Bug 64846

Summary: Total Throughput calculation on html report is not correct
Product: JMeter Reporter: Ashok Reddy <ashokreddy1208>
Component: MainAssignee: JMeter issues mailing list <issues>
Status: NEW ---    
Severity: major CC: p.mouawad
Priority: P1    
Version: 5.3   
Target Milestone: JMETER_5.3.0   
Hardware: PC   
OS: Linux   
Attachments: sum of all sample request throughput is not matching with total tps
Please find schema for same test plan
TPS per sample occurrence duration vs TPS per test duration

Description Ashok Reddy 2020-10-23 13:53:12 UTC
Created attachment 37530 [details]
sum of all sample request throughput is not matching with total tps

sum of all sample request throughput is not matching with total throughput:

Example:-
sample 1 throughput - 9528.37
sample 2 throughput  - 2219.28
sample 3 throughput - 1900.87
sample 4 throughput - 3443.52

---------------------------------
Sum of all samples -  9528.37+2219.28+1900.87+3443.52 = 17092.04  but in HTML report total throughput is showing as 7603.34


can you plz help how this calculation happens

output summary:-
summary = 10500000 in 00:23:01 = 7601.6/s Avg:    54 Min:     3 Max:  2479 Err:   441 (0.00%)
Comment 1 Philippe Mouawad 2020-10-24 14:18:03 UTC
Can you open your test plan and run :

Tools > Generate Schematic Overview

and attach result here.

Thanks
Comment 2 Ashok Reddy 2020-10-27 04:02:17 UTC
Created attachment 37535 [details]
Please find schema for same test plan
Comment 3 Ashok Reddy 2020-11-03 10:16:50 UTC
any update on this issue
Comment 4 piotr.mirek 2021-07-14 14:52:30 UTC
Created attachment 37954 [details]
TPS per sample occurrence duration vs TPS per test duration

My guess is that jmeter TPS is calculated for each sampler bases with it's start/end time, which will give different results when you do the same for whole group. Please see the the attachment.

As you can see for "reset-wiremock" example (single request sent), JMeter calculated throughput for time between the first sample going out timestamp and last sample coming in timestamp, which results with 3.2 TPS for single request over 300 seconds (1 sample and duration 0.312s -> 3.2 TPS). 

It's a matter of question, how you would like to calculate TPS for each sampler - per sampler occurrence, or per whole test