[SPARK-1916] SparkFlumeEvent with body bigger than 1020 bytes are not read properly

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 0.9.0
• Fix Version/s: 0.9.2, 1.0.1
• Component/s: DStreams
• Assignee: David Lemieux
• Reporter: David Lemieux
• Created: 23/May/14 19:28
• Updated: 03/Jun/14 08:21
• Resolved: 28/May/14 22:52

Description

The readExternal implementation on SparkFlumeEvent will read only the first 1020 bytes of the actual body when streaming data from flume.

This means that any event sent to Spark via Flume will be processed properly if the body is small, but will fail if the body is bigger than 1020.
Considering that the default max size for a Flume Avro Event is 32K, the implementation should be updated to read more.

The following is related : http://apache-spark-user-list.1001560.n3.nabble.com/Spark-Streaming-using-Flume-body-size-limitation-tt6127.html

Issue Links

• is related to: SPARK-1998 SparkFlumeEvent with body bigger than 1020 bytes are not read properly