----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/64806/#review194462 -----------------------------------------------------------
src/resource_provider/http_connection.hpp Lines 167 (patched) <https://reviews.apache.org/r/64806/#comment273226> This migh still trigger `detected` depending on how `detector->detect` is implemented. That means `detected` might be called multiple times? I'd suggest we fix ConstantEndpointDetector to properly handle discard. - Jie Yu On Dec. 22, 2017, 1:36 p.m., Benjamin Bannier wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/64806/ > ----------------------------------------------------------- > > (Updated Dec. 22, 2017, 1:36 p.m.) > > > Review request for mesos, Jie Yu and Jan Schlicht. > > > Bugs: MESOS-8349 > https://issues.apache.org/jira/browse/MESOS-8349 > > > Repository: mesos > > > Description > ------- > > The expectation for disconnection of the resource provider driver is > that disconnection handlers of the resource provider would be invoked > and a new connection would be detected. > > Since the existing implementation did not account for the possibility of > a remote endpoint already being detected, the code did now show the > expected behavior. This is fixed in this patch by explicitly triggering > an endpoint redetection if a disconnection was detected. We also modify > the driver's 'start' method to implicitly discard any pending endpoint > detections. > > > Diffs > ----- > > src/resource_provider/http_connection.hpp > 3d5088dc1eeeb64a932a731d10e4c12564e79034 > src/tests/resource_provider_manager_tests.cpp > f58ab6b4c98eff92ae23acacefbcb9066b5ace91 > > > Diff: https://reviews.apache.org/r/64806/diff/1/ > > > Testing > ------- > > `make check` > > > Thanks, > > Benjamin Bannier > >