----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/39152/#review101992 -----------------------------------------------------------
src/tests/containerizer/mesos_containerizer_tests.cpp (line 307) <https://reviews.apache.org/r/39152/#comment159517> Modifying the environment doesn't seem like a great thing for a test to be doing, since it might impact other tests -- we should at least reset LIBPROCESS_IP to its previous value at the end of the test. That still isn't great (e.g., if we want to run multiple tests concurrently in the same OS process), but I'm not sure there's a better way. - Neil Conway On Oct. 9, 2015, 12:37 a.m., Greg Mann wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/39152/ > ----------------------------------------------------------- > > (Updated Oct. 9, 2015, 12:37 a.m.) > > > Review request for mesos and Kapil Arya. > > > Bugs: MESOS-3553 > https://issues.apache.org/jira/browse/MESOS-3553 > > > Repository: mesos > > > Description > ------- > > If DNS is not available on the agent node and a task is launched which > explicitly specifies the executor's environment, LIBPROCESS_IP will not be > passed through and the default hostname lookup after spawning the executor > process will throw an error. This patch alters the agent to always pass > LIBPROCESS_IP, even when the executor environment is specified. > > > Diffs > ----- > > src/slave/containerizer/containerizer.cpp > 25c87e9f948b7efe8b9a853c403bee69982d6c4c > src/tests/containerizer/mesos_containerizer_tests.cpp > 5bc7d408bda0c249e1b66747d8bd87e688362e6c > > Diff: https://reviews.apache.org/r/39152/diff/ > > > Testing > ------- > > `make check` > > > Thanks, > > Greg Mann > >
