----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/54524/#review158568 -----------------------------------------------------------
Looks good, is it possible to write this patch such that it continues to work for single role frameworks? That way we could land this and have a DISABLED test for this on the multi-role side without breaking the single role framework case. src/master/master.hpp (line 2703) <https://reviews.apache.org/r/54524/#comment229352> Can you use = delete here? - Benjamin Mahler On Dec. 8, 2016, 7:10 a.m., Jay Guo wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/54524/ > ----------------------------------------------------------- > > (Updated Dec. 8, 2016, 7:10 a.m.) > > > Review request for mesos, Benjamin Mahler and Guangya Liu. > > > Bugs: MESOS-6685 > https://issues.apache.org/jira/browse/MESOS-6685 > > > Repository: mesos > > > Description > ------- > > Currently `resources()` method in `struct Role` of master.hpp returns > totalUsedResources and totalOfferedResources for a framework, which is > inaccurate for multi-role frameworks. A filter is added to filter out > actual resources for specific role in a framework. > > > Diffs > ----- > > src/master/master.hpp d3c20d7cbb9137d3b05cae3d67f004d5e7475c8a > src/master/master.cpp 353e6ea802e197b4456c1647f78d9984a50f1c9d > > Diff: https://reviews.apache.org/r/54524/diff/ > > > Testing > ------- > > This patch is just to showcase a possible solution for accounting resources > used by a role in multi-role framework. It fails certain test cases since > `allocation_info` is not filled yet. > > > Thanks, > > Jay Guo > >
