----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/11131/#review21893 -----------------------------------------------------------
src/linux/cgroups.cpp <https://reviews.apache.org/r/11131/#comment45187> On second thought, let's just keep it simple for now as I imagine it's alright to freeze an empty cgroup. So stop -> kill -> empty -> freeze -> kill -> thaw -> empty. Then we don't need to be concerned with only continuing the chain when the kill doesn't work. Disregard SIGCONT, I was mislead by some comments in killtree.sh but it's not necessary when using SIGKILL. I apologize for the conflicting feedback!! To be clear, we haven't made any progress either in terms of figuring out what the problem is so we're agreed on this fix! :) - Ben Mahler On June 14, 2013, 1:14 a.m., Brenden Matthews wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/11131/ > ----------------------------------------------------------- > > (Updated June 14, 2013, 1:14 a.m.) > > > Review request for mesos. > > > Description > ------- > > Changed cgroups killTasks() sequence. > > The sequence was broken into 2 steps: > 1. stop, kill, continue; then > 2. freeze, kill, thaw (if not empty) > > Review: https://reviews.apache.org/r/11131 > > > Diffs > ----- > > src/linux/cgroups.cpp 8d94fe63610c4c7a48f92d260fcd526b7a83942e > > Diff: https://reviews.apache.org/r/11131/diff/ > > > Testing > ------- > > Used in production at airbnb. > > make -j10 check && cd hadoop && make hadoop-2.0.0-mr1-cdh4.2.1 && make > hadoop-0.20.205.0 && make hadoop-0.20.2-cdh3u3 > > > Thanks, > > Brenden Matthews > >
