[SPARK-10540] HadoopFsRelationTest's "test all data types" is flaky

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 1.5.1, 1.6.0
• Component/s: SQL
• Labels: flaky-test
• Assignee: Cheng Lian
• Reporter: Yin Huai
• Created: 10/Sep/15 18:10
• Updated: 18/Sep/15 19:20
• Resolved: 18/Sep/15 19:19

Description

https://amplab.cs.berkeley.edu/jenkins/job/Spark-1.5-SBT/297/AMPLAB_JENKINS_BUILD_PROFILE=hadoop1.0,label=spark-test/testReport/junit/org.apache.spark.sql.sources/ParquetHadoopFsRelationSuite/test_all_data_types/

Probably it is caused by we sort results based on the strings of rows (https://github.com/apache/spark/blob/master/sql/core/src/test/scala/org/apache/spark/sql/QueryTest.scala#L118-L131) and the toString function of the user class used by the UDT is not overridden.

Issue Links

• links to: [Github] Pull Request #8702 (yhuai)
• links to: [Github] Pull Request #8705 (yhuai)
• links to: [Github] Pull Request #8768 (liancheng)