On 30.06.2014 20:26, Mark Phippard wrote: > On Mon, Jun 30, 2014 at 1:00 PM, Darly Senecal Baptiste > <dsenec...@gmail.com <mailto:dsenec...@gmail.com>> wrote: > > Looking at the network trace packet wireshark PoV, looks like that > the server tried to send data to the client (javahl), but saw that > client's data buffer window is full (like ZeroWindow packet) and > client is sending a zero window to the server. And finally, the > client is sending a series of RESET packets to server. And these > RST packets appears when javaHL crashed > > > If JVM is crashing it should be producing a dump. Find it and send it. > > Also, as I mentioned earlier, you should see if you can recreate the > crash using the command line.
Not necessarily. I fixed a bug in JavaHL just today that was specific to the JavaHL native code and was not triggered by a similar code path used by the command-line client. That bug wasn't in any way related to this problem, though; it was in trunk JavaHL and not in any released version; but that doesn't mean there aren't any such bugs in released versions of JavaHL, too. -- Brane -- Branko Čibej | Director of Subversion WANdisco // Non-Stop Data e. br...@wandisco.com
