> On Sept. 6, 2017, 9 p.m., Qian Zhang wrote: > > src/tests/containerizer/ports_isolator_tests.cpp > > Lines 1051-1052 (patched) > > <https://reviews.apache.org/r/62003/diff/1/?file=1808287#file1808287line1051> > > > > Can you please elaborate a bit about this? What do you mean for `the > > original nested container status gets swallowed`? > > James Peach wrote: > When we are not using nested containers, we get a > `REASON_CONTAINER_LIMITATION` status update which comes directly from the > isolator. However, when we use nested containers, we actually get a > `REASON_COMMAND_EXECUTOR_FAILED` reason; the task status that comes from the > isolator never goes anywhere.
> However, when we use nested containers, we actually get a > REASON_COMMAND_EXECUTOR_FAILED reason; the task status that comes from the > isolator never goes anywhere. Is it because there is only one nested container and when that nested container is destroyed by Mesos containerizer, the default executor will know it and terminate itself, so it has no chance to send REASON_CONTAINER_LIMITATION status for that nested container before it's self termination? If so, what if we launch a task group which has two tasks? when one of the task is destroyed due to the limitation, will we get a `REASON_CONTAINER_LIMITATION` status update for it? - Qian ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/62003/#review184660 ----------------------------------------------------------- On Aug. 31, 2017, 7:16 a.m., James Peach wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/62003/ > ----------------------------------------------------------- > > (Updated Aug. 31, 2017, 7:16 a.m.) > > > Review request for mesos, Qian Zhang and Jiang Yan Xu. > > > Bugs: MESOS-7675 > https://issues.apache.org/jira/browse/MESOS-7675 > > > Repository: mesos > > > Description > ------- > > Added `network/ports` isolator nested container tests using the v1 > TaskGroups API. This tests that rogue port usage by a nested task is > detected both with and without agent recovery, and that a well-behaved > task is preserved across agent recovery. > > > Diffs > ----- > > src/tests/containerizer/ports_isolator_tests.cpp PRE-CREATION > > > Diff: https://reviews.apache.org/r/62003/diff/1/ > > > Testing > ------- > > make check (Fedora 26) > > > Thanks, > > James Peach > >
