----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/16150/#review30238 -----------------------------------------------------------
benh and I went over this today, this is a partial review. src/slave/container/isolator.hpp <https://reviews.apache.org/r/16150/#comment57886> Rather than using an Option with None() as a sentinel, can this take the full state and perform any cleanup as part of the single call to recover()? src/slave/container/isolator.hpp <https://reviews.apache.org/r/16150/#comment57890> Why does this return a string script? Couldn't the script be executed by the isolator during the isolate() call? src/slave/container/isolator.hpp <https://reviews.apache.org/r/16150/#comment57887> ExecutorInfo contains resources, why are resources being passed separately here? src/slave/container/isolator.hpp <https://reviews.apache.org/r/16150/#comment57888> Cleanup? Destroy seems to imply that this is destroying the underlying processes. src/slave/container/isolator.hpp <https://reviews.apache.org/r/16150/#comment57889> In general we place the API (in this case: Isolator) above the libprocess Process. src/slave/container/isolator.cpp <https://reviews.apache.org/r/16150/#comment57891> CHECK_NOTNULL(process.get()) src/slave/container/isolators/cpu/posix.cpp <https://reviews.apache.org/r/16150/#comment57897> Can this use pstree instead? (pstree didn't exist when I first implemented posix usage collection). There seems to be a lot of redundant code in order to split apart the cpu and memory usage() reporting for posix. Is there a major gain from doing this? I would prefer to see less code by avoiding the posix inheritance. src/slave/container/isolators/posix.hpp <https://reviews.apache.org/r/16150/#comment57895> Please use virtual to signify that this is a virtual function that you are overloading, here and elsewhere. src/slave/container/isolators/posix.hpp <https://reviews.apache.org/r/16150/#comment57892> Can this return a Failure instead? src/slave/container/isolators/posix.hpp <https://reviews.apache.org/r/16150/#comment57893> return Future<Nothing>(); src/slave/container/isolators/posix.hpp <https://reviews.apache.org/r/16150/#comment57896> You don't need this, unless you want this to show that it is unimplemented? src/slave/container/isolators/posix.hpp <https://reviews.apache.org/r/16150/#comment57894> No need for a promise, see above. Also, the downside of a promise is that if a single person discards the future then all clients of watch() will get the discarded future, whereas with returning the Future each client has a different Future. - Ben Mahler On Dec. 11, 2013, 4:09 a.m., Ian Downes wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/16150/ > ----------------------------------------------------------- > > (Updated Dec. 11, 2013, 4:09 a.m.) > > > Review request for mesos, Benjamin Hindman, Ben Mahler, Chi Zhang, Niklas > Nielsen, samya, and Jason Dusek. > > > Repository: mesos-git > > > Description > ------- > > Isolators perform isolator for the MesosContainerizer > > Isolator interface and implementations of Posix CPU and Mem isolators (no > isolation, just usage()) > > > Diffs > ----- > > src/slave/container/isolator.hpp PRE-CREATION > src/slave/container/isolator.cpp PRE-CREATION > src/slave/container/isolators/cpu/posix.hpp PRE-CREATION > src/slave/container/isolators/cpu/posix.cpp PRE-CREATION > src/slave/container/isolators/mem/posix.hpp PRE-CREATION > src/slave/container/isolators/mem/posix.cpp PRE-CREATION > src/slave/container/isolators/posix.hpp PRE-CREATION > > Diff: https://reviews.apache.org/r/16150/diff/ > > > Testing > ------- > > > Thanks, > > Ian Downes > >
