> On Feb. 24, 2015, 10:25 a.m., Adam B wrote: > > src/tests/fetcher_tests.cpp, line 435 > > <https://reviews.apache.org/r/31362/diff/1/?file=874124#file874124line435> > > > > What non-0 return status do you expect? > > Bernd Mathiske wrote: > Our Subprocess (see subprocess.hpp) implementation does not report the > exit code from the program it ran. Instead it returns what waitpid() reports, > which is a child process ID in this case. This is different every time. I > think this behavior of Subprocess is less useful than it could be, but that's > a separate issue.
Revisited. I got that wrong. Subprocess returns the status in this piece of code: waitpid(pid, &status, WNOHANG) This means I can actually get the child process status. Now using this statement: // mesos-fetcher always exits with EXIT(1) on failure. EXPECT_EQ(1, WIFEXITED(status1.get().get())); - Bernd ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/31362/#review73838 ----------------------------------------------------------- On Feb. 24, 2015, 9:28 a.m., Bernd Mathiske wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/31362/ > ----------------------------------------------------------- > > (Updated Feb. 24, 2015, 9:28 a.m.) > > > Review request for mesos, Adam B, Ben Mahler, and Niklas Nielsen. > > > Bugs: MESOS-2390 > https://issues.apache.org/jira/browse/MESOS-2390 > > > Repository: mesos > > > Description > ------- > > Now the containerizer/fetcher sets the HADOOP_HOME variable for mesos-fetcher > again if the slave flag hadoop_home is set. Added a test that checks that > HDFS fetching is not broken and also ensures that the flag gets translated to > the environment variable and then gets applied in mesos-fetcher. Created a > mock hadoop implementation script for this. This script has the exact same > side effects as a real haddop client in the scope of our testing. Using this, > Mesos testing has no extra external dependencies (on Hadoop). > > Slave flag frameworks_home does not need to be an evironment variable. It is > now part of FetcherInfo only, but it also gets tested now that it works. > > > Diffs > ----- > > include/mesos/fetcher/fetcher.proto > facb87b92bf3194516f636dcc348e136af537721 > src/launcher/fetcher.cpp fed0105946da579a38357a30e7ae56e646e05b89 > src/slave/containerizer/fetcher.cpp > d290f95251def3952c5ee34f600e1d71467f6293 > src/tests/fetcher_tests.cpp 2438620e2db94c3c56336fa5d8e69a18fe8f3bac > src/tests/mock_hadoop.sh PRE-CREATION > > Diff: https://reviews.apache.org/r/31362/diff/ > > > Testing > ------- > > make check > > > Thanks, > > Bernd Mathiske > >
