----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/38051/#review99453 -----------------------------------------------------------
Can you write a test for this? src/master/master.cpp (lines 5552 - 5554) <https://reviews.apache.org/r/38051/#comment156341> This comment is a bit confusing. The master receiving a second terminal update doesn't mean the slave sent it by mistake. It could happen if the executor sends it too. I would change this as follows // Once a task's state has been transitioned to terminal state, no further // terminal updates should result in a state change. These are the same // semantics that are enforced by the slave. - Vinod Kone On Sept. 5, 2015, 11:41 p.m., Yong Qiao Wang wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/38051/ > ----------------------------------------------------------- > > (Updated Sept. 5, 2015, 11:41 p.m.) > > > Review request for mesos and Vinod Kone. > > > Bugs: MESOS-2864 > https://issues.apache.org/jira/browse/MESOS-2864 > > > Repository: mesos > > > Description > ------- > > Only update the task status when its old status is not terminal. > > > Diffs > ----- > > src/master/master.cpp 5589eca > > Diff: https://reviews.apache.org/r/38051/diff/ > > > Testing > ------- > > UT: > > 1. make successfully! > 2. make check successfully! > 3. Run test framework successfully! > > > Thanks, > > Yong Qiao Wang > >