Return-Path: Mailing-List: contact jmeter-user-help@jakarta.apache.org; run by ezmlm Delivered-To: mailing list jmeter-user@jakarta.apache.org Received: (qmail 67371 invoked from network); 23 Sep 2003 07:24:23 -0000 Received: from unknown (HELO mx1.pin.se) (81.93.130.15) by daedalus.apache.org with SMTP; 23 Sep 2003 07:24:23 -0000 Received: from hof (h50n2fls308o1003.telia.com [81.224.183.50]) by mx1.pin.se (Postfix) with ESMTP id 7548F1415D; Tue, 23 Sep 2003 09:24:34 +0200 (CEST) From: "Lars-Erik Helander" To: "JMeter Users List" , Subject: RE: Running multiple GUI sessions Date: Tue, 23 Sep 2003 09:20:05 +0200 Message-ID: MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit X-Priority: 3 (Normal) X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2910.0) Importance: Normal X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4807.1700 X-Spam-Rating: daedalus.apache.org 1.6.2 0/1000/N In my search for a specific functionality I came across this discussion held a while ago. In my case the proposed solution is not ideal, and that is why I would like to comment on it (and propose a possible path forward). In my scenario I have several machines that act as load-generators. Some of these machines are equipped with special hardware in order to be able to create special type of stimuli to my test objects. One example of "special" hardware are GSM modems used to load my test object via a GSM network. The total setup is quite complex with different kinds of machines and different kinds of protocols , including creation of background load to my databases etc. So now we have a real world situation where it is not possible to run the identical test plan on all machines involved. A natural way to go would then be to follow the principle of multiple controlling GUIs, but then I think you loose one of the major benefits of JMeter, namely the possibility to control, and save the result of, a particular test run from/in a single entity, that you later can study. Further the possibility to do regression tests by reapplying the same test pattern is of tremendous value. It will be hard to trust that you reapply your test patterns identically, if this is not supported by the tool you use, and as said before JMeter does a great job doing this, under "normal" circumstances. I have made a small "hack" that allowed me to specify for each ThreadGroup the hostnames of the machines that that particular ThreadGroup should be started on, and that simple piece of functionality solves the problem for me. I am happy do dicsuss this or other ways to solve my problem. And if a way is found that needs modifications to JMeter I would hope that it could become an integral part of a future JMeter release. Regards Lasse Helander