----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/10109/#review18335 -----------------------------------------------------------
src/slave/slave.cpp <https://reviews.apache.org/r/10109/#comment38542> Since the slave wouldn't register with master until after all executors have been re-registered/cleaned up, I wasn't expecting a framework to be shutdown in the middle of recovery. In general, the master shouldn't be sending messages to an un-registered slave. The only scenario I can think of, is when the master sends a message to the old slave, but was received by the new slave (as seen in MESOS-365). I think we should guard against these messages in the slave. - Vinod Kone On March 25, 2013, 1:47 a.m., Benjamin Hindman wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/10109/ > ----------------------------------------------------------- > > (Updated March 25, 2013, 1:47 a.m.) > > > Review request for mesos and Vinod Kone. > > > Description > ------- > > This is just part of a chain ... probably won't be a commit in the end. > > > Diffs > ----- > > src/slave/slave.cpp 091ec5ed19924aef31b761e68b70b8d042f9a9b7 > > Diff: https://reviews.apache.org/r/10109/diff/ > > > Testing > ------- > > N/A > > > Thanks, > > Benjamin Hindman > >
