----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/71886/#review219699 -----------------------------------------------------------
src/slave/containerizer/mesos/isolators/cgroups/subsystems/cpu.cpp Lines 107 (patched) <https://reviews.apache.org/r/71886/#comment307915> For executor containers, you have logic in `launchExecutor()` which inspects the value of `flags.cgroups_enable_cfs`, and also sets the limits equal to the resource requests when appropriate. It seems strange to me to have some of that logic in the agent code, and some of that logic in the isolators. I would expect that we either put it all in the agent, or all in the isolators. What about adding code to the `LaunchContainer` API handler which does this for tasks, and then the isolator doesn't have to care about the semantics of the `cgroups_enable_cfs` flag and the limits being set to the resource requests in some cases. - Greg Mann On Feb. 26, 2020, 12:12 p.m., Qian Zhang wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/71886/ > ----------------------------------------------------------- > > (Updated Feb. 26, 2020, 12:12 p.m.) > > > Review request for mesos, Andrei Budnik and Greg Mann. > > > Bugs: MESOS-10047 > https://issues.apache.org/jira/browse/MESOS-10047 > > > Repository: mesos > > > Description > ------- > > Set container's `cpu.cfs_quota_us` to its CPU resource limit. > > > Diffs > ----- > > src/slave/containerizer/mesos/isolators/cgroups/subsystems/cpu.cpp > 960bd141430387e076a8fab1948d07719613ed90 > > > Diff: https://reviews.apache.org/r/71886/diff/4/ > > > Testing > ------- > > > Thanks, > > Qian Zhang > >
