----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/19764/#review39813 -----------------------------------------------------------
Please add 'benh' to the reviewers on this one as well. As discussed, would it be beneficial to also time individual operations on a full Registry in the benchmarks? That way, we could more closely measure the penalty we're taking from the copying in a single batch. How long does copying take for 50,000 slaves? More importantly, I'm curious how much benefit this code change is providing via the benchmarks (i.e. how many copies are avoided per batch). src/state/protobuf.hpp <https://reviews.apache.org/r/19764/#comment72451> Likely a copy was being elided here, but let's update this to a const&! src/state/protobuf.hpp <https://reviews.apache.org/r/19764/#comment72450> Ditto here. - Ben Mahler On April 1, 2014, 1:13 a.m., Jiang Yan Xu wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/19764/ > ----------------------------------------------------------- > > (Updated April 1, 2014, 1:13 a.m.) > > > Review request for mesos, Ben Mahler and Vinod Kone. > > > Bugs: MESOS-1155 > https://issues.apache.org/jira/browse/MESOS-1155 > > > Repository: mesos-git > > > Description > ------- > > See summary. > > > Diffs > ----- > > src/state/protobuf.hpp 802617b48586d8fc682d5efda3e2b86205e988b6 > src/state/state.hpp bebfe02dcdcbaec8742d173e2233817288722bd8 > > Diff: https://reviews.apache.org/r/19764/diff/ > > > Testing > ------- > > mesos-tests.sh –verbose –gtest_filter=“RegistrarTest*” > –gtest_also_run_disabled_tests > On my test machine it costs about 40-60ms to copy a registry of 20,000 slaves. > > > Thanks, > > Jiang Yan Xu > >
