> On 2012-05-28 08:09:05, Gordon Sim wrote: > > Diff 3 seems to revert (as compared with diff 2) the correct dequeing of > > removed messages; was that intentional?
Yes. We are now assuming the caller has dequeued any messages after the trunc position, so we don't do this potentially expensive check again. - Alan ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/5231/#review8142 ----------------------------------------------------------- On 2012-05-25 17:55:24, Alan Conway wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/5231/ > ----------------------------------------------------------- > > (Updated 2012-05-25 17:55:24) > > > Review request for qpid, Gordon Sim and Kenneth Giusti. > > > Summary > ------- > > In the new HA code a backup may sometimes be ahead of the new primary > after a > fail-over. In that case the backup truncates it's queues to the same > position > as the primary so it can continue replicating. > > > Diffs > ----- > > /trunk/qpid/cpp/src/qpid/broker/MessageDeque.h 1342442 > /trunk/qpid/cpp/src/qpid/broker/MessageDeque.cpp 1342442 > /trunk/qpid/cpp/src/qpid/broker/MessageMap.h 1342442 > /trunk/qpid/cpp/src/qpid/broker/MessageMap.cpp 1342442 > /trunk/qpid/cpp/src/qpid/broker/Messages.h 1342442 > /trunk/qpid/cpp/src/qpid/broker/PriorityQueue.h 1342442 > /trunk/qpid/cpp/src/qpid/broker/PriorityQueue.cpp 1342442 > /trunk/qpid/cpp/src/qpid/broker/Queue.h 1342442 > /trunk/qpid/cpp/src/qpid/broker/Queue.cpp 1342442 > /trunk/qpid/cpp/src/tests/Makefile.am 1342442 > /trunk/qpid/cpp/src/tests/QueueTest.cpp 1342442 > /trunk/qpid/cpp/src/tests/brokertest.py 1342442 > > Diff: https://reviews.apache.org/r/5231/diff > > > Testing > ------- > > Passing make check, added unit tests to QueueTests.cpp > > > Thanks, > > Alan > >
