TBH, reading your latest input and trying to figure out the problem is hard, as the previous context is lost. IMHO, you'll get better chances to get an answer if you articulate clearly your whole/current situation again, without missing anything, but without repeating irrelevant parts.
On Thu, Dec 7, 2023 at 4:05 PM Jun Zhuang <thornbird...@yahoo.com.invalid> wrote: > Figured out, it's the name of the variables I defined for the Gaussian > timer. In my user defined variables list, I defined the following > > [image: Inline image] > > then used them in the Gaussian timer like > > [image: Inline image] > > somehow that got JMeter confused. When I reduce the value of those > variables, say set them to 100/300 or both to 0, I got higher throughput. > Same after I changed their names by removing the ThinkTime_ part. > > Anyone has an idea why? > > Thanks, > Jun > On Thursday, December 7, 2023 at 11:12:10 AM EST, Jun Zhuang < > thornbird...@yahoo.com> wrote: > > > In the following screenshot, my understanding is the timer in the > *Transaction: > Load Login Page *only works in that scope, does not matter whether it's > applied in the beginning or end, is that right? If so, then what's causing > the significantly slowness with the no-timer block? > > [image: Inline image] > > On Thursday, December 7, 2023 at 10:58:40 AM EST, Tong Sun < > suntong...@gmail.com> wrote: > > > > > On Thu, Dec 7, 2023 at 10:50 AM Jun Zhuang <thornbird...@yahoo.com.invalid> > wrote: > > My apologies, my previous email was a mess, re-sending. > > ----------------- > > I am seeing unexpected behavior with the timer scoping. I am not using any > timer for requests in the 1st half of my test plan (DB operations only) and > only using the Gaussian timer (2 - 4 secs) in the 2nd half of the test, > i.e., in the === Create Sample Batching Job === simple controller, but it > seems to be applied universally anyway. The reason I am saying "applied > universally" is because the test runs very fast in GUI mode (~ 1 min) when > I use the run without pause option or in non-GUI mode by setting the timer > to 0. > > When I use the 2-4 seconds setting in non-GUI mode, the test ran for more > than 5 MINUTES and mostly in the 1st half the DB query part. > > The timer is placed within the last request in each of the transactions, > so my understanding is it will only happen after the last request in each > transaction. > > [image: Inline image] > > > > > ---------- Forwarded message --------- > From: *Dmitri T* <glin...@live.com> > Date: Thu, Dec 7, 2023 at 10:39 AM > Subject: Re: JMeter timer scoping issue > To: JMeter Users List <user@jmeter.apache.org>, Jun Zhuang < > thornbird...@yahoo.com> > > Timers are executed *before* each Sampler in their scope. If you want it > to be executed after all Samplers you need to add a Flow Control Action > < > https://jmeter.apache.org/usermanual/component_reference.html#Flow_Control_Action > > > sampler and make the timer a child of this sampler. Or Flow Control > Action sampler can create delay on its own exactly where it's placed. > > More information: > > * Scoping Rules > <https://jmeter.apache.org/usermanual/test_plan.html#scoping_rules> > * A Comprehensive Guide to Using JMeter Timers > <https://www.blazemeter.com/blog/jmeter-timer> > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: user-unsubscr...@jmeter.apache.org > For additional commands, e-mail: user-h...@jmeter.apache.org
