Thanks Philippe I did read that, but still confused when to use one over the other. Do you have any examples? Maybe that would make it a little more clear. However, whether the CTP is calculated based on when each thread last ran or when any thread last ran, the goal of the timer to reach the target throughput. Why when using shared mode does the throughput jump from 250 tps to 333 tps when changing CTP from 20,000 to 20,001? Thanks again. - Marcelo
> Date: Wed, 18 Feb 2015 21:19:09 +0100 > Subject: Re: Constant Throughput Timer limitation or bug? > From: [email protected] > To: [email protected] > > Hi, > As per: > http://jmeter.apache.org/usermanual/component_reference.html#Constant_Throughput_Timer > > > - all active threads in current thread group - the target throughput is > divided amongst all the active threads in the group. Each thread will delay > as needed, based on when it last ran. > - all active threads in current thread group (shared) - as above, but > each thread is delayed based on when any thread in the group last ran. > > Regards > > On Wed, Feb 18, 2015 at 9:00 PM, Marcelo Jara <[email protected]> > wrote: > > > Not sure why my original email's format is off. > > > > Updating CTP to use "all active threads in current thread group" seems to > > work. Not sure why using "shared" exhibits the odd results. > > > > > > > From: [email protected] > > > To: [email protected] > > > Subject: Constant Throughput Timer limitation or bug? > > > Date: Wed, 18 Feb 2015 14:36:49 -0500 > > > > > > I am using Jmeter 2.12 and noticed that during a test, the transactions > > per second was much higher than I had configured. Using a value of 15000 > > (250 TPS) works as expected. If I go to 15001, the TPS jumps to 333. Here > > are some more results: > > CTP = 15000, expected = 250 tps, result 250 tpsCTP = 15001, expected = 250 > > tps, result 333 tps > > CTP = 20000, expected = 333 tps, result 333 tpsCTP = 20001, expected = 333 > > tps, result 500 tps > > CTP = 30000, expected = 500 tps, result 500 tpsCTP = 30001, expected = 500 > > tps, result 1000 tps > > > > > > > > > I was able to create a simple test plan that exhibits the same issue. > > Thread Group (100 threads, loop forever,scheduler with duration of 60) > > ---- Debug Sampler (all defaults) ---- CTP (all active threads in > > current thread group (shared) > > > > > > > > > Anyone else run into this? > > > Thanks, > > > Marcelo > > > > > > > > -- > Cordialement. > Philippe Mouawad.
