You are viewing a plain text version of this content. The canonical link for it is here.
Posted to java-user@axis.apache.org by Martin Gainty <mg...@hotmail.com> on 2013/02/01 03:19:03 UTC
RE: AW: AW: AW: a question about Axis2 performance test with JMeter
Josef und Schuler
und Schüler
I suggest running Jmeter at command line thri mvn with maven-jmeterplugin at http://wiki.apache.org/jmeter/JMeterMavenPlugin
concerning Josef's suggestion of callback you could create a simple AxisCallback as seen here
AxisCallback callback = new AxisCallback() { /**
* This is called when we receive a message.
*
* @param msgContext the (response) MessageContext
*/
public void onMessage(MessageContext msgContext) {
OMElement result = msgContext.getEnvelope().getBody().getFirstElement();
TestingUtils.compareWithCreatedOMElement(result);
log.debug("result = " + result);
} /**
* This gets called when a fault message is received.
*
* @param msgContext the MessageContext containing the fault.
*/
public void onFault(MessageContext msgContext) {
fail("Fault received");
}
/**
* This gets called ONLY when an internal processing exception occurs.
*
* @param e the Exception which caused the problem
*/
public void onError(Exception e) {
} /** This is called at the end of the MEP no matter what happens, quite like a finally block. */
public void onComplete() {
finish = true;
try {
notify();
}
catch(Exception excp)
{
System.err.println("AsyncServiceTest::onComplete throws Exception message="+excp.getMessage());
}
}
}; sender = new ServiceClient(configcontext, service);
sender.setOptions(options); sender.sendReceiveNonBlocking(operationName, method, callback);
log.info("send the request");
synchronized (callback) {
if (!finish) {
callback.wait(45000);
if (!finish) {
throw new AxisFault(
"Server was shutdown as the async response take too long to complete");
}
}
}
ServiceClient object calls: public void sendReceiveNonBlocking(QName operation, OMElement elem, AxisCallback callback)
throws AxisFault {
MessageContext mc = new MessageContext();
fillSOAPEnvelope(mc, elem);
OperationClient mepClient = createClient(operation);
// here a blocking invocation happens in a new thread, so the
// progamming model is non blocking
mepClient.setCallback(callback);
mepClient.addMessageContext(mc);
mepClient.execute(false);
}
callback will route toonMessage for normal processing
onError for handling error
onFault for handling faults
I dont understand the need for a new protocol when HTTP 1.1 using NIO connector will handle any callback nicelyperhaps you can explain?
Viel Gluck!
Martin
______________________________________________
Verzicht und Vertraulichkeitanmerkung
Diese Nachricht ist vertraulich. Sollten Sie nicht der vorgesehene Empfaenger sein, so bitten wir hoeflich um eine Mitteilung. Jede unbefugte Weiterleitung oder Fertigung einer Kopie ist unzulaessig. Diese Nachricht dient lediglich dem Austausch von Informationen und entfaltet keine rechtliche Bindungswirkung. Aufgrund der leichten Manipulierbarkeit von E-Mails koennen wir keine Haftung fuer den Inhalt uebernehmen.
Subject: AW: AW: AW: a question about Axis2 performance test with JMeter
Date: Thu, 31 Jan 2013 14:41:20 +0100
From: josef.stadelmann@axa-winterthur.ch
To: java-user@axis.apache.org
Hi Yunxi as you said, fireAndForget() sends without expecting a response OK given the web-service intends to answer this fireAndForget() request you need to pass it an address OK,this is normally done in an addressing header also engaging the addressing.mar of axis2 (requestor and responder) each side how do you intend to do that ? well I think for your service requestor and for your web-service this is no problem, you have done so ! What you in fact implement that way is some sort of an asynchronous request response behavior (a protocol) which you might achieve differently too using axis2. BUT In any case, if the requestor does not wait for the responder the responder needs to know the network address & port of the http-listener maybe a callback routine which handles this late returning asynchronous response event Regarding JMeterit must play very exactly the role of your service requestorit is the requestor, and hence it has to send a request and all the responder needs to know to call back later at asynchronous time, laterit has to prepare for the responder the same environment as said aboveby passing listeners address and portby passing a call back routine which shall handle the response event In this regard it is very logically what JMeter needs to do. BUTI think you can't do that with JMeter as it stands today.JMeter is very menu driven, and that limits you (potentially). I would suggest to try to contact the developer of JMeter offline and ask about such further going possibilities. Also I encourage others in the forum to participate on this interesting area, given they know (and I know there are some) much more then I do. Josef Von: Yunxi Zhang [mailto:zhangyx395@hotmail.com]
