> On Nov. 11, 2014, 1:24 p.m., Alexander Rukletsov wrote: > > src/slave/flags.hpp, lines 335-339 > > <https://reviews.apache.org/r/26736/diff/9/?file=753212#file753212line335> > > > > Does it make sense to reuse `Flags::executor_shutdown_grace_period`? > > Timothy Chen wrote: > I'm not quite sure we want to tie the executor shutdown to docker stop > time, as the concepts doesn't seem to be identical. I'm a bit fuzzy with the > executor shutdown altogether, that what are we including into "executor" that > is specified in the flag. The executor I believe usually is just for > CommandExecutor, so if we start including even into docker containers we > might want to rename the flag?
I think what we want to have at the end is some sort of task-dependent shutdown timeout. Now we have `executor_shutdown_grace_period` flag, that is available in all containerizers and will be propagated to all executors via CommandInfo. So the docker finalization is not identical to executor shutdown, they seem to me similar enough to share the same flag and mechanism. How about we put a TODO here for now and revisit it when the finalization concept is fleshed out? Maybe it is even worth to file a JIRA, what do you think? - Alexander ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/26736/#review60778 ----------------------------------------------------------- On Nov. 13, 2014, 8:33 p.m., Ryan Thomas wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/26736/ > ----------------------------------------------------------- > > (Updated Nov. 13, 2014, 8:33 p.m.) > > > Review request for mesos, Benjamin Hindman and Timothy Chen. > > > Bugs: MESOS-1925 > https://issues.apache.org/jira/browse/MESOS-1925 > > > Repository: mesos-git > > > Description > ------- > > Ensure docker calls stop if a docker_stop_timeout is provided Added the flag > docker_stop_timeout that defaults to 0, if the timeout is 0 the a docker kill > will be run, otherwise a docker stop with the timeout in seconds is used. A > negative value will result in a Failure. > > > Diffs > ----- > > src/docker/docker.hpp 2dc692c51d6f9d45d8d7c59f8678991fd80c6e89 > src/docker/docker.cpp 0c0a1bfd3c1260ebdf342d0a6e0e04d7b6a111a4 > src/slave/containerizer/docker.cpp 5978ec2fcf91c127ed7a8f12ee3a47e57f4c0f1b > src/slave/flags.hpp 4ec5954b2e532c9fb3ea83d4d7c06ca4d5a5ca74 > src/tests/docker_containerizer_tests.cpp > 66552ad5cb3b6ff978abd7bf388453f874c860f3 > src/tests/docker_tests.cpp ff06a01d7f20656c05bcf5afd3bc10059dc5378d > > Diff: https://reviews.apache.org/r/26736/diff/ > > > Testing > ------- > > make check > sudo bin/mesos-tests.sh --gtest_filter="DOCKER" --verbose > support/mesos-style.py > > > Thanks, > > Ryan Thomas > >
