Idea ID: 2872312

JMeter Test from LRE gets much Lower TPS than expected when response time has no issue.

Status : New Idea

When no timer set in JMeter script, no pacing set in runtime settings, JMeter test execution from LRE is able to get similar response time as direct JMeter run, but TPS is very low comparing to direct JMeter run and excepted value calculating from the LRE test parameters and results (e.g. No. of VUs, response time).

With 15 concurrent users, average response time is 0.22 sec, which is correct. Without any timer/pacing, TPS should be 60+ (able to achieve it from direct JMeter run). But TPS from LRE run is just around 12. Total transaction count is tally with this TPS, which is much lower than expected.

Environment:

LRE2021 R1

JMeter version is customized 5.4

In LRE, from Analysis result file:

Average Transaction Response time = 0.22 second.

TPS = 12.

 

In JMeter directly:

Average Transaction Response time = 0.19 second.

TPS = 77.

The TPS in LRE is much lower than in JMeter.

OCTIM10C623179 

Tags:

Labels:

Protocols