Gesendet: Mittwoch, 30. Januar 2013 12:55
An: java-user@axis.apache.org
Betreff: Re: AW: AW: a question about Axis2 performance test with JMeter Hi, Josef, Thank you for your reply. To be honest, i am a PhD student and i've been designing and developing my own protocol for two Web Services. Basically, to apply my protocol, the two Web Services should be both Web Services (i would call them as a service provider and a service requester instead of a server or a client). Due to the unique characteristics of my protocol, each Web Service can determine whether or not to send out a response message after processing a receiving message. To enable each Web Service to decide whether to send out a response message, i tried using the function fireAndForget(element) with the RawXMLInOnlyMessageReceiver, since this function can send out a message without any concerns about receiving a response, after processing an incoming message through the use of the RawXMLInOnlyMessageReceiver. In order to verify my protocol, i only want to test its performance. i then found the JMeter, which seemed as an appropriate testing tool for testing the performance of Web Services, and i've gone through the tutorial already. Normally, in order to use the fireAndForget() function, a receiver's EndpointReference as an address is required. There is no issue to enable two Web Services to communicate in this deployment, since i can set the EndpointReference of the counterpart in each Web Service. However, when the JMeter is used to simulate an initial message sender that can send a message to a Web Service and wait for the reply, i would assume that the Web Service should be able to know the address of the JMeter. That's why i asked my initial question. By the way, since i'm using Mac OS, so software and tools for Windows OS may not be suitable to my testing platform. Hope you can understand my situation and give me more suggestions on the performance testing. Best wishes, Yunxi On 29 Jan 2013, at 19:06, Stadelmann Josef wrote:
Good you understand that.
To be honest and say the true, I never used JMeter but I used Visual Studios
Universal Test and Performance Measurement facilites to get a clue about how
well we designed our .NET WCF-3.5 Web-Service SOAP/XML Client talking to
or OpenVMS / TOMCAT / Apache Axis2 / Spezpla Web Service Server - talking
to our integrated SPg-Legacy (some 120'000 Lines of Pascal Code on OpenVMS
of course ;-)).
This tools are somewhat complex to explain in 5 minutes, but I have read
just briefly over the NetBeans Tutorial and I come fast into the Apache
User Guide for JMeter, also real worth to read, as released by Apache.
JMeter Plugin brings almost a lot components one needs to understand a bit
This my impression;;
The Help is therefore real worth to read;
The components available to JMeter should be understood first
(at least the big picture needs to be understood in proper context (your setup));
Else I think you AND I get lost in space.
You can measure whatever you want ! i.e. access times to remote running FTP
servers to catch down 2 files by 3 concurrent users each catching down a file
3 times ... get the recording and show a graphical output.
So you need to know what you want to measure! I would not suggest to start
first with Web Services but get familiar with the tools as such and more
simpler things.
BUT you can measure the time from when a SOAP/XML package leaves the
client and when it returns to client code control.
and if you like do a great job you should best start at
http://jmeter.apache.org/usermanual/index.html
then read click you to the user manual
http://jmeter.apache.org/usermanual/index.html
Or take the NetBeans Tour
http://netbeans.org/kb/docs/javaee/ecommerce/test-profile.html
AND
Just by doing a very simple example taking the NetBeans On-Line Help;
an awful a lot of information is provided.
So Question: what are you all doing with so many component?
You need to understand the big picture first!
A sampler which will sample data for you, in your case it is a
sampler able to sample SOAP messages as they arrive from the server.
Be warned: I just saw Axis but not Axis2
So that's all for today/night - interesting - isn't it?
AND
if you can give me more details - then I might be able to help you
for what to look for.
