> On Dec. 11, 2012, 10:44 p.m., Benjamin Hindman wrote: > > src/slave/slave.cpp, line 104 > > <https://reviews.apache.org/r/8171/diff/4/?file=235830#file235830line104> > > > > The problem with doing these checks is that it makes it hard to create > > "simulations" of clusters where some slaves might be reporting more CPUs > > than are physically available. > > Ben Mahler wrote: > Yes, that's a good point. Perhaps we should just isolate the check to the > cgroups isolation module initialization. > > i.e. when cpuset is enabled, ensure the slave doesn't think there are > more cpus than available (this is cause a CHECK to fail when filling up the > cpusets if it doesn't hold)
This is exactly what I was thinking. - Benjamin ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/8171/#review14320 ----------------------------------------------------------- On Dec. 7, 2012, 7:25 p.m., Vinod Kone wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/8171/ > ----------------------------------------------------------- > > (Updated Dec. 7, 2012, 7:25 p.m.) > > > Review request for mesos, Benjamin Hindman and Ben Mahler. > > > Description > ------- > > Now we properly set cpu/mem/disk/ports based on what's missing in slave flags. > > Also moved usage from os to fs. > > > Diffs > ----- > > src/slave/slave.cpp 7deb4574943aae4cfc5da5d6b3f600042686975f > src/tests/exception_tests.cpp 13355d08788432ed07679daf24c2d74cc12a7f11 > third_party/libprocess/include/stout/fs.hpp > 9e62a1b91bc9fac092818ffb3c8bcec46b0bd26d > third_party/libprocess/include/stout/os.hpp > 76e5e0624af36a0021755fb4acf7f76bfb81a823 > > Diff: https://reviews.apache.org/r/8171/diff/ > > > Testing > ------- > > make check > > > Thanks, > > Vinod Kone > >
