diff options
author | Hari Shreedharan <hshreedharan@apache.org> | 2015-05-13 16:43:30 -0700 |
---|---|---|
committer | Andrew Or <andrew@databricks.com> | 2015-05-13 16:43:30 -0700 |
commit | 61d1e87c0d3d12dac0b724d1b84436f748227e99 (patch) | |
tree | 0511f757c1aacf1b9b3a7926ae2c041fd227733a /core/src/main/scala/org/apache/spark/shuffle | |
parent | bb6dec3b160b54488892a509965fee70a530deff (diff) | |
download | spark-61d1e87c0d3d12dac0b724d1b84436f748227e99.tar.gz spark-61d1e87c0d3d12dac0b724d1b84436f748227e99.tar.bz2 spark-61d1e87c0d3d12dac0b724d1b84436f748227e99.zip |
[SPARK-7356] [STREAMING] Fix flakey tests in FlumePollingStreamSuite using SparkSink's batch CountDownLatch.
This is meant to make the FlumePollingStreamSuite deterministic. Now we basically count the number of batches that have been completed - and then verify the results rather than sleeping for random periods of time.
Author: Hari Shreedharan <hshreedharan@apache.org>
Closes #5918 from harishreedharan/flume-test-fix and squashes the following commits:
93f24f3 [Hari Shreedharan] Add an eventually block to ensure that all received data is processed. Refactor the dstream creation and remove redundant code.
1108804 [Hari Shreedharan] [SPARK-7356][STREAMING] Fix flakey tests in FlumePollingStreamSuite using SparkSink's batch CountDownLatch.
Diffstat (limited to 'core/src/main/scala/org/apache/spark/shuffle')
0 files changed, 0 insertions, 0 deletions