On 17/06/2010, João Tiago Ferreira <[email protected]> wrote: > > > > -----Original Message----- > > From: sebb [mailto:[email protected]] > > > Sent: quinta-feira, 17 de Junho de 2010 15:57 > > To: JMeter Users List > > Subject: Re: Jmeter JMS support for Tibco EMS > > > > On 17/06/2010, João Tiago Ferreira <[email protected]> > > wrote: > > > > Logs attached in jira. > > > > I think you mean Bugzilla ... > > > Yes Bugzilla :p > > > > > > Can you also attach details of the source changes you needed to make? > > > I modified org.apache.jmeter.protocol.jms.sampler.Receiver.run() line 84 in > release 2.3.4 from: > MessageAdmin.getAdmin().putReply(reply.getJMSMessageID(), reply); > To: > MessageAdmin.getAdmin().putReply(reply.getJMSCorrelationID(), reply); > > > > > > > Responses bellow > > > > > > > > > -----Original Message----- > > > From: sebb [mailto:[email protected]] > > > > > > Sent: quinta-feira, 17 de Junho de 2010 14:29 > > > To: JMeter Users List > > > Subject: Re: Jmeter JMS support for Tibco EMS > > > > > > OK. > > > > > > On 17/06/2010, João Tiago Ferreira <[email protected]> > > wrote: > > > > Hi, i will post my logs in this issue: > > https://issues.apache.org/bugzilla/show_bug.cgi?id=46142 . You agree? > > > > > > > > João > > > > > > > > > > > > -----Original Message----- > > > > From: sebb [mailto:[email protected]] > > > > Sent: quinta-feira, 17 de Junho de 2010 13:33 > > > > To: JMeter Users List > > > > Subject: Re: Jmeter JMS support for Tibco EMS > > > > > > > > On 17/06/2010, João Tiago Ferreira > > <[email protected]> wrote: > > > > > Hi > > > > > > > > > > I am using jmeter to perform some functional tests in our > > application. I > > > > > need to send a request for the JMS server and wait for a > > response. > > > > > > > > > > Our use case for the the use of correlation and message ids is > > that the > > > > > messageID of the request is used to set the correlationID of > > the response. > > > > > > > > OK. > > > > > > > > > Using jmeter JMS request response the correlation is not > > working as > > > > > expected, neither in 2.3.4 or nightly builds. > > > > > > > > I'm not surprised about 2.3.4, however I thought the nightly > > builds were better. > > > > > > > > Have you tried both settings of "Use Request Message Id As > > Correlation Id"? > > > > > > > > > > > > Not checking that option requires setting the correlationID in JMS > > Properties in the test, which I think makes no sense in Tibco EMS > > neither other JMS servers... > > > > > > > Why not? > > > > Surely one use case would be for the sender to supply the > > correlationId that they want the service to return? > > > > This would allow the sender to choose something more meaningful than > > the messageId - over which the sender has no control. > > > > > > > > > > > > > > > > > > > I send the logs as attachments. > > > > > > > > Please don't send attachments to the list. > > > > In this case they got lost anyway. > > > > > > > > Post them somewhere public and send the link, or create a > > Bugzilla > > > > issue and attach them there. That is probably best in this case. > > > > > > > > > > > > > > > > > > > As far as I can understand the JMS Sampler is using the > > messageID of the > > > > > response to match the messageID of the request, > > > > > > > > I don't think it does that, but I'll check. > > > > > > > > > while it should use the correlationID of the response. > > > > > > > > Sometimes - I think that depends on how the JMS service is > > configured. > > > > > > > > > I easily modified the source to fit my needs, but wanted to > > contribute my > > > > > testing with Tibco EMS. I also noted that several bug reports > > exist about > > > > > this issue. > > > > > > > > Yes, this is a tricky area, partly because JMS knowledge within > > the > > > > team is limited, and partly because the protocol does not seem to > > be > > > > particularly well-defined with regard to how correlationId is > > treated. > > > > > > > > AIUI, not all JMS providers behave the same way: > > > > Some may copy the request messageId to the response > > correlationId. > > > > Others may copy the request correlationId to the response > > correlationId. > > > > There may be other behaviours. > > > > > > > > I could not find any documentation that defines what the > > "correct" behaviour is. > > > > As far as I could tell, the setting of the response correlationId > > is > > > > server- or application-defined. [If you have details, please post > > a > > > > link to the docs!] > > > > > > > > > In my application is responsibility of the application to set the > > correlationID from the request messageID. > > > > OK. > > > > This tends to confirm what I wrote about the lack of a standard, and > > therefore the JMeter behaviour needs to be confgurable. > > > Yes the reference isn't clear and some use cases tend to break other, so > making this configurable is the way to go...
It looks like there are 4 options altogether, depending on which fields to use to correlate the messages: * request field: messageId or correlationId * response field: messageId or correlationId The current nightly code supports only the following 2 options: + request:messageId == reply:messageId + request:correlationId == reply:correlationId However AIUI your use case is request:messageId == reply:correlationId which is not currently supported. I think the solution is to allow the request and response "match" fields to be chosen independently. --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
