Hi Eric,
thanks for providing this fix and keep up the good work! I can confirm
that it is working now. Issue 1 as well as issue 2 are resolved now.
Nice to hear that :-)
Yes, you should be able to test the cluster aware stickiness also in
this build. Well, I just uploaded the build to [1] and tested with the
hessian service code that you sent to me even with the list size as
15000, when it goes to 20 000 the client times out.....
Hmmm, I could successfully execute this test with 500000 entries. To use
more I would need to increase the heap size. Using 1000000 I received an
out of memory error. ;-) But this was expected.
Well, is that on the ESB side or the client/hessian(tomcat) side, we
store the message into a file when the size is too much, I am wondering
how ESB goes out of memory??
Also the issue with the fault seems to be due to a html response which
is not expected as a hessian fault. Is this how hessian sends faults
(i.e. as an html message), could you please attach the tcpmon message
trace of a fault from tomcat to synapse which is failing?
No problem. I will do so and get back to you as soon as possible.
I hope you can change the current implementation (error handling) in
that way, that no matter what sequence of bytes will be send as part of
a (error) response, the esb will not block and stop handling further
requests.
We can fix the issue of blocking ESB, but if the fault comes as an html
message there is no guarantee that we can pass that to the client side
because the contentType is different??? :-(
Thanks,
Ruwan
Regards,
Eric
_______________________________________________
Esb-java-dev mailing list
[email protected]
http://wso2.org/cgi-bin/mailman/listinfo/esb-java-dev
