> On March 3, 2016, 1:37 a.m., Vinod Kone wrote: > > src/master/http.cpp, lines 458-462 > > <https://reviews.apache.org/r/44278/diff/3/?file=1278474#file1278474line458> > > > > should these be BadRequest as well?
I think `Forbidden` is more apt here. _There is nothing wrong with the request object per se i.e. being malformed etc._ In both the cases, the framework is trying to perform an operation that it is not allowed to do i.e. trying to send a non-subscribe call without subscribing/trying to send a non-subscribe call when previously subscribed as a driver based framework. - Anand ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/44278/#review121756 ----------------------------------------------------------- On March 2, 2016, 10:36 p.m., Greg Mann wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/44278/ > ----------------------------------------------------------- > > (Updated March 2, 2016, 10:36 p.m.) > > > Review request for mesos, Anand Mazumdar and Vinod Kone. > > > Bugs: MESOS-3583 > https://issues.apache.org/jira/browse/MESOS-3583 > > > Repository: mesos > > > Description > ------- > > Added HTTP scheduler stream IDs. > > In some failure scenarios involving highly-available HTTP schedulers with > multiple instances, it's possible for a non-leading instance to successfully > make HTTP calls to the master. This patch enables the master to use HTTP > scheduler stream IDs to uniquely identify each HTTP subscription stream, > preventing any non-leading scheduler instance from making calls to the > master. The patch also adds stream ID support to the HTTP scheduler library. > > > Diffs > ----- > > src/master/http.cpp 5e9e28e904ba0045ee27eb828f47231632a91d74 > src/master/master.hpp 13c6ff153e77c527822309e787942eb463d59e7d > src/scheduler/scheduler.cpp 7ea1c2567f37a73160bca346a25bb2f0c54e71a0 > > Diff: https://reviews.apache.org/r/44278/diff/ > > > Testing > ------- > > `make check` was used to test on both OSX and CentOS 7.1. > > > Thanks, > > Greg Mann > >