> On Oct. 24, 2016, 5:21 p.m., Neil Conway wrote: > > src/slave/slave.cpp, line 787 > > <https://reviews.apache.org/r/53067/diff/2/?file=1544356#file1544356line787> > > > > I wanted to avoid using "terminating" because that describes a > > particular state in the agent's state machine; the agent might be in > > `finalize()` without `state == TERMINATING`.
"terminating" was in reference to the fact that `finalize` gets called when `terminate` libprocess event is sent to the actor, not in relation to the state machine. This is consistent with how we log `Master::finalize`. Also note that "shutting down..." already happens and makes more sense inside `Slave::shutdown` message because it refers to the fact that it received a shutdown event. - Vinod ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/53067/#review153709 ----------------------------------------------------------- On Oct. 24, 2016, 5:20 p.m., Neil Conway wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/53067/ > ----------------------------------------------------------- > > (Updated Oct. 24, 2016, 5:20 p.m.) > > > Review request for mesos and Vinod Kone. > > > Repository: mesos > > > Description > ------- > > When an agent ID is terminated (e.g., due to "--recover=cleanup"), it > seems important to notify the operator that the agent will be assigned a > new agent ID when it is next started. > > > Diffs > ----- > > src/slave/slave.cpp 881c10ac61d1e4fdeabdc42c0a41508c36f49040 > > Diff: https://reviews.apache.org/r/53067/diff/ > > > Testing > ------- > > Visual inspection of log output. > > > Thanks, > > Neil Conway > >
