> On Sept. 25, 2017, 11:41 p.m., Vinod Kone wrote: > > include/mesos/master/master.proto > > Lines 205-206 (patched) > > <https://reviews.apache.org/r/62475/diff/4/?file=1833352#file1833352line205> > > > > The last sentence is not strictly true right? A new agent with a new id > > but old volumes/reservations can still come back right? If so, we should > > mention that possibility.
hmm, thats possible. The reason I did not mention it was that it's not possible without the operator manually tinkering and removing the `latest` symlink. What do you think? - Anand ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/62475/#review186171 ----------------------------------------------------------- On Sept. 24, 2017, 5:06 p.m., Anand Mazumdar wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/62475/ > ----------------------------------------------------------- > > (Updated Sept. 24, 2017, 5:06 p.m.) > > > Review request for mesos and Vinod Kone. > > > Bugs: MESOS-7443 > https://issues.apache.org/jira/browse/MESOS-7443 > > > Repository: mesos > > > Description > ------- > > This change introduces the `MARK_AGENT_GONE` call that can be > used by operators to assert that a given agent has failed. It > is specially useful for stateful frameworks to ascertain whether > its safe to move the workload to a new agent. > > > Diffs > ----- > > include/mesos/master/master.proto b94e90287982e620749c10bec77cf0af10318415 > include/mesos/v1/master/master.proto > 7499fa4f62ab18dd3cd4827461717bc9c688dc49 > > > Diff: https://reviews.apache.org/r/62475/diff/4/ > > > Testing > ------- > > make check > > > Thanks, > > Anand Mazumdar > >