Hi Eric,
Let me jump into the game :-)
Hubert, Eric wrote:
Hi Asankha,
propably the last update from me for today. Just as an additional information. I let the tcpmon between the ESB and JBoss and conducted the load test with a single thread. No problem, but of course the speed decreased (only 4 tps, I had between 30 and 40 without tcpmon). So I used 10 starting threads and 20 end threads and started another test for 60 seconds. My small CPU was almost saturated on the machine running soapUI, JBoss and tcpmon, soapUI showed around 50 tps, no problem. I then did a last test without tcpmon in between and again only one thread and had about 35 tps. After 6 seconds boom... Same stacktrace as already posted. So somehow tcpmon "fixes" this problem.
I think tcpmon keeps the connection between both parties and that is why
this seems fixed, in other wards, now the http connection is between
synapse and the tcpmon and not within the synapse and JBOSS. Can you get
the tcpdump from the OS without using the tcpmon so the http connection
will still be between synapse and JBOSS. Any way I will work with
Asankha on this today.
I really don't know what's going on here and how this is all related to some recent changes. Do you have weekly builds from synapse somewhere? I would take the time to narrow down the timefram of the change, which introduced this problem, without knowing what it actually can be.
I suspect this is due to the fix that we did recently for the pipe-leek
issue. You could turn on the debug logs of the transport in the
log4j.properties file. (org.apache.synapse.transport log level to DEBUG)
and attach the log with the exception stack trace.
I don't know whether this would help. Unfortunately the time between the
working v4 and v5 is pretty long. About one month with a great number of
commits in axis2 and synapse.
Again this should be due to the transport changes so we need to look at
org.apache.synapse.transport.
Thanks,
Ruwan.
Would it make sense to selectively turn on some logs? E.g. put a general level to trace. Set a global category to warn and only some classes to trace? I don't know the affected code, so I don't know if that would make sense or not.
Regards,
Eric
________________________________
Von: [EMAIL PROTECTED] im Auftrag von Hubert, Eric
Gesendet: Mo 05.05.2008 21:16
An: [email protected]
Betreff: AW: AW: AW: AW: [esb-java-dev] AsynchronousCloseException
Hi Asankha,
ok, here the requested tcpdump information:
request from soapUI:
POST http://r11:10081/WSTest/HelloWorldService HTTP/1.1
Host: r2:10081
SOAPAction: ""
Content-Type: text/xml; charset=UTF-8
Transfer-Encoding: chunked
Connection: Keep-Alive
User-Agent: Synapse-HttpComponents-NIO
e4
<?xml version='1.0' encoding='UTF-8'?><soapenv:Envelope
xmlns:soapenv="http://schemas.xmlsoap.org/soap/envelope/";
xmlns:ser="http://service.jamba.de/";><soapenv:Body <http://service.jamba.de/%22%3E%3Csoapenv:Body>
>
<ser:helloWorld />
</soapenv:Body></soapenv:Envelope>
0
response from JBoss 4.2.2 GA (using modified Tomcat 5.5)
HTTP/1.1 200 OK
Server: Apache-Coyote/1.1
X-Powered-By: Servlet 2.4; JBoss-4.2.2.GA (build: SVNTag=JBoss_4_2_2_GA
date=200710221139)/Tomcat-5.5
Content-Type: text/xml;charset=UTF-8
Transfer-Encoding: chunked
Date: Mon, 05 May 2008 19:03:42 GMT
f4
<env:Envelope xmlns:env='http://schemas.xmlsoap.org/soap/envelope/'><env:Header></env:Header><env:Body><ns2:helloWorldResponse
xmlns:ns2="http://service.jamba.de/";><return>Hello <http://service.jamba.de/%22%3E%3Creturn%3EHello>
World!</return></ns2:helloWorldResponse></env:Body></env:Envelope>
0
I can't see anything suspicious... What about you?
Regards,
Eric
------------------------------------------------------------------------
_______________________________________________
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