Josef
18.1 Samplers
FTP Request
HTTP Request
JDBC Request
Java Request
SOAP/XML-RPC Request
WebService(SOAP) Request
LDAP Request
LDAP Extended Request
Access Log Sampler
BeanShell Sampler
BSF Sampler
JSR223 Sampler
TCP Sampler
JMS Publisher
JMS Subscriber
JMS Point-to-Point
JUnit Request
Mail Reader Sampler
Test Action
SMTP Sampler
18.2 Logic Controllers
Simple Controller
Loop Controller
Once Only Controller
Interleave Controller
Random Controller
Random Order Controller
Throughput Controller
Runtime Controller
If Controller
While Controller
Switch Controller
ForEach Controller
Module Controller
Include Controller
Transaction Controller
Recording Controller
18.3 Listeners
Sample Result Save Configuration
Graph Full Results
Graph Results
Spline Visualizer
Assertion Results
View Results Tree
Aggregate Report
View Results in Table
Simple Data Writer
Monitor Results
Distribution Graph (alpha)
Aggregate Graph
Mailer Visualizer
BeanShell Listener
Summary Report
Save Responses to a file
BSF Listener
JSR223 Listener
Generate Summary Results
Comparison Assertion Visualizer
18.4 Configuration Elements
CSV Data Set Config
FTP Request Defaults
HTTP Authorization Manager
HTTP Cache Manager
HTTP Cookie Manager
HTTP Request Defaults
HTTP Header Manager
Java Request Defaults
JDBC Connection Configuration
Login Config Element
LDAP Request Defaults
LDAP Extended Request Defaults
TCP Sampler Config
User Defined Variables
Random Variable
Counter
Simple Config Element
18.5 Assertions
Response Assertion
Duration Assertion
Size Assertion
XML Assertion
BeanShell Assertion
MD5Hex Assertion
HTML Assertion
XPath Assertion
XML Schema Assertion
BSF Assertion
JSR223 Assertion
Compare Assertion
SMIME Assertion
18.6 Timers
Constant Timer
Gaussian Random Timer
Uniform Random Timer
Constant Throughput Timer
Synchronizing Timer
BeanShell Timer
BSF Timer
JSR223 Timer
18.7 Pre Processors
HTML Link Parser
HTTP URL Re-writing Modifier
HTML Parameter Mask
HTTP User Parameter Modifier
User Parameters
BeanShell PreProcessor
BSF PreProcessor
JSR223 PreProcessor
18.8 Post-Processors
Regular Expression Extractor
XPath Extractor
Result Status Action Handler
BeanShell PostProcessor
BSF PostProcessor
JSR223 PostProcessor
18.9 Miscellaneous Features
Test Plan
Thread Group
WorkBench
SSL Manager
HTTP Proxy Server
HTTP Mirror Server
Property Display
Debug Sampler
Debug PostProcessor
18.10 Reports
Report Plan
Report Table
HTML Report Writer
Report Page
Line Graph
Bar Chart
-----Ursprüngliche Nachricht-----
Von: Yunxi Zhang [mailto:zhangyx395@hotmail.com]
Gesendet: Dienstag, 29. Januar 2013 12:19
An: java-user@axis.apache.org
Betreff: Re: AW: a question about Axis2 performance test with JMeter
Hi, Josef,
Thank you for your reply.
So, according to your explanation, should i use RawXMLInOutMessageReceiver. If i am going to use RawXMLInOutMessageReceiver, will the Axis2 reply to JMeter automatically or do i need to set the reply address manually?
Many Thanks,
Yunxi
On 29 Jan 2013, at 08:32, Stadelmann Josef wrote:
The RawXMLInonlyMessageReceiver hais by definition no SOAP/XML return message!There is a return to the client on the TCP/IP layer as an ACK, saying that the TCP/IP package has arrived and its content, the SOAP Message has been delivered.But that's it.Josef -----Ursprüngliche Nachricht-----Von: Yunxi Zhang [mailto:zhangyx395@hotmail.com]Gesendet: Montag, 28. Januar 2013 15:14An: java-user@axis.apache.orgBetreff: a question about Axis2 performance test with JMeter Hi, I've developed my own Web Service by using the POJO method of Axis2, and the receiver i'm using is the RawXMLInonlyMessageReceiver. Now i want to use JMeter to test the performance of my Web Service. I know JMeter can send soap messages to a Web Service, but how should i set the responding address in my Web Service to enable JMeter to obtain the correct response soap message? Has anyone got any idea about it? Regards Yunxi---------------------------------------------------------------------To unsubscribe, e-mail: java-user-unsubscribe@axis.apache.orgFor additional commands, e-mail: java-user-help@axis.apache.org ---------------------------------------------------------------------To unsubscribe, e-mail: java-user-unsubscribe@axis.apache.orgFor additional commands, e-mail: java-user-help@axis.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: java-user-unsubscribe@axis.apache.org
For additional commands, e-mail: java-user-help@axis.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: java-user-unsubscribe@axis.apache.org
For additional commands, e-mail: java-user-help@axis.apache.org