You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@jmeter.apache.org by bu...@apache.org on 2014/03/03 10:17:57 UTC

[Bug 56206] New: Response times reported by TCP sampler in a thread group scale with the number of threads

https://issues.apache.org/bugzilla/show_bug.cgi?id=56206

            Bug ID: 56206
           Summary: Response times reported by TCP sampler in a thread
                    group scale with the number of threads
           Product: JMeter
           Version: 2.11
          Hardware: PC
            Status: NEW
          Severity: major
          Priority: P2
         Component: Main
          Assignee: issues@jmeter.apache.org
          Reporter: bobruto@gmail.com

Created attachment 31359
  --> https://issues.apache.org/bugzilla/attachment.cgi?id=31359&action=edit
Python echo server and Jmeter config

A single tcp sampler within a thread group is run a few times in a row (re-use
connection option is off).
It sends request to an echo service which has a more or less fixed response
delay
With just 1 thread, response times reported by jmeter match up with the echo
server.
If you increase the number of threads, response times reported by Jmeter scale
in direct proportion with that number.

Test scenario also includes a timer alongside the sampler which one can play
with (as well as with the thread group's ramp up time) to observe various
effects (2 threads, 1sec ramp up time, 200ms timer = 800ms sample times instead
of 1000 without the timer)


Steps to reproduce:
just run the attached echo server (Python) and the Jmeter test plan with
various thread group settings


Expected results:
Sample times around 500ms regardless of the number of threads (at least until
the echo server gets overwhelmed)



Observed on both Windows 7 x64 and Ubuntu 13.10
java version "1.7.0_10"
Java(TM) SE Runtime Environment (build 1.7.0_10-b18)
Java HotSpot(TM) 64-Bit Server VM (build 23.6-b04, mixed mode)

Jmeter 2.11.r1554548

-- 
You are receiving this mail because:
You are the assignee for the bug.

[Bug 56206] Response times reported by TCP sampler in a thread group scale with the number of threads

Posted by bu...@apache.org.
https://issues.apache.org/bugzilla/show_bug.cgi?id=56206

--- Comment #3 from Sebb <se...@apache.org> ---
Created attachment 31361
  --> https://issues.apache.org/bugzilla/attachment.cgi?id=31361&action=edit
Sample TCP text plan with different thread counts

The attached plan shows the TCP Sampler with thread counts of 1,2,20.

The average response time is approx the same for all the thread groups.

-- 
You are receiving this mail because:
You are the assignee for the bug.

[Bug 56206] Response times reported by TCP sampler in a thread group scale with the number of threads

Posted by bu...@apache.org.
https://issues.apache.org/bugzilla/show_bug.cgi?id=56206

Sebb <se...@apache.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|---                         |WORKSFORME

--- Comment #2 from Sebb <se...@apache.org> ---
This is clearly not a JMeter problem.

As already noted on the user list, multiple threads running a TCP request
against a multi-threaded HTTP server show that the TCP sampler behaves
correctly.

-- 
You are receiving this mail because:
You are the assignee for the bug.

[Bug 56206] Response times reported by TCP sampler in a thread group scale with the number of threads

Posted by bu...@apache.org.
https://issues.apache.org/bugzilla/show_bug.cgi?id=56206

bobruto@gmail.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 OS|                            |All

--- Comment #1 from bobruto@gmail.com ---
...And I'll be the first to notice that the echo server attached is
single-threaded. The fact that the tcp client can write the request to socket
output stream before the server accepts the connection can throw one off, but
otherwise Jmeter seems to behave as it should.
Now the real question is why the actual software we're testing with JMeter
demonstrates similarly serial nature when it's supposed to handle incoming
requests in parallel, but that's for another issue tracker

-- 
You are receiving this mail because:
You are the assignee for the bug.

[Bug 56206] Response times reported by TCP sampler in a thread group scale with the number of threads

Posted by bu...@apache.org.
https://issues.apache.org/bugzilla/show_bug.cgi?id=56206

bobruto@gmail.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |CLOSED
         Resolution|WORKSFORME                  |INVALID

-- 
You are receiving this mail because:
You are the assignee for the bug.