I would like to thank you very much for your time and help...it worked out
...:)
On Fri, Aug 16, 2013 at 8:07 AM, Flavio Cysne <flaviocy...@gmail.com> wrote:
> JMeter Test Plan will look like this:
>
> Test Plan
> - CSV Data Set Config (Recycle on EOF: True, Stop thread on EOF: false,
> Sharing mode: All threads)
> - Thread Group (threads: 250, rampup: 0, loop: 16)
> - HTTP Request 1
> - Synchronizing Timer (threads: 250)
> - Constant Timer (delay: ${__javaScript(${__threadNum()}*1818)}
> // 60000 / 33 = ~1818
> - HTTP Request 2
> ...
> - HTTP Request n
>
>
>
> 2013/8/16 umesh prajapati <praumes...@gmail.com>
>
> > Thank you very much for making me understand what constant timer is. I
> see
> > you mentioned about synchronizing timer . So you have mentioned that
> > synchronizing timer will start all the threads at the same time. But in
> my
> > case, I would like 33 users to login in 1 minute. And lets say i have 250
> > users in my csv file but I would like to run my test for 2 hrs. And to
> > accomplish this I am increasing the number of loops. To complete one loop
> > lets say it takes 8 mins. so I will need around 15-16 loop counts.
> >
> > Number of threads: 250
> > Ramo up: 0
> > Loop: 16
> >
> > Constant timer to be added before http sampler : (60000/33) ms
> >
> > So do I need synchronizing timer. Because if I do add synchronizing timer
> > it will start all 250 users at once (just to be clear when you say start
> > all at once) did you mean all 250 user login request will be sent at same
> > time.
> > On Aug 16, 2013 7:18 AM, "Flavio Cysne" <flaviocy...@gmail.com> wrote:
> >
> > > When you use Synchronizing Timer all the threads will start at the same
> > > time, when the number of threads started reaches the configured value.
> > > In your test plan Ramp-up will delay the start of the threads, but next
> > > loop execution will break down your test, as you have stated.
> > > Then Constant Timer is used to overcome this issue, delaying the start
> > time
> > > of the threads as ramp-up does, but not messing thing up after the
> first
> > > loop.
> > > BTW, your script Threads Group Ramp-up have to be 0 (zero).
> > >
> > >
> > > 2013/8/16 umesh prajapati <praumes...@gmail.com>
> > >
> > > > Thank you but I am still confused. So you mean constant time also
> does
> > > the
> > > > same job like ramp-up period. Lets say I have following configuration
> > > >
> > > > Number of threads : 250
> > > > Ramp up period : 450
> > > > Loop : 3
> > > >
> > > > From what I have understood about rampup. It takes 450 seconds to
> > > complete
> > > > 250 threads. So the delay between thread 1 and thread 2 is 450/250
> > > seconds.
> > > >
> > > > So if I also add constant timer to this configuration lets say
> > (450/250)
> > > > seconds. I don't get why do we need constant timer when ramp up is
> > doing
> > > > the same thing.
> > > > On Aug 16, 2013 4:51 AM, "Flavio Cysne" <flaviocy...@gmail.com>
> wrote:
> > > >
> > > > > No. The Constant Timer is the delay between the start of one thread
> > to
> > > > the
> > > > > start of the next thread. No relation among threads execution time
> at
> > > > all.
> > > > > So if thread 1 starts and take 1 second to complete, and Constant
> > Timer
> > > > has
> > > > > 300ms delay, then the second thread will start before 1st thread
> > ends.
> > > > >
> > > > > Ramp-up also put this delay in your threads, the same way constant
> > > timer
> > > > > will do. If you want to leverage concurrency remove the timer.
> > > > >
> > > > >
> > > > > 2013/8/15 umesh prajapati <praumes...@gmail.com>
> > > > >
> > > > > > If I put the timer, than it means 2nd request has to wait until
> the
> > > > first
> > > > > > request is complete. This means if first request is taking time
> to
> > > > > process
> > > > > > the request than 2nd request will not start until the first
> request
> > > is
> > > > > > completed. Which means there won't be a load but I am doing a
> load
> > > > test.
> > > > > > On Aug 14, 2013 12:50 PM, "Flavio Cysne" <flaviocy...@gmail.com>
> > > > wrote:
> > > > > >
> > > > > > > only in the first http sampler
> > > > > > >
> > > > > > >
> > > > > > > 2013/8/14 umesh prajapati <praumes...@gmail.com>
> > > > > > >
> > > > > > > > one quick question do i add the timer on thread group or the
> > http
> > > > > > reques
> > > > > > > > sampler
> > > > > > > >
> > > > > > > >
> > > > > > > > On Wed, Aug 14, 2013 at 9:25 AM, umesh prajapati <
> > > > > praumes...@gmail.com
> > > > > > > > >wrote:
> > > > > > > >
> > > > > > > > > @ flavio Thanks ...will try it out and let you know the
> > result.
> > > > > > > > >
> > > > > > > > >
> > > > > > > > > On Wed, Aug 14, 2013 at 8:35 AM, Flavio Cysne <
> > > > > flaviocy...@gmail.com
> > > > > > > > >wrote:
> > > > > > > > >
> > > > > > > > >> I think you misunderstood me.
> > > > > > > > >>
> > > > > > > > >> "I know in my application, one user is not allowed to
> login
> > > > > multiple
> > > > > > > > >> times until
> > > > > > > > >> the session thats logged in is logged out."
> > > > > > > > >>
> > > > > > > > >> Loop 2 threads can't start until all Loop 1 threads are
> > over.
> > > > > > > > >> Synchronizing
> > > > > > > > >> Timer with "Number of simulated users" equal to the number
> > of
> > > > > > threads
> > > > > > > > will
> > > > > > > > >> ensure this.
> > > > > > > > >>
> > > > > > > > >> "@Flavio I dont need to start all my thread at once,..."
> > > > > > > > >>
> > > > > > > > >> That's why I suggest you to use a Constant timer with an
> > > > > expression.
> > > > > > > > >>
> > > > > > > > >> "...But I would like to test the real scenario, where 2000
> > > users
> > > > > > login
> > > > > > > > in
> > > > > > > > >> 2
> > > > > > > > >> hrs. So, basically 1 min 33 users."
> > > > > > > > >>
> > > > > > > > >> Modify that 300 in Constant Timer expression for the value
> > > > > > equivalent
> > > > > > > to
> > > > > > > > >> time gap for 33 users within 1 minute. (60000ms / 33
> users =
> > > > > > > > ~1818ms/user)
> > > > > > > > >>
> > > > > > > > >> Regards
> > > > > > > > >>
> > > > > > > > >
> > > > > > > > >
> > > > > > > >
> > > > > > >
> > > > > >
> > > > >
> > > >
> > >
> >
>
