[Resolved] Minimum statistic increases during a test

Hi!

Strange, I've never heard of this. Just for the record, what statistic are we talking here (i.e. the minimum of what)? Minimum Response size, min Time taken, etc.?

When does this usually happen? Is there any possibility that you could send the loadUI Project (don't worry, referred soapUI Projects are not included) and the results folder for an execution where this happens to us (usually to be found in c:\userName\.loadui\results)?

Thanks for reporting this!

Henrik
SmartBear Software

Hi

Thanks for responding so soon.

It's the "min" statistic as choosen in the "statistics" analysis component and reported in the attached csv as well as the table log component. I assume it's minimum response time.
I've attached the results.csv and project xml.

Have a great holiday!

cheers

Iain

Ok, this is expected since you are using the Statistics component, which has a "sliding window" for its statistics (have a look in the component's settings dialog, under tab Periods > Chart period (min) ). So min will by default be the minimum value recorded during the last minute.

With this being said, we encourage everyone interested in stats to use the new Statistics Workbench instead (if this sounds confusing, have a look here for an introduction to loadUI statistics). In the Statistics Workbench, you will get the minimum value from the start of the test.

Regards

Henrik
SmartBear Software

Thanks Henrik

That seems logical now.

I've tried using the statistics workbench in the past with the aim of exporting the results (via "export raw data") and manipulating those results in Excel in order to provide consistent reporting (consistent with, in this case, output from JMeter) but the exported results are odd.

I've pasted below what I get - each metric has the same figure for each timestamped request (except the second to last), which I guess makes sense from a statistical point of view.
So I'm guessing the way to get the metrics for the entire overall run is to calculate max,min,average etc from one of the columns.Easy enough.
The sample at timestamp 0:04:00 does look odd though in that each cell varies and has a standard deviation reading (where the other rows do not). Is there any reason that this row is the odd one out?

TIMESTAMP MEDIAN 90TH MIN AVERAGE MAX STD_DEV
0:00:51 38562 38562 38562 38562 38562 0
0:00:54 29801 29801 29801 29801 29801 0
0:01:35 34742 34742 34742 34742 34742 0
0:01:45 68698 68698 68698 68698 68698 0
0:02:03 74709 74709 74709 74709 74709 0
0:02:30 65656 65656 65656 65656 65656 0
0:02:34 81761 81761 81761 81761 81761 0
0:02:43 67010 67010 67010 67010 67010 0
0:03:02 62532 62532 62532 62532 62532 0
0:03:24 96154 96154 96154 96154 96154 0
0:03:46 58250 58250 58250 58250 58250 0
0:03:52 52298 52298 52298 52298 52298 0
0:04:00 98163.5 99906.3 95985 98163.5 100342 2178.5
0:04:03 86838 86838 86838 86838 86838 0
"
cheers

Iain

IMounsey-Smith wrote:
The sample at timestamp 0:04:00 does look odd though in that each cell varies and has a standard deviation reading (where the other rows do not). Is there any reason that this row is the odd one out?

It looks like all the other time intervals just got one single sample (which is why standard deviation is zero and all other values exactly the same).

So, I would guess that between 0:03:52 and 0:04:00 there were more than one sample.

Does this make sense?

Henrik
SmartBear Software

PS. Maybe looking on the Seconds zoom level helps explaining things?

Yes it does
Appreciate the quick responses
Thanks
Iain