> On June 13, 2019, 8:35 p.m., Benjamin Mahler wrote: > > src/sched/sched.cpp > > Lines 2342-2348 (original), 2347-2369 (patched) > > <https://reviews.apache.org/r/70854/diff/1/?file=2149406#file2149406line2347> > > > > I think we want to put most of this type of logic into the > > SchedulerProcess rather than the MesosSchedulerDriver. It's already where > > we do the ID assignment. > > > > The filling in of hostname/user seems ok to leave in the scheduler > > driver since that's where it's done during initialization.
Well, my initial thinking was that it is generally better to report a local error at the call site, and not via a callback. However, after having a look into some existing framework implementations (say, Spark), I've realized that frameworks are obligated to provide an error() callback in their Scheduler implementation - and they do provide it, but they are not obligated to check the status returned by the SchedulerDriver (and they usually don't). - Andrei ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/70854/#review215887 ----------------------------------------------------------- On June 13, 2019, 7:04 p.m., Andrei Sekretenko wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/70854/ > ----------------------------------------------------------- > > (Updated June 13, 2019, 7:04 p.m.) > > > Review request for mesos and Benjamin Mahler. > > > Repository: mesos > > > Description > ------- > > This patch allows the caller to fill `FrameworkInfo.id` passed into > MesosSchedulerDriver::updateFramework() with the FrameworkID provided > by the scheduler driver. Previously the caller was required to keep > this field empty. > > > Diffs > ----- > > include/mesos/scheduler.hpp c5e61d4fd7aae5067d8c7d8dec878dc82c86e4ec > src/sched/sched.cpp 281236bb53e1c6ed77b69bf954e27705595ffb2a > > > Diff: https://reviews.apache.org/r/70854/diff/1/ > > > Testing > ------- > > > Thanks, > > Andrei Sekretenko > >
