----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/20958/#review43447 -----------------------------------------------------------
src/slave/containerizer/mesos_containerizer.cpp <https://reviews.apache.org/r/20958/#comment77601> O_NONBLOCK is not set on these, right? :) src/slave/containerizer/mesos_containerizer.cpp <https://reviews.apache.org/r/20958/#comment77604> Is there a "race" or interaction between this code, and the ABORT() calls? Possibly an interaction between these operations and libc/libc++ stdio? setvbuf()? Unsure. src/slave/containerizer/mesos_containerizer.cpp <https://reviews.apache.org/r/20958/#comment77602> Should contain a mode (0666). src/slave/containerizer/mesos_containerizer.cpp <https://reviews.apache.org/r/20958/#comment77603> Ditto - Tobias Weingartner On May 19, 2014, 10:57 p.m., Ian Downes wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/20958/ > ----------------------------------------------------------- > > (Updated May 19, 2014, 10:57 p.m.) > > > Review request for mesos, Ben Mahler, Jie Yu, and Vinod Kone. > > > Repository: mesos-git > > > Description > ------- > > We *may* have observed deadlocks in the forked child when testing some other > new code so this makes the execute() function in MesosContainerizer > completely async-signal safe. > > Reviewers: I'd appreciate it if you could carefully check that I've made this > completely async signal safe. > > > Replace all stout calls with just the appropriate system call. > > Construct Envp and pass into child and use execle for the environment. > > Use open and dup2 rather than freopen. > > Determine gid and uid and pass into child. > > > Diffs > ----- > > src/slave/containerizer/mesos_containerizer.cpp > 2a4816ec34f2298b71e8591c1e92d77f94ef4cb3 > > Diff: https://reviews.apache.org/r/20958/diff/ > > > Testing > ------- > > make check > > > Thanks, > > Ian Downes > >
