On 22 October 2013 19:16, Kirk Pepperdine <[email protected]> wrote: > > On 2013-10-22, at 5:22 PM, Milamber <[email protected]> wrote: > >> >> Le 21/10/2013 01:55, sebb a ecrit : >>> On 19 October 2013 16:47, Milamber <[email protected]> wrote: >>>> Le 19/10/2013 01:26, sebb a ecrit : >>>> >>>>> [Trying again - please do not hijack this thread.] >>>>> >>>>> The Constant Throughput Timer (CTT) calculates the desired wait time, >>>>> and if this is less than zero - i.e. a sample should already have been >>>>> generated - it could trigger the creation of a failed Assertion (or >>>>> similar) >>>>> showing the time difference. >>>>> >>>>> Would this be sufficient to detect all CO occurrences? >>>> >>>> An option in CTT element which allow to mark as fail the(s) sampler(s) in >>>> the scope with a less than zero wait time seems a good point to inform >>>> users. >>>> The failed message can indicate the delay to help users to fix the scenario >>>> (i.e. add more VU and reduce frequency of CTT) >>> Or CTT would save the delay somewhere, and a new Assertion could be >>> created to report it. >>> >>> That might work better if there are more places where delays could be >>> detected. >> >> Yes good idea. > > I think I suggested a separate detector. Could it not be a listener instead?
Yes, a listener would have access to all the sample results. The reason I suggested the CTT is that the element "knows" the expected transaction rate, so can detect late samples easily and cheaply. Other code will either have to be provided with the expected rate or have to analyse the data to deduce the information. Analysis which may be expensive. [However if the detection is only done in GUI mode that would not be an issue.] > Regards, > Kirk > > > --------------------------------------------------------------------- > 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]
