> On June 7, 2013, 5:50 a.m., Benjamin Hindman wrote: > > src/master/master.cpp, line 1655 > > <https://reviews.apache.org/r/11229/diff/3/?file=301601#file301601line1655> > > > > I'm not sure I understand fully why we need to filter 'tasks' into > > 'slaveTasks'. Can't we just do the 'foreachvalue (Task* task, > > utils::copy(slave->tasks))' below and have a nested for loop which looks in > > 'tasks' to see if the slave knows about it?
that would be a O(n^2) operation instead of O(n). also, I'm not filtering, as much as converting a vector to map. am I missing something? > On June 7, 2013, 5:50 a.m., Benjamin Hindman wrote: > > src/tests/fault_tolerance_tests.cpp, line 1456 > > <https://reviews.apache.org/r/11229/diff/3/?file=301604#file301604line1456> > > > > s/PendingUpdates/IncompleteTasks/? Just going off your naming above ... thank you. forgot to change the name after we changed the semantics in the slave. > On June 7, 2013, 5:50 a.m., Benjamin Hindman wrote: > > src/tests/fault_tolerance_tests.cpp, lines 1525-1527 > > <https://reviews.apache.org/r/11229/diff/3/?file=301604#file301604line1525> > > > > To really be sure about this you'll want to pause before you post the > > NewMasterDetectedMessage. N/A since the retried TASK_FINISHED ensures the receipt of TASK_LOST, if any. > On June 7, 2013, 5:50 a.m., Benjamin Hindman wrote: > > src/tests/fault_tolerance_tests.cpp, line 1501 > > <https://reviews.apache.org/r/11229/diff/3/?file=301604#file301604line1501> > > > > Ideally we just drop the one, and the test later shows the > > TASK_FINISHED makes its way through ... done - Vinod ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/11229/#review21561 ----------------------------------------------------------- On June 7, 2013, 8:37 p.m., Vinod Kone wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/11229/ > ----------------------------------------------------------- > > (Updated June 7, 2013, 8:37 p.m.) > > > Review request for mesos, Benjamin Hindman and Ben Mahler. > > > Description > ------- > > Reopened the discarded review (includes refactor of task reconciliation) to > fix task reconciliation. > > > This addresses bug MESOS-447. > https://issues.apache.org/jira/browse/MESOS-447 > > > Diffs > ----- > > src/master/master.hpp 8e7b74cccf46bea8dd5cb8543bf083aed64d370a > src/master/master.cpp a2e4b905f1ef5c00560917c133b27ac978988807 > src/slave/slave.hpp 26dc96e5f2fdc0711fc49a9ea80b7f037bf93c29 > src/slave/slave.cpp 8ce1646f2f804bc8dc20506d11078191f0274654 > src/tests/fault_tolerance_tests.cpp > ef570b7e4b61df5e452628a5ea7c75888a0797ec > > Diff: https://reviews.apache.org/r/11229/diff/ > > > Testing > ------- > > make check > > > Thanks, > > Vinod Kone > >
