----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/68131/#review206962 -----------------------------------------------------------
Just a high level comment, it's unfortunate that this test depends so much on timing w.r.t. to the 200ms polling interval. For example, if we actually serve the first request in less than 200ms then we wouldn't be measuring the benefits of batching. An idea for how to fix this: * Issue batches, e.g. N requests at once, when they all finish, another N requests, only need to repeat a few times. This means that the benchmark will always show how batching can help high load, and we're not assuming requests take longer than e.g. 200ms to get processed on the master. We don't need an interval here because we can just proceed with the next poll as soon as everything finished from the first one. src/tests/master_benchmarks.cpp Lines 512 (patched) <https://reviews.apache.org/r/68131/#comment290128> The review summary needs to updated to reflect the renaming? - Benjamin Mahler On Aug. 6, 2018, 10:30 a.m., Alexander Rukletsov wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/68131/ > ----------------------------------------------------------- > > (Updated Aug. 6, 2018, 10:30 a.m.) > > > Review request for mesos, Benno Evers and Benjamin Mahler. > > > Bugs: MESOS-8975 > https://issues.apache.org/jira/browse/MESOS-8975 > > > Repository: mesos > > > Description > ------- > > See summary. > > > Diffs > ----- > > src/tests/master_benchmarks.cpp b6d6dc7c1752491e2da854018966374b624d6682 > > > Diff: https://reviews.apache.org/r/68131/diff/2/ > > > Testing > ------- > > See https://reviews.apache.org/r/68132/ > > > Thanks, > > Alexander Rukletsov > >
