.edu/~moorthy as:

> genlog --graph=./moorthy.gr --run-for-seconds=1800 --requests-per-min=3000 --num-users=100
	

Voila, logs are generated as if being created by a web server.


Caveats

* If you want 3000 requests per minute, you'll get less than that. genlog calculates how long to sleep between
requests based on the requests/min spec, but doesnt account for how long it takes to generate an entry. For me on a
2Ghz p4 with loads of memory it takes about .02 seconds to generate a single log entry.
* You need a website for users to traverse. Its a beautiful idea it seems, that you can create a shell of a website
and put little fake users on it, but creating the shells of websites is not normal practice for most folks. You can
download Dr. Mukkai Krishnamoorthy's website from /research/graphs/moorthy.gr. You can also download Dr. John Punin's
Java Course. These files are XGMML graphs with extra meta-data which was obtained by setting a modified version of
w3c's webbot loose to spider through a given website. Please email me if you would like to create your own graphs and
would like help.
* If you would like to just get log data and process it after, you'll need to do some programming or convince me to
do it. Right now I generate logs as in real time. Asking for 1 request per minute will actually take 1 minute to
generate a request, rather than pump out logs with 1 minute between the timestamps.


> genlog --help

genlog 0.0 
Send patches and bug reports to morria@cs.rpi.edu

Usage: genlog [--version] [--help] --graph
        [--run-for-seconds] [--number-requests] [--requests-per-min]
        [--user-timeout] [--num-users] [--no-wait]

  -v,--version                          Print version information and quit
  -h,--help                             Print this help information and quit
  -q,--quiet                            Be quiet ( no debug output to stderr )
  -s,--run-for-seconds=<SECONDS>        Generate log entries for SECONDS seconds
  -n,--number-requests=<REQUESTS>       Generate REQUESTS log entries
  -r,--requests-per-min=<RPM>           Generate RPM requests per minute
  -t,--user-timeout=<TIMEOUT>           Have users tend leave by TIMEOUT seconds
  -u,--num-users=<USERS>                Try to have USERS random active users at any time
  -w,--no-wait                          Do not pause between log entries ( timestamps are
                                         not adjusted )