----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/52174/#review150127 -----------------------------------------------------------
src/master/master.cpp (line 5925) <https://reviews.apache.org/r/52174/#comment217995> No need for this hack. You can look in the `recovered` map for the FrameworkInfo of a recovered framework that has not resubscribed yet. - Vinod Kone On Sept. 22, 2016, 9:12 p.m., Neil Conway wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/52174/ > ----------------------------------------------------------- > > (Updated Sept. 22, 2016, 9:12 p.m.) > > > Review request for mesos and Vinod Kone. > > > Bugs: MESOS-6226 > https://issues.apache.org/jira/browse/MESOS-6226 > > > Repository: mesos > > > Description > ------- > > We previously assumed that when marking a task unreachable, we would > have access to the `FrameworkInfo` for that task's framework. However, > that is not the case if the master has failed over and the framework has > not yet reregistered with the new master. > > In that situation, we don't have the framework's FrameworkInfo, and > hence we cannot determine if the framework is partition-aware. For now, > we assume the framework is partition-aware, which isn't great behavior > (MESOS-6232). > > > Diffs > ----- > > src/master/master.cpp 66a672f6d16233e96b29e330a9e6c474546fa851 > src/tests/partition_tests.cpp 7c38f0efa414447e6292b2d6b334fb9879c92eb5 > > Diff: https://reviews.apache.org/r/52174/diff/ > > > Testing > ------- > > `make check` on OSX. > > > Thanks, > > Neil Conway > >