Re: issues

[Mike Stover <ms...@apache.org>]

On Wednesday 13 June 2001 10:39, Leigh.White@syntegra.com wrote:
> Following on from my previous email, I am basically using jmeter to call a
> jsp which in turn calls a EJB method remotely.  This is a dummy EJB and
> just outputs a string, ultimately though I would like to use jmeter on our
> actual J2EE app.  We will also be using OptimizeIT to catch any performance
> issues that arise during this load testing.  In order to test this I have
> it plugged into my weblogic server and am monitoring the results of my
> Jmeter tests.
>
> I have some issues with Jmeter that are more than likely down to me than
> the tool itself, I have detailed these below, if anyone could help it would
> be great.
>
> I can create the script no bother, and run it however:
>
> 1)	I get a lot of exceptions in the jmeter console mainly
> ConnectExceptions mentioning the sockets. (I am not using hhtps only http),
> sometimes I get a netAddress or something I can't remember exactly as I
> cannot recreate one now I wan to know what it is.

I have seen this type of behavior, and it seems to be related to the site I'm 
hitting at the time.  Using more threads, the number of failed connections 
increases, I've found.  Do you get errors when you have just 1 thread?  10?  
See if there's a correlation.

>
> 2)	I run my script, which is quite simple, but it only seems to last a
> second or two, I thought Jmeter would loop is this the case or will it only
> execute through the script once.

If I remember correctly, your script contained a OnceOnly controller.  All 
samples below that controller will be performed only once and then dropped.  
Remove the OnceOnly controller and it should loop forever.

>
> 3)	I have run my script with the threads set to 500.  Does this mean it
> will test 500 threads total, if so how many is it doing at a time.  And how
> can I increase the overall test duration or vary the simulated load?

500 threads means JMeter will try to spin up 500 threads and then run them 
all simultaneously.  I'd be surprised if your computer handled it well unless 
you have multiple processors.  You can put any number you want in there, but, 
in the end, JMeter's ability to stress your server is limited by the power of 
your machine.  To truly increase the stress load, you need to use RMI and run 
several machines at once.  JMeter is designed to do this, however, the 
functionality needs someone to work on it because it is currently a little 
broken.  ( I believe there are some serialization issues).

>
> 4)	After I have run my script, I cannot re-run it, I stop it, clear the
> visualizers and then run, and nothing happens.  I suspect this could be to
> do with the exceptions I am encountering, but I have to keep exiting Jmeter
> then going back in.  Is there anything that anyone recognises could be at
> fault here (aside from me :-)

Not sure - I've never seen that.  It's possible that having the whole test 
contained in the OnceOnly controller is interfering with JMeter's ability to 
stop the test (ironically enough!).  I re-run JMeter repeatedly all the time.

Hmm, nope - I just tried that out, and it still worked fine.  Maybe it just 
takes JMeter a very long time to shut down 500 threads?  Ok, nope tried that 
too (and actually, it still worked quite well with that many threads).  Have 
you tried the simplest test scripts and still had these problems?

-- 
Mike Stover
mstover1@apache.org

---------------------------------------------------------------------
To unsubscribe, e-mail: jmeter-user-unsubscribe@jakarta.apache.org
For additional commands, e-mail: jmeter-user-help@jakarta.apache.org