At the server end, the request was accepted successfully and processed like any other request. As far as the timing is concerned, which we thought to be culprit here, there are some weird finding. As soon as the request is fired, it's registered in the weblogic access log at server end. However, there's 15 second delay between this and the time when the first service log appears in application log file.
The whole request takes about 2 second, but it's after another 43 seconds that we got this error in calling system. It's a one-off error, which we have never seen before. Some reading tells me that weblogic was asked to something impossible and it gave up. Some other sources suggests that it might be due to some bug in weblogic. Any ideas about this one?