----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/5529/#review8495 -----------------------------------------------------------
I don't think the AsynchIO code is the correct place to be catching this exception and closing the connection. The exception handler that is currently there is there to stop code errors from propagating further and to give a clue what might have gone wrong, because we actually saw bugs like that. However IMO the codec is not supposed to throw and previously didn't. Can't you handle this the same way as the recent connection limiting code by failing authentication instead? - Andrew Stitcher On June 22, 2012, 8:18 p.m., Alan Conway wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/5529/ > ----------------------------------------------------------- > > (Updated June 22, 2012, 8:18 p.m.) > > > Review request for qpid, Andrew Stitcher and Gordon Sim. > > > Description > ------- > > A HA backup broker rejects client connections by throwing out of > ConnectionObserver::opened. > This is not an error, but generates a lot of noise in the form of error > log messages. > This patch throws a ClosedException in this case and has the exception > handler generate a debug > message instead of an error if it catches ClosedException. > > > Diffs > ----- > > /trunk/qpid/cpp/src/qpid/ha/BackupConnectionExcluder.h 1353017 > /trunk/qpid/cpp/src/qpid/ha/ConnectionObserver.cpp 1353017 > /trunk/qpid/cpp/src/qpid/sys/AsynchIOHandler.cpp 1353017 > > Diff: https://reviews.apache.org/r/5529/diff/ > > > Testing > ------- > > make check > > > Thanks, > > Alan Conway > >
