> On Feb. 23, 2016, 10:17 p.m., Vinod Kone wrote: > > src/scheduler/scheduler.cpp, line 216 > > <https://reviews.apache.org/r/43661/diff/1/?file=1252901#file1252901line216> > > > > Subscribe contains FrameworkInfo. So this check means that a framework > > cannot update its FrameworkInfo unless it gets a disconnected callback? Is > > there a way for the scheduler force the disconnection?
As per our offline discussion: The driver also only allowed to update the `FrameworkInfo` after a scheduler failover. I would add a comment about this behavior explicitly for the library in a subsequent review. Does this sound reasonable? Feel free to reopen the issue if this does not suffice. - Anand ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/43661/#review120398 ----------------------------------------------------------- On Feb. 17, 2016, 7:22 p.m., Anand Mazumdar wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/43661/ > ----------------------------------------------------------- > > (Updated Feb. 17, 2016, 7:22 p.m.) > > > Review request for mesos and Vinod Kone. > > > Bugs: MESOS-3570 > https://issues.apache.org/jira/browse/MESOS-3570 > > > Repository: mesos > > > Description > ------- > > This change adds a check to ensure that we don't `Subscribe` again in case we > already have a `SubscribedResponse` stream set. With pipelining, we won't be > able to send more then 1 subscribe calls on the `Subscribe` connection. > > In case, the scheduler sends two subscribe calls simultaneously before the > subscribed response stream could be established. We would invoke the > `disconnected`/`connected` callback, making the scheduler subscribe again. > > > Diffs > ----- > > src/scheduler/scheduler.cpp 99a7d0dfff7b0c61decc9ff6d9e6d46ef13a7e75 > > Diff: https://reviews.apache.org/r/43661/diff/ > > > Testing > ------- > > make check > > > Thanks, > > Anand Mazumdar > >
