diff options
author | Marcelo Vanzin <vanzin@cloudera.com> | 2014-06-04 22:56:49 -0700 |
---|---|---|
committer | Patrick Wendell <pwendell@gmail.com> | 2014-06-04 22:56:49 -0700 |
commit | b77c19be053125fde99b098ec1e1162f25b5433c (patch) | |
tree | c6dcfc077d0add1c8fcf49e16ab492366e742b8a /data/lr_data.txt | |
parent | abea2d4ff099036c67fc73136d0e61d0d0e22123 (diff) | |
download | spark-b77c19be053125fde99b098ec1e1162f25b5433c.tar.gz spark-b77c19be053125fde99b098ec1e1162f25b5433c.tar.bz2 spark-b77c19be053125fde99b098ec1e1162f25b5433c.zip |
Fix issue in ReplSuite with hadoop-provided profile.
When building the assembly with the maven "hadoop-provided"
profile, the executors were failing to come up because Hadoop classes
were not found in the classpath anymore; so add them explicitly to
the classpath using spark.executor.extraClassPath. This is only
needed for the local-cluster mode, but doesn't affect other tests,
so it's added for all of them to keep the code simpler.
Author: Marcelo Vanzin <vanzin@cloudera.com>
Closes #781 from vanzin/repl-test-fix and squashes the following commits:
4f0a3b0 [Marcelo Vanzin] Fix issue in ReplSuite with hadoop-provided profile.
Diffstat (limited to 'data/lr_data.txt')
0 files changed, 0 insertions, 0 deletions