On 29/08/2007, Pieter Ennes <[EMAIL PROTECTED]> wrote: > Hi Sebb, > > sebb wrote: > > On 24/08/07, Pieter Ennes <[EMAIL PROTECTED]> wrote: > >> Hi list, > >> > >> We have a situation where our scripts are fairly short and need to > >> start frequently, so that the expense of starting up a Java > >> environment each time is very high compared to the actual script > >> execution. > > > > Can you not run the scripts in a loop, with suitable delays at the end > > of the loop? > > Good point, but not an option I think, since the scripts will become > many and that would mean running out of memory resources pretty quickly > on the machine.
Why? The memory resources will only increase if you use additional listeners. > >> Would there be any way to submit test scripts to a running jmeter > >> server without using Java to do the actual submission? > > > > Probably not, as JMeter uses RMI for client-server communication. > > Ok, no problem, we can go for some Java programming I guess :) > However, since I'm not really into that yet, I seem to fail in finding > out what the preferred/documented way is to talk to a running jMeter > server through RMI. Could you give me a hint on where to start? Any > pointer to (e.g. jMeter) example code would do. Sorry, too complicated to explain - you'll have to look at the JMeter code. And you will need to keep some of the JMeter code anyway to handle the responses. > One more question: Am I correct that the original idea (submit file, > wait for server to execute script, submit next file) will work when > programmed in Java? Or is there no such thing as 'submit XML file to > server'... > The server expects to receive the script via RMI; there is no such thing as sending an XML script across. > >> Should I be looking into Beanshell? > >> Or something like PHP/Java integration? > >> (http://php.net/manual/en/ref.java.php) > > > > Can't see how these would help. > > > > == > > > > The JMeter GUI allows one to run a script, load another file, and then run > > that. > > In theory, this functionality could be added to non-GUI mode. > > > > However you could design a mechanism to define the tests and the > > timings, and would then need to do the necessary Java coding. > > Ok, a problem here would probably be that our input files are dynamical, > and thus not known in advance. The scripts can use variables read from files or generated on the fly. If the test is not known in advance, then I don't see how JMeter can execute them - it cannot make up its own tests. > Merci, > -- > - Pieter > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [EMAIL PROTECTED] > For additional commands, e-mail: [EMAIL PROTECTED] > > --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
