On 21 October 2011 15:56, Adrian Speteanu <[email protected]> wrote: > The BS Server was a nice addition. If the Throughput timer is shared by all > threads can you still change its value dynamically?
Yes, as the page says, you can use the server to set a property (which is global to JMeter) and then reference the property in th CTT. The BSH server runs in its own thread - not sure it's possible to set JMeter variables (which are local to a thread). > Regards, > Adrian S > > On Fri, Oct 21, 2011 at 2:42 AM, sebb <[email protected]> wrote: > >> On 21 October 2011 00:40, sebb <[email protected]> wrote: >> > On 21 October 2011 00:36, David Luu <[email protected]> wrote: >> >> Thanks, the beanshell server looks useful for what I'd want. >> > >> > That's very flexible but it would be simpler if the variable were set >> > in the test plan itself. >> > >> > e.g. if you wanted to ramp up gradually, you could use a counter with >> > a suitable increment to define the throughput variable. >> >> I meant to add: for experimental purposes, I suggest using a test plan >> with some Java samplers; you can then test without annoying the >> neighbours ;-). >> >> >> On Thu, Oct 20, 2011 at 10:43 AM, sebb <[email protected]> wrote: >> >> >> >>> On 20 October 2011 18:14, Adrian Speteanu <[email protected]> wrote: >> >>> > Hey David, >> >>> > >> >>> > I've used both. >> >>> > >> >>> > Its not much to compare :). If you understand the Constant Throughput >> >>> Timer, >> >>> > then you understand both and it's easy to see that its major >> limitation >> >>> is >> >>> > that it holds (if possible) the throughput constant at the value you >> >>> define >> >>> > (which was the intent). However if you want to easily create a more >> >>> complex >> >>> > test scenario like: >> >>> >> >>> The "constant" can also be varied. If the constant is expressed as a >> >>> variable or function reference, that allows the throughput to be >> >>> varied at will. >> >>> >> >>> See for example: >> >>> >> >>> >> >>> >> http://jakarta.apache.org/jmeter/usermanual/best-practices.html#beanshell_server >> >>> >> >>> >> >>> > 1. run a test with throughput 100 for 2 min >> >>> > 2. continue for 2 more minutes with throughput 200 >> >>> > 3. continue for 2 more minutes with throughput 500 >> >>> > to see how the application responds to these changes in the load >> >>> received, >> >>> > without the plugin you had to use more thread groups and configure >> the >> >>> > scheduling so you get the same effect. The plugin makes these use >> cases >> >>> > simple and more easy to understand. >> >>> > >> >>> > You can make things a lot more complicated than that if you'd like >> using >> >>> > both methods and a lot of imagination. >> >>> > >> >>> > On Tue, Oct 11, 2011 at 5:19 AM, David Luu <[email protected]> >> wrote: >> >>> > >> >>> >> I was wondering if anybody has used both, and wonder how they >> compare. >> >>> >> Pros/cons of each. >> >>> >> >> >>> >> And how one could best simulate the Throughput Shaping Timer with >> the >> >>> >> constant throughput timer via variable/property that defines the >> >>> throughput >> >>> >> and changing it during test run. Particularly how would you >> implement >> >>> the >> >>> >> changing of the throughput variable and where to place in test plan. >> >>> >> >> >>> >> FYI, I'm refering to this JMeter plugin: >> >>> >> >> >>> >> http://code.google.com/p/jmeter-plugins/wiki/ThroughputShapingTimer >> >>> >> >> >>> > >> >>> >> >>> --------------------------------------------------------------------- >> >>> 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] >> >> > --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
