On 8 February 2012 22:36, Robin D. Wilson <rwils...@gmail.com> wrote: > So the note from the manual is actually what I would like to capture > (basically the round-trip time for request/response) - but it > is clear that's not what JM2.6 is doing... > > I'm using the Summary Report listener. It shows columns for: > > Label # Samples Average Min Max Std. Dev. Error > % Throughput KB/Sec Avg. Bytes > > I had assumed that Average was the average of whatever 'elapsed time' JMeter > was using for the request, but either I'm mistaken, or > JM2.6 isn't correctly measuring the elapsed time for the request.
It's the average of the samples that have the same Label. So it will be somewhere between Min and Max. > My test cgi script is pretty simple, it just sleeps for 2 seconds, then > returns an HTML document that is basically 15,700 (or so) > bytes of useless text. I was trying to simulate the slowdown I'm seeing in > JMeter2.6 vs. JMeter2.4 from my standard test cases. > > There is something else I don't understand - related to the 'average' shown > in the Summary Report listener... If I do 10,000 > iterations of a single request, and spread those over 100 threads - a rough > estimate of my elapsed time should be something like > '(samples/threads) * average'. But I'm getting numbers that are waaay out of > whack with what I'm actually seeing: > > JM2.6: 10000 samples, 100 threads, 756 average ... I should get ~75.6 > seconds total elapsed time to run the test. (Plus any > JMeter overhead.) This assumes that all 100 threads start simultaneously and finish simultaneously. > The actual time is 210 seconds. > > JM2.4: 10000 samples, 100 threads, 767 average ... I should get ~76.7 > seconds total elapsed time to run the test. (Plus any > JMeter overhead.) > The actual time is 211 seconds Same here. > So the average is _waaaay_ off from what I'm thinking it should be, but the > total time actually is in-line with the expected > 'average' for the request/response I'm sending... Not necessarily, which is why I asked for the elapsed times you are seeing. > While we're at it, the 'throughput' has the same problem - except that it is > _vastly_ different from JM2.4 to JM2.6. Specifically, > the throughput for the above JM2.6 test says: 48.4/sec, whereas the > throughput for the above JM2.4 test is 128.3/sec. > > The _real_ throughput (based on the samples/elapsed time in seconds) is > actually around 47/sec (so JM2.6 appears to be closer to > accurate on that). Again, the throughput calculations are based on total test time. Are you sure the test run times are comparable? There's clearly something else going on here. Please recheck the individual sample response times and see how they compare to the average. If there still appears to be a problem, create a Bugzilla issue and attach: - JMX test case - log files for JMeter 2.4 and 2.6 - CSV result files for 2.4 and 2.6 > -- > Robin D. Wilson > Sr. Director of Web Development > KingsIsle Entertainment, Inc. > VOICE: 512-777-1861 > www.KingsIsle.com > > > -----Original Message----- > From: sebb [mailto:seb...@gmail.com] > Sent: Wednesday, February 08, 2012 3:09 PM > To: JMeter Users List > Subject: Re: Average times are confusing me... > > On 8 February 2012 20:59, Robin D. Wilson <rwils...@gmail.com> wrote: >> So, I created a test case that simply performs a 'POST' to a cgi-bin test >> script. The test script simple pauses for 2 seconds, > then >> returns a page that is full of useless text. >> >> When I run this on my JMeter 2.6 GUI, I see that the 'average' sample time >> for this page (this is the only sampler in the test) is >> ~750ms... BUT I _KNOW_ that each page takes at least 2 seconds to return >> from the request. Because I have a 'sleep 2' at the top > of >> the page. >> >> So is the 'average' time actually measuring the time it takes to receive the >> response (after the response starts), or is it >> measuring the time it takes to receive the response after the request ends >> (which is what I really want)? > > Neither. > The average is the average of the response (elapsed) times for the > relevant set of results. > The response (elapsed) time is described here: > > http://jmeter.apache.org/usermanual/glossary.html > > What do the individual elapsed times show? > [e.g. use Table or Tree View Listener] > > Which Listener are you using that shows the average times? > >> -- >> Robin D. Wilson >> Sr. Director of Web Development >> KingsIsle Entertainment, Inc. >> VOICE: 512-777-1861 >> www.KingsIsle.com >> >> >> >> >> --------------------------------------------------------------------- >> To unsubscribe, e-mail: user-unsubscr...@jmeter.apache.org >> For additional commands, e-mail: user-h...@jmeter.apache.org >> > > --------------------------------------------------------------------- > To unsubscribe, e-mail: user-unsubscr...@jmeter.apache.org > For additional commands, e-mail: user-h...@jmeter.apache.org > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: user-unsubscr...@jmeter.apache.org > For additional commands, e-mail: user-h...@jmeter.apache.org > --------------------------------------------------------------------- To unsubscribe, e-mail: user-unsubscr...@jmeter.apache.org For additional commands, e-mail: user-h...@jmeter.apache.org
