MysqlCollector plugin added to JMeter wiki: http://wiki.apache.org/jakarta-jmeter/MysqlCollectorPlugin (check attachments for the patch).
Please excuse the lack of conforming to programming standards, I know theres a lot of bad things (tm) in there, if more experienced developers can give some advice / patches it could be a lot more useful. the core logic is there, and it works for our test plans. Have just finished the SNMP sampler, starting on the GUI.... Any comments on streamlining development of this (I've started a sourceforge project for the mysql plugin, but need to be able to have drop-in functionality and run it as a project seperate from JMeter). Regards, Brett On Wed, Feb 24, 2010 at 12:39 PM, Thibaut Raballand < [email protected]> wrote: > That's sounds like great stuff ! :) > And more generic that the stuff I do to fill in the holes. > > Tibo > > On Wed, Feb 24, 2010 at 11:05, Brett Cave <[email protected]> wrote: > > > Just finished the MysqlCollectorGui & MysqlCollector classes, nothing too > > complicated, just finalising db schema and fixing up prepared statements, > > then will share. it works great, but there's no doubt plenty of room for > > improvement. > > > > Also, have just finished downloading snmp4j, next step is to add a > sampler > > that polls SNMP on target hosts to get resource usage and add the results > > into the collector. I would say SNMP is pretty generic and implemented on > > most servers anyway, and it beats running a "jmeter-agent" like some of > the > > load testing frameworks. (then again, an agent might not be a bad > idea...) > > > > Regards, > > Brett > > > > On Wed, Feb 24, 2010 at 1:40 AM, Deepak Shetty <[email protected]> > wrote: > > > > > >Could you go into a little more detail about how you use a listener to > > > write > > > >data to the DB > > > you dont need a listener, you can do it after the test has run. > > > If your result file is CSV this is trivial. If XML then its fairly easy > > to > > > parse and insert. > > > > > > On Tue, Feb 23, 2010 at 2:55 PM, James Hill <[email protected]> > > > wrote: > > > > > > > Could you go into a little more detail about how you use a listener > to > > > > write > > > > data to the DB? I've been looking at doing it as part of the Ant task > > > that > > > > calls JMeter but if there's an easier way I'd love to find it :) > > > > > > > > Also, what do you use to collect load/mem/cpu usage from the servers? > > I'm > > > > considering sar to do this, but seeing as there's an existing license > > for > > > > Spotlight on Unix I'm not sure I need to (seeing as it collects that > > info > > > > anyway). However, it could be handy for another project where SoU > isn't > > > in > > > > use. > > > > > > > > I like the idea of the php website to collate and display the > results. > > > When > > > > I have some spare time I'd like to put together a USB drive with > > JMeter, > > > > MySQL and relevant scripts and howto's that can be used on just about > > any > > > > site I end up at. Simplify the startup time. As you point out > Thibaut, > > it > > > > takes time to get to that point but it must save a lot of hassle in > the > > > > long > > > > run. > > > > > > > > On Wed, Feb 24, 2010 at 1:58 AM, Thibaut Raballand < > > > > [email protected]> wrote: > > > > > > > > > Hi, > > > > > > > > > > As for us, > > > > > - We send the results of each run directly from JMeter to a mysql > DB > > > > (with > > > > > a > > > > > listener) > > > > > - We collect load / mem / cpu usage from the servers to the same DB > > > > > automatically > > > > > - We have a PHP web site the correlate automatically those datas > > > > > > > > > > Sure, you need some time to put all this up and running, but it's > > worth > > > > it. > > > > > > > > > > Regards, > > > > > Tibo > > > > > > > > > > On Tue, Feb 23, 2010 at 14:45, Brett Cave <[email protected]> > > wrote: > > > > > > > > > > > hi, > > > > > > > > > > > > been using jmeter for a few weeks now, and wondering how other > > users > > > > > > correlate target load / mem / cpu usage into jmeter reporting? My > > > > current > > > > > > method is to enable SNMP and use a seperate RRD-tool based system > > to > > > > > > generate graphs, and then correlate the target resource usage > with > > > the > > > > > load > > > > > > injection manually. This is a manual process, and i would like to > > get > > > > > data > > > > > > specific to each test i run (load testing currently runs a number > > of > > > > > tests, > > > > > > 1 by 1). > > > > > > > > > > > > Regards, > > > > > > Brett > > > > > > > > > > > > > > > > > > > > >
