----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/44790/#review125510 -----------------------------------------------------------
Fix it, then Ship it! src/linux/systemd.cpp (lines 73 - 74) <https://reviews.apache.org/r/44790/#comment188328> let's add a `NOTE` here explaining that returning an error implies the child will be killed. I know you added this to the header, but it's great to have it with the implementation. src/linux/systemd.cpp (line 81) <https://reviews.apache.org/r/44790/#comment188327> `is not configured as enabled` - Joris Van Remoortere On March 14, 2016, 3:44 p.m., Joerg Schad wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/44790/ > ----------------------------------------------------------- > > (Updated March 14, 2016, 3:44 p.m.) > > > Review request for mesos and Joris Van Remoortere. > > > Bugs: MESOS-4839 > https://issues.apache.org/jira/browse/MESOS-4839 > > > Repository: mesos > > > Description > ------- > > Previously the Error behavior (in particular where the child process is > killed) was not consistent among Error cases. > We should establish the convention that the Hook itself returns an error, > while the subprocess code is resposible for > killing the child process in case of such Error. > > > Diffs > ----- > > src/linux/systemd.hpp 6b240b9cc2fe559d111a4235e70f00e07f3d4495 > src/linux/systemd.cpp e120d2ceffe6bb400859f2a132f02a99f561f856 > > Diff: https://reviews.apache.org/r/44790/diff/ > > > Testing > ------- > > sudo make check (on Linux) > > > Thanks, > > Joerg Schad > >
