Re: Review Request 50187: Removed the `os::sleep` from `Clock::settle`.
--- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/50187/#review142778 --- Ship it! Thanks! - Benjamin Mahler On July 19, 2016, 10:08 a.m., Neil Conway wrote: > > --- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/50187/ > --- > > (Updated July 19, 2016, 10:08 a.m.) > > > Review request for mesos and Benjamin Mahler. > > > Bugs: MESOS-3760 > https://issues.apache.org/jira/browse/MESOS-3760 > > > Repository: mesos > > > Description > --- > > `Clock::settle` is used to wait until all pending libprocess events > have been handled. Test cases should typically use it only when there is > no other way to achieve the proper synchronization between two events. > > Previously, `Clock::settle` also contained an `os::sleep` call to > workaround broken test cases that assumed `settle` provided stronger > guarantees than described above (e.g., some test cases assumed that > doing `http::get` followed by a `Clock::settle` ensured that the remote > side of the HTTP connection will have seen the request). Currently, > there are relatively few such test cases, so it is better to fix those > test cases (or add a `sleep` call to them) and remove the `sleep` from > `Clock::settle`. > > > Diffs > - > > 3rdparty/libprocess/src/process.cpp > 9661386afd4fddd1877d55941fa403afc9230280 > 3rdparty/libprocess/src/tests/metrics_tests.cpp > 5a82f4f49aecd03d12687de629516be5b7895036 > > Diff: https://reviews.apache.org/r/50187/diff/ > > > Testing > --- > > `make check` with lots of iterations. > > > Thanks, > > Neil Conway > >
Re: Review Request 50187: Removed the `os::sleep` from `Clock::settle`.
--- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/50187/#review142755 --- Patch looks great! Reviews applied: [50187] Passed command: export OS='ubuntu:14.04' BUILDTOOL='autotools' COMPILER='gcc' CONFIGURATION='--verbose' ENVIRONMENT='GLOG_v=1 MESOS_VERBOSE=1'; ./support/docker_build.sh - Mesos ReviewBot On July 19, 2016, 10:08 a.m., Neil Conway wrote: > > --- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/50187/ > --- > > (Updated July 19, 2016, 10:08 a.m.) > > > Review request for mesos and Benjamin Mahler. > > > Bugs: MESOS-3760 > https://issues.apache.org/jira/browse/MESOS-3760 > > > Repository: mesos > > > Description > --- > > `Clock::settle` is used to wait until all pending libprocess events > have been handled. Test cases should typically use it only when there is > no other way to achieve the proper synchronization between two events. > > Previously, `Clock::settle` also contained an `os::sleep` call to > workaround broken test cases that assumed `settle` provided stronger > guarantees than described above (e.g., some test cases assumed that > doing `http::get` followed by a `Clock::settle` ensured that the remote > side of the HTTP connection will have seen the request). Currently, > there are relatively few such test cases, so it is better to fix those > test cases (or add a `sleep` call to them) and remove the `sleep` from > `Clock::settle`. > > > Diffs > - > > 3rdparty/libprocess/src/process.cpp > 9661386afd4fddd1877d55941fa403afc9230280 > 3rdparty/libprocess/src/tests/metrics_tests.cpp > 5a82f4f49aecd03d12687de629516be5b7895036 > > Diff: https://reviews.apache.org/r/50187/diff/ > > > Testing > --- > > `make check` with lots of iterations. > > > Thanks, > > Neil Conway > >
Review Request 50187: Removed the `os::sleep` from `Clock::settle`.
--- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/50187/ --- Review request for mesos and Benjamin Mahler. Bugs: MESOS-3760 https://issues.apache.org/jira/browse/MESOS-3760 Repository: mesos Description --- `Clock::settle` is used to wait until all pending libprocess events have been handled. Test cases should typically use it only when there is no other way to achieve the proper synchronization between two events. Previously, `Clock::settle` also contained an `os::sleep` call to workaround broken test cases that assumed `settle` provided stronger guarantees than described above (e.g., some test cases assumed that doing `http::get` followed by a `Clock::settle` ensured that the remote side of the HTTP connection will have seen the request). Currently, there are relatively few such test cases, so it is better to fix those test cases (or add a `sleep` call to them) and remove the `sleep` from `Clock::settle`. Diffs - 3rdparty/libprocess/src/process.cpp 9661386afd4fddd1877d55941fa403afc9230280 3rdparty/libprocess/src/tests/metrics_tests.cpp 5a82f4f49aecd03d12687de629516be5b7895036 Diff: https://reviews.apache.org/r/50187/diff/ Testing --- `make check` with lots of iterations. Thanks, Neil Conway