> On Sept. 1, 2016, 5:48 a.m., Qian Zhang wrote: > > src/slave/slave.cpp, line 1882 > > <https://reviews.apache.org/r/51477/diff/2/?file=1487469#file1487469line1882> > > > > I do not think we need this, because if we get here, `kill` must be > > still `false`.
Good catch! I added an explicit invariant check here. > On Sept. 1, 2016, 5:48 a.m., Qian Zhang wrote: > > src/slave/slave.cpp, line 1951 > > <https://reviews.apache.org/r/51477/diff/2/?file=1487469#file1487469line1951> > > > > Not yours, but I think we do not need this blank line. We might want to clean this up in a separate patch. > On Sept. 1, 2016, 5:48 a.m., Qian Zhang wrote: > > src/slave/slave.cpp, line 1754 > > <https://reviews.apache.org/r/51477/diff/2/?file=1487469#file1487469line1754> > > > > I think we can `break` from here. hmm, we can't `break` here. Let's say we had a task group containing tasks: T1, T2 & T3. We put them in the pending queue in `_runTask()`. A scheduler sends a `killTask()` for T2 before `__runTask()` is invoked. We remove T2 from pending in `killTask()`. When we are in `__runTask`, pending is T1 & T3. If we break in the loop just after `T2`, we would leave behind `T3` still in `framework->pending`. > On Sept. 1, 2016, 5:48 a.m., Qian Zhang wrote: > > src/slave/slave.cpp, line 2542 > > <https://reviews.apache.org/r/51477/diff/2/?file=1487469#file1487469line2542> > > > > Here we send a `KillTaskMessage` to executor, but the task to be killed > > may be part of a task group, so that means it is executor's responsibility > > to kill the tasks in the whole group, right? What if executor does not do > > it? Will we enforce it in agent? We have never enforced this on the agent (historically for the single task case too) - Anand ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/51477/#review147401 ----------------------------------------------------------- On Sept. 6, 2016, 9:25 p.m., Anand Mazumdar wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/51477/ > ----------------------------------------------------------- > > (Updated Sept. 6, 2016, 9:25 p.m.) > > > Review request for mesos and Vinod Kone. > > > Bugs: MESOS-6076 > https://issues.apache.org/jira/browse/MESOS-6076 > > > Repository: mesos > > > Description > ------- > > This changes implements the `runTaskGroup()` handler on the > agent ensuring that task group is sent atomically to the executor > via the `LAUNCH_GROUP` event. It also refactors `runTask()`/`_runTask()` > to go through a common handler function. Also, it ensures that all > tasks in `framework->pending`/`queuedTasks` that are killed before > running the task group result in all the tasks being killed. > > Review: https://reviews.apache.org/r/51477/ > > > Diffs > ----- > > src/slave/slave.hpp 4add4c0180ea56039e0d5009bad4d9346128bde6 > src/slave/slave.cpp 11664779ed78c0a5913598bb7dd1bb0e793d6b93 > > Diff: https://reviews.apache.org/r/51477/diff/ > > > Testing > ------- > > make check > > > Thanks, > > Anand Mazumdar > >
