----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/33759/#review82268 -----------------------------------------------------------
This is computing different statistics than I was expected, see comment below, happy to discuss further? src/slave/containerizer/isolators/cgroups/cpushare.cpp <https://reviews.apache.org/r/33759/#comment132977> This is computing the scheduling latency statistics across pids rather than across time. I was thinking it's more informative to know the scheduing latency statistics of the aggregate container during the usage() sampling interval. - Ian Downes On May 1, 2015, 10:53 a.m., Cong Wang wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/33759/ > ----------------------------------------------------------- > > (Updated May 1, 2015, 10:53 a.m.) > > > Review request for mesos, Chi Zhang, Ian Downes, and Jie Yu. > > > Bugs: MESOS-2683 > https://issues.apache.org/jira/browse/MESOS-2683 > > > Repository: mesos > > > Description > ------- > > Expose the accumulated time spent waiting on a runqueue for each task in cpu > cgroup, it is useful to monitor kernel schedule latency. This value is read > from /proc/<pid>/schedstat, kernel needs to enable CONFIG_SCHEDSTATS. > > > Diffs > ----- > > include/mesos/mesos.proto 967b1e3bbfb3f6b71d5a15d02cba7ed5ec21816f > src/slave/containerizer/isolators/cgroups/cpushare.cpp > 6a5b2b5c6e2844fe1a10815956569194b6f56681 > src/slave/flags.hpp d3b1ce117fbb4e0b97852ef150b63f35cc991032 > src/slave/flags.cpp d0932b04e3825abb6173efe0d1aee199aa356932 > > Diff: https://reviews.apache.org/r/33759/diff/ > > > Testing > ------- > > Manual tests on local host > > > Thanks, > > Cong Wang > >