> On Dec. 3, 2017, 2:33 p.m., Zhitao Li wrote: > > src/slave/slave.cpp > > Lines 6159 (patched) > > <https://reviews.apache.org/r/64267/diff/1/?file=1906550#file1906550line6159> > > > > Not necessarily this diff, but I wonder what kind of high level > > separation we want to keep between `Slave` and `Containerizer` in the long > > run. For example, the `docker_store_dir` is more of a containerizer > > internal configuration rather than logically belonging to the `Slave` class > > here. > > > > Maybe a more structured alternative is to either create the > > containerizer with an `ImageGcConfig`, or pass it down to the `pruneImages` > > API, and let `containerizer` to handle related logic? > > > > This is just food for thought for now.
Agree. We should think about this question in Unified Artifect store design. Added a TODO. - Gilbert ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/64267/#review192656 ----------------------------------------------------------- On Dec. 1, 2017, 5:10 p.m., Gilbert Song wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/64267/ > ----------------------------------------------------------- > > (Updated Dec. 1, 2017, 5:10 p.m.) > > > Review request for mesos, Jie Yu, Qian Zhang, Vinod Kone, and Zhitao Li. > > > Bugs: MESOS-8294 > https://issues.apache.org/jira/browse/MESOS-8294 > > > Repository: mesos > > > Description > ------- > > Supported basic auto image gc in the agent. > > > Diffs > ----- > > src/slave/slave.hpp 643d8559244f0842e82b293c1ef99cb26e111059 > src/slave/slave.cpp a9aa987ade67542b85aeb51356d00a4c14c6be9d > > > Diff: https://reviews.apache.org/r/64267/diff/1/ > > > Testing > ------- > > make check > > > Thanks, > > Gilbert Song > >
