On 14/12/05, Iago Toral Quiroga <[EMAIL PROTECTED]> wrote: > El mar, 13-12-2005 a las 18:19, sebb escribió: > > On 13/12/05, Iago Toral Quiroga <[EMAIL PROTECTED]> wrote: > > > However, is there any way to get threads of different machines execute > > > diferent requests? > > > > Yes, make sure the data files are have different data in them. > > That works fine. Thanks sebb! > > Regarding to the problems with constant troughput timer, I discovered > the following: > > I cannot reproduce the problem in my development environment. So that, > in my dev. environment the constant troughput timer produces a 10 > requests per second test sending requests to the server at each second > as it should be, but the same test, in the work environment produces the > problem I had talked about before: sends bunches of 100 requests each 10 > seconds. > > This is quite weird, I can't understand why is this happening. > > In my development environment I'm using: > - Debian Linux, kernel 2.6.12 > - Jakarta Jmeter 2.1.1 > - Apache 1.3 > - Java 1.5.0_04 > - CPU: Celeron 2.4GHz > - RAM: 512MB > > While in the working environment I'm using: > - Windows XP proffesional SP1 > - Jakarta Jmeter 2.1.1 > - Apache 1.3 > - Java 1.5.0_04 > - CPU: Celeron 2.4GHz > - RAM: 512MB > > Any ideas? Any of you have experienced a behaviour like this one before? > May this be happening because of a Windows or Windows JVM issue? or may > it be a consequence of some kind of network configuration?
Could be a JVM threading / priority issue, I suppose. > the key question here is how does jmeter decide when to launch a set of > requests to the server and how many when using a constant throughput > timer? > The code is in http://svn.apache.org/repos/asf/jakarta/jmeter/trunk/src/components/org/apache/jmeter/timers/ConstantThroughputTimer.java See the delay() method. S. --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
