Hi Ruwan, it is a normal scenario and I can reproduce it at will with your snapshot version v5. Snapshot version v4 and any version before run without problems. Regarding the stack I had the same impression. Here is again a description with what I did: 1) I deployed a Hello World SOAP WebService to JBoss 4.2.2 GA 2) Using soapUI 2.0.2 I created a test request based on the WSDL <soapenv:Envelope xmlns:soapenv="http://schemas.xmlsoap.org/soap/envelope/"; xmlns:ser="http://service.jamba.de/";> <soapenv:Header/> <soapenv:Body> <ser:helloWorld/> </soapenv:Body> </soapenv:Envelope> Test responds looks like that: <env:Envelope xmlns:env="http://schemas.xmlsoap.org/soap/envelope/";> <env:Header/> <env:Body> <ns2:helloWorldResponse xmlns:ns2="http://service.jamba.de/";> <return>Hello World!</return> </ns2:helloWorldResponse> </env:Body> </env:Envelope> 3) I created a load test with strategy "thread", 1 start thread, 1 end thread and conducted a test for 60 seconds against the service directly --> no problem 4) I configured the service in synapse <syn:proxy name="HelloWorldService" startOnLoad="true" statistics="enable"> <syn:target endpoint="HelloWorld"> <syn:outSequence> <syn:send/> </syn:outSequence> </syn:target> </syn:proxy> <syn:endpoint name="HelloWorld"> <syn:address uri="http://r11:10080/WSTest/HelloWorldService"/> </syn:endpoint> 5) In soapUI I changed the endpoint to the ESB address an rerun the same load test. After about 5 seconds (99 requests served, I receive the exception posted (always the same stack trace and always after the same time). 98-99 requests without error and then this exception. 6) I rerun the same test with WSO2 ESB 1.6 and your snapshot build v3 and v4 and had no problem. What additional information do you need? I'll ask a colleague to conduct the same test on Linux today. Regards, Eric
________________________________ Von: [EMAIL PROTECTED] im Auftrag von Ruwan Linton Gesendet: Mo 05.05.2008 01:24 An: [email protected] Betreff: Re: AW: [esb-java-dev] AsynchronousCloseException Hi Eric, We do not recommend to host ESB in windows for production, but it should run on Windows for the normal scenarios. If you can consistently reproduce this issue, could you please tell us the exact procedure to reproduce this issue? By looking at the exception stack trace I also think this is purely on the nio implementation level, and I cannot think of how this regression has been injected, because we have not changed the httpcore-nio dependency. Anyway it is good for us to know the issue, if there are any. So please tell us the exact procedure. Thanks, Ruwan Hubert, Eric wrote: > Hi Asankha, > > ok, most of the time I only tested on Linux, as this will be our production > environment. Right now I'm at home and only have a few virtualized > Linux-boxes on some windows machines available. That's why I tested straight > on Windows, as I did not think much about the differences in the underlying > nio implementation for such basic tests. > > I did not put the ESB in maintenance mode. I just sequentially sent messages > via soapUI to the ESB and got that error, without changing something on the > ESB (no maintenance mode, no graceful restart)... It was just ment as a > warm-up test to check the environment, and I'm happy that I did this before. > Otherwise I might have thought of a bug in the new graceful shutdown code. ;-) > > I re-run the exact same test against version 1.6 of WSO2 ESB (on the same > machine, with the same configuration) - no problems. So this looks like a > regression. I'm not sure whether I would be able to trigger the same problem > under Linux. > > If you need some further information to reproduce the problem, please just > ask. I'm willing to help. > > Regards, > Eric > > > > ________________________________ > > Von: [EMAIL PROTECTED] im Auftrag von Asankha C. Perera > Gesendet: So 04.05.2008 20:14 > An: [email protected] > Betreff: Re: [esb-java-dev] AsynchronousCloseException > > > Eric > > > > > Since your production environment is Linux, I would not recommend > trying to isolate the above problem in Windows. One basic difference with the > NIO platform is that we read using channels, and then need to switch into > Java streams for the rest of the processing, and we use Pipe's in Unix, and > Simulated (i.e. Buffered) pipes in Windows. Also, its good to stick to JDK > 1.5.0_14 wherever possible. > > If you put the ESB into maintenance during the 'load test', it may have > caused some problem to the client as new connections would have been refused > by the ESB.. I am not sure I understand exactly how you did the test. I would > try to reproduce this tomorrow, and thought of using the SimpleStockQuote > service, the Java ApacheBench client and the ESB.. and putting the ESB into > graceful shutdown? Is this what you tried as well? > > asankha > > > > ------------------------------------------------------------------------ > > _______________________________________________ > Esb-java-dev mailing list > [email protected] > http://wso2.org/cgi-bin/mailman/listinfo/esb-java-dev > _______________________________________________ Esb-java-dev mailing list [email protected] http://wso2.org/cgi-bin/mailman/listinfo/esb-java-dev
<<winmail.dat>>
_______________________________________________ Esb-java-dev mailing list [email protected] http://wso2.org/cgi-bin/mailman/listinfo/esb-java-dev
