Even with WS-RM network failure propogated to the application
-------------------------------------------------------------
Key: CXF-1156
URL: https://issues.apache.org/jira/browse/CXF-1156
Project: CXF
Issue Type: Bug
Components: WS-* Components
Affects Versions: 2.0.2
Reporter: Bharath Ganesh
Consider the scenario:
1. WS-RM is enabled on the endpoint.
2. The client sends a request to the server endpoint(req-resp). The request
message is passed to the application at the destination end, after going
through the RM Destination interceptors. But the RM -destination interceptors
do not send the acknowledgment to the client, as the acknowledgment is
piggybacked along with the response.
3. Now assume the processing takes a longer time at the server side, the client
would get a read timeout and client thread would die.
4. But the RM timer task at the client side would wake up and send the request
again, making the server code to get executed twice.(rather many times)
Ideally:
In this case it was a read timeout exception which was easy to be replicated.
Assume it was a network failure while the response was being send from server
to client. The client application thread would get the network exception, and
die out. But the RM timer thread would try out a resend again. Ideally the
application should never get the exception, instead get the response received
by the RM-retrial timer thread should be given to the application.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
