I am running RPT 188.8.131.52 on Redhat Linux on a machine with 16 GB of RAM. RPT is configured to use up to 8 GB of RAM (-Xmx8192m). I'm using the generic service client to record a large web service transaction. Based on testing the same transaction in SoapUI, I expect the transaction to take about six minutes, and return between five and ten megabytes of XML in the response.
When I select "Invoke", the Sending Request message appears and, for about 5 minutes, the UI is still responsive and CPU utilization is around 30% of one CPU core. Monitoring the process, memory utilization is about 600 MB during this phase. Then the UI freezes (greys out), the sliding status bar stops, and CPU utilization for the single core hits 100% and stays there. Memory utilization jumps to 1.8 GB and remains constant.
Oddly enough it appears that the service being exercised is actually completing successfully. At any time after the initial 5 minutes or so, before the UI freezes, I can check the database the request is updating and see that the target records have been updated as expected. So it appears that the generic service client is freezing/failing while receiving and parsing the response.
I have allowed the process to run in excess of two hours on several occasions. In one instance, the UI "unfroze" with a garbled result. When I clicked anywhere on the RPT interface it froze again.
This *looks* like a memory problem, but the RPT process seems to have allocated nowhere near the maximum amount of RAM it is provisioned. Is there another setting somewhere to allocate memory to the Generic service client? Are there some other settings I can tweek? Or is this a known bug?