----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/18675/#review36557 -----------------------------------------------------------
Ship it! Chatted w/ BenM online. I was confused by the fact that the Registrar::recover() was needed irrespective whether MasterInfo was passed through or not. I'm fine with this change in that case. Afterthe clean up around Option<Owned<Promise<Registry> > > this is much easier to follow. - Vinod Kone On March 6, 2014, 8:12 p.m., Ben Mahler wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/18675/ > ----------------------------------------------------------- > > (Updated March 6, 2014, 8:12 p.m.) > > > Review request for mesos, Benjamin Hindman and Vinod Kone. > > > Bugs: MESOS-764 > https://issues.apache.org/jira/browse/MESOS-764 > > > Repository: mesos-git > > > Description > ------- > > As a safety measure, we would like to force a version change on the Registry > as a result of the recovery process. This helps prevent a "rogue" Master > (that still believes it is leading) from performing writes to the Registry. > > This change adds a 'Recover' operation which adds the latest MasterInfo to > the Registry. Unfortunately, since the Registrar is injected to the Master, I > had to add 'MasterInfo' as an argument to the 'recover' function, thus > altering the design somewhat. > > The current semantics are that there one must call 'recover' before calling > other operations, otherwise these other operations will fail. > > > Diffs > ----- > > src/master/registrar.hpp 20734afc69055197e9ab90d42253c56e4af4b97c > src/master/registrar.cpp 37337c07b24a96e71910b7c83085d159361a1188 > src/tests/registrar_tests.cpp 3bf42bd77a10470a2afc6fd8e1da30d6134e792c > > Diff: https://reviews.apache.org/r/18675/diff/ > > > Testing > ------- > > Added a small test. > > > Thanks, > > Ben Mahler > >
