Then copy the libmysqll.dll from D:\wamp\bin\mysql\mysql5.0.51a\bin into your \R\R-2.7.0\library\RMySQL\libs directory.

After that you should be able to run similar reports to those identified above. Substitute the quartz() function (for Mac OSX) to the windows() function for WinXP.

Also instead of using a C:\my.cnf you can also pass in connection properties with your dbConnect call:
con< -dbConnect(m, user="root", port="3306", host="myhost", dbname="mydb")

I would use Ruby|Perl to open up your existing SiteScope\log files, parse them by line, and then use a mysqlimport statement at the end of the script.

For example, this line in a SiteScope log:
00:03:06 02/23/2008goodIF.1IF Server Ping 0.01 sec2:2216ok10.0100


Could be translated into a csv using ruby so that it reads something like:
02-23-2008 00:03:06,good,IF.1,IF Server,Ping,0.01,10.0,100

Then at the end of your ruby script suck that data into MySql using a puts %x{mysql import command}

I’m doing something similar here

A mysql import command on windows looks like this:
mysqlimport -h host -P 3306 -s -u myimportuser --password=mypassword --fields-optionally-enclosed-by=""" --fields-terminated-by=, --lines-terminated-by="\r\n" --local perfstats D:\path\to\your\file_created_by_ruby.csv

I’ve left some current Ruby scripts that do something very similar at work, so will post them on Monday for you.

I’m tending to use SiteScope less and less though. I hate its lack of flexibility (in presenting data) and I prefer to tweak my own monitors launched via scheduled tasks or cron as appropriate, that way I can control exactly what counters I want, and how they are formatted. For example, its not easy (possible?) to setup a custom DTrace monitor for Solaris using SiteScope, and it’s sometimes easier (to get access) to just ask a remote sysadmin to run a set of native commands he’s used to rather than impose the clunky use of SiteScope -> ssh -> …

For example, on windows, for batch collection I just use typeperf as in
typeperf -cf D:\monitors\perfcounters.txt -si 55 -sc 5 -s \\%1 -o

For solaris or *ix, I just use vmstat, iostat and sar mostly via cron. Or other arrangements depending on the system and level of access granted …

For app servers I use JMX, called via a ruby script as per that previous link I showed you.

I import all that data usually into a MySql database, which you can still use SiteScope as a reporting tool against that data (if you wish). But like in this article, tools like R let you quickly analyze tons of data, without necessarily worrying about how you set the monitor up in the first place. In other words, I’m leaning towards (and I know some people will hate this approach ), monitor *everything*, report on select few … way of doing things…

I’ve even mucked around with a SiteScope ‘faker’ which can re-present data from a MySql database when a LoadRunner controller is running, giving you the perception of live counters during a load test run All SiteScope really does is present XML data via a http port 8888 in this case …

Hmm, too much coffee this morning

Cheers,
Tim Koopmans

Since I’m using SiteScope to collect performance data, I could have it log into a MySQL DB… now you’ve got me thinking.

How do you get your data into MySQL?