Aman Sinha created DRILL-1834:
---------------------------------
Summary: Misleading error message when querying an empty Parquet
file
Key: DRILL-1834
URL: https://issues.apache.org/jira/browse/DRILL-1834
Project: Apache Drill
Issue Type: Bug
Components: Execution - Flow
Affects Versions: 0.7.0
Reporter: Aman Sinha
It is possible that a CTAS may fail and create an empty Parquet file. When we
run a query against this file, we get a misleading error message from the
planner that hides the original IOException, although the log file does have
the original exception:
{code:sql}
0: jdbc:drill:zk=local> select count(*) from dfs.`/tmp/empty.parquet`;
Query failed: Query failed: Unexpected exception during fragment
initialization: Internal error: Error while applying rule
DrillPushProjIntoScan, args
[rel#77:ProjectRel.NONE.ANY([]).[](child=rel#76:Subset#0.ENUMERABLE.ANY([]).[],$f0=0),
rel#68:EnumerableTableAccessRel.ENUMERABLE.ANY([]).[](table=[dfs,
/tmp/empty.parquet])]
{code}
The cause of the exception is in the logs:
Caused by: java.io.IOException: Could not read footer:
java.lang.RuntimeException: file:/tmp/empty.parquet is not a Parquet file (too
small)
at
parquet.hadoop.ParquetFileReader.readAllFootersInParallel(ParquetFileReader.java:195)
~[parquet-hadoop-1.5.1-drill-r4.jar:0.7.0-SNAPSHOT]
at
parquet.hadoop.ParquetFileReader.readAllFootersInParallel(ParquetFileReader.java:208)
~[parquet-hadoop-1.5.1-drill-r4.jar:0.7.0-SNAPSHOT]
at
parquet.hadoop.ParquetFileReader.readFooters(ParquetFileReader.java:224)
~[parquet-hadoop-1.5.1-drill-r4.jar:0.7.0-SNAPSHOT]
at
org.apache.drill.exec.store.parquet.ParquetGroupScan.readFooter(ParquetGroupScan.java:208)
~[drill-java-exec-0.7.0-SNAPSHOT-rebuffed.jar:0.7.0-SNAPSHOT]
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
