Issue with reqs/sec calculation for the report?

Gatling (Open-Source)
#1

Hi,

I’m not sure if I’m looking at this the wrong way. Stats is not my forte. I have 4 requests in a test; Each pair performs a given query that hits our ElasticSearch server; one of them goes through HAProxy, the other hits the ElasticSearch server directly.

When I look at the report I’m seeing that while the mean times are different, the reqs/sec remain the same across all similar requests.

Here’s a screenshot of the report.

There’s six requests in there (actual names vary a bit):

ElasticSearch Query 1 (through HAProxy)
ElasticSearch Query 1 (direct)

ElasticSearch Query 2 (through HAProxy)
ElasticSearch Query 2 (direct)

oam_logout…
oam_logout… redirect 1

If we take a look at the first two we can see that the mean for the first is 1898, for the second it is 3062 (it takes about twice as long to get a response, no?). I’d expect the reqs/sec to be around half for the second one, and yet they are both exactly 5.031. The same goes for the other pair (which is a completely different ES query).

Is this a bug? Or am I misunderstanding the stats?

Regards,
Roberto

#2

Forgot to mention that I am using the following versions:

`

<properties>
   <maven.compiler.source>1.7</maven.compiler.source>
   <maven.compiler.target>1.7</maven.compiler.target>
   <scala.version>2.11.4</scala.version>
   <encoding>UTF-8</encoding>
   <gatling.version>2.1.4</gatling.version>
   <gatling.maven.plugin.version>2.1.2</gatling.maven.plugin.version>
   <scala-maven-plugin.version>3.2.0</scala-maven-plugin.version>
</properties>

`

The simulations are being called from maven.

`

% mvn --version ✹ ✭
Apache Maven 3.2.5 (12a6b3acb947671f09b81f49094c53f426d8cea1; 2014-12-14T11:29:23-06:00)
Maven home: /usr/local/Cellar/maven/3.2.5/libexec
Java version: 1.7.0_75, vendor: Oracle Corporation
Java home: /Library/Java/JavaVirtualMachines/jdk1.7.0_75.jdk/Contents/Home/jre
Default locale: en_US, platform encoding: UTF-8
OS name: “mac os x”, version: “10.10.1”, arch: “x86_64”, family: “mac”

`

#3

You’re looking at mean values on the whole run duration. Same count => same rps