----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/32843/#review80853 -----------------------------------------------------------
src/master/master.cpp <https://reviews.apache.org/r/32843/#comment130964> It looks like we can do this without having the extra `killTask` method: just have this convert to the `Kill` `Call` and call `kill`. This is the aproach we used in `launchTasks` as well by converting to an `Accept` `Call`, so seems nice to keep things consistent? src/master/master.cpp <https://reviews.apache.org/r/32843/#comment130968> TASK_ERROR is a terminal state, right? What does it mean to send TASK_ERROR but leave the task running? For example, a reconciliation request with the wrong slave id will _not_ reply with TASK_ERROR like we do here, since we ignore the slave id when we can find the task. - Ben Mahler On April 20, 2015, 8:02 p.m., Vinod Kone wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/32843/ > ----------------------------------------------------------- > > (Updated April 20, 2015, 8:02 p.m.) > > > Review request for mesos and Ben Mahler. > > > Bugs: MESOS-1127 > https://issues.apache.org/jira/browse/MESOS-1127 > > > Repository: mesos > > > Description > ------- > > Having SlaveID will help us do better reconciliation when the task is not > found. Also, updated master to handle Call::Kill. > > > Diffs > ----- > > include/mesos/scheduler/scheduler.proto > 783a63ad1cc0edd86605d638046fb959cb6e97e8 > src/master/master.hpp c10e7c08c191acef9d31d98018a47a2a952a4dfc > src/master/master.cpp e30b951eda2b3b0d5b2a80716f0b32c6bbe041bc > src/scheduler/scheduler.cpp 584b042e32865fdf875bf41ebcfb7f9c327d882a > src/tests/scheduler_tests.cpp 4a89a7a88b50bb8c254f5076661ce07ac9fc7657 > > Diff: https://reviews.apache.org/r/32843/diff/ > > > Testing > ------- > > make check > > > Thanks, > > Vinod Kone > >
