[ORC-285] Empty vector batches of floats or doubles get java.io.EOFException

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.3.4, 1.4.1
• Fix Version/s: 1.4.3, 1.5.0
• Assignee: Owen O'Malley
• Reporter: Owen O'Malley
• Created: 27/Dec/17 17:12
• Updated: 09/Mar/20 19:38
• Resolved: 18/Jan/18 17:44

Description

The FloatTreeReader and DoubleTreeReader both fail if given a set of empty values. This often happens when the float or double data is inside a list where all of the values are empty.

The stack trace looks like:

java.io.IOException: Error reading file: /Users/piyush.mukati/tmp/11-task_1511779500016_298243_r_000031-r-00031.orc
	at org.apache.orc.impl.RecordReaderImpl.nextBatch(RecordReaderImpl.java:1191)
	at org.apache.orc.tools.ScanData.main(ScanData.java:67)
	at org.apache.orc.tools.Driver.main(Driver.java:109)
Caused by: java.io.EOFException: Read past EOF for compressed stream Stream for column 36 kind DATA position: 0 length: 0 range: 0 offset: 0 limit: 0
	at org.apache.orc.impl.SerializationUtils.readFully(SerializationUtils.java:119)
	at org.apache.orc.impl.SerializationUtils.readLongLE(SerializationUtils.java:102)
	at org.apache.orc.impl.SerializationUtils.readDouble(SerializationUtils.java:98)
	at org.apache.orc.impl.TreeReaderFactory$DoubleTreeReader.nextVector(TreeReaderFactory.java:763)
	at org.apache.orc.impl.TreeReaderFactory$StructTreeReader.nextVector(TreeReaderFactory.java:1835)
	at org.apache.orc.impl.TreeReaderFactory$StructTreeReader.nextVector(TreeReaderFactory.java:1835)
	at org.apache.orc.impl.TreeReaderFactory$StructTreeReader.nextVector(TreeReaderFactory.java:1835)
	at org.apache.orc.impl.TreeReaderFactory$ListTreeReader.nextVector(TreeReaderFactory.java:2003)
	at org.apache.orc.impl.TreeReaderFactory$StructTreeReader.nextVector(TreeReaderFactory.java:1835)
	at org.apache.orc.impl.TreeReaderFactory$ListTreeReader.nextVector(TreeReaderFactory.java:2003)
	at org.apache.orc.impl.TreeReaderFactory$StructTreeReader.nextVector(TreeReaderFactory.java:1835)
	at org.apache.orc.impl.TreeReaderFactory$StructTreeReader.nextBatch(TreeReaderFactory.java:1817)
	at org.apache.orc.impl.RecordReaderImpl.nextBatch(RecordReaderImpl.java:1184)
	... 2 more

Issue Links

• is caused by: HIVE-9711 ORC Vectorization DoubleColumnVector.isRepeating=false if all entries are NaN
• is related to: HIVE-16480 ORC file with empty array<double> and array<float> fails to read
• is related to: SPARK-23340 Upgrade Apache ORC to 1.4.3
• links to: GitHub Pull Request #205
• links to: GitHub Pull Request #477