> On March 15, 2016, 11:34 p.m., Ben Mahler wrote: > > Looking pretty good. Would be great to have a CHANGELOG update here that > > outlines the deprecation and what we're advising users to do in each > > version.
I'll update the changelog in the next patch, hope this is fine. > On March 15, 2016, 11:34 p.m., Ben Mahler wrote: > > src/docker/executor.hpp, line 74 > > <https://reviews.apache.org/r/44661/diff/4/?file=1299831#file1299831line74> > > > > Just to be clear, is the intention to remove this in 0.30? If so, can > > you say that explicitly? Or are we giving time for users to set kill > > policies and planning to remove it 6 months from 0.29.0? I think our *current* strategy is to remove in 6 months. That's why I put the version when we started the deprecation cycle. > On March 15, 2016, 11:34 p.m., Ben Mahler wrote: > > src/slave/flags.cpp, line 532 > > <https://reviews.apache.org/r/44661/diff/4/?file=1299834#file1299834line532> > > > > poly? Yeah, a small pony sneaked in : ). > On March 15, 2016, 11:34 p.m., Ben Mahler wrote: > > src/slave/containerizer/docker.cpp, line 967 > > <https://reviews.apache.org/r/44661/diff/4/?file=1299832#file1299832line967> > > > > Hm.. I'm a bit confused about the deprecation taking place here. Could > > you outline the steps in this description? Specifically, what does the > > 0.28, 0.29, and final versions look like and what are we advising users to > > do in each version? > > > > This information would be great to have in the CHANGELOG deprecation > > message I suggested above. I think we can remove this timeout now, but it's hard to say whether there are any users relying on the unfortunate fact that the containerizer does not kill immediately. Hence I decided to remove it via a deprecation cycle. On the other side, deprecating a bug is a bit strange. What do you think? - Alexander ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/44661/#review123787 ----------------------------------------------------------- On March 15, 2016, 2:28 p.m., Alexander Rukletsov wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/44661/ > ----------------------------------------------------------- > > (Updated March 15, 2016, 2:28 p.m.) > > > Review request for mesos, Ben Mahler and Timothy Chen. > > > Bugs: MESOS-4910 > https://issues.apache.org/jira/browse/MESOS-4910 > > > Repository: mesos > > > Description > ------- > > Instead, a combination of `executor_shutdown_grace_period` > agent flag and task kill policies should be used. > > > Diffs > ----- > > docs/configuration.md 739d4ff9aeeb1ba70ce11033168d63d37b6ef56b > src/docker/executor.hpp abbc419533ab40312e917931a2fc2ce78b38da41 > src/slave/containerizer/docker.cpp fb9188a19a5cd8211d4f36f9647ebb70de560109 > src/slave/flags.hpp feb095da4521f678c96f4cc53bdfda262d350388 > src/slave/flags.cpp 4d10818105627738e258116647ccada374e3d7b9 > > Diff: https://reviews.apache.org/r/44661/diff/ > > > Testing > ------- > > The complete chain was tested. See https://reviews.apache.org/r/44662/. > > > Thanks, > > Alexander Rukletsov > >
