[
https://issues.apache.org/jira/browse/DRILL-7934?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17354073#comment-17354073
]
ASF GitHub Bot commented on DRILL-7934:
---------------------------------------
vvysotskyi commented on a change in pull request #2238:
URL: https://github.com/apache/drill/pull/2238#discussion_r642094351
##########
File path:
exec/java-exec/src/main/java/org/apache/drill/exec/store/parquet/ParquetGroupScanStatistics.java
##########
@@ -115,7 +118,7 @@ public void collect(Collection<T> metadataList) {
previousCount.setValue(Statistic.NO_COLUMN_STATS);
}
ColumnMetadata columnMetadata =
SchemaPathUtils.getColumnMetadata(schemaPath, metadata.getSchema());
- TypeProtos.MajorType majorType = columnMetadata != null ?
columnMetadata.majorType() : null;
+ TypeProtos.MajorType majorType = columnMetadata != null ?
columnMetadata.majorType() : NULL;
Review comment:
It is used to collect metadata for partition columns only. Even if some
specific field is not determined as a partition, it still may have its metadata
and be used in row group pruning later. But if you will set the type to NULL,
it may cause issues for this case when it is used later for some specific types
like `FIXED_LEN_BYTE_ARRAY` or `BINARY`. So it is better do not mark such field
as partition at all and avoid possible issues with determining comparator (see
`ParquetGroupScanStatistics.getTypeForColumn()` method usage).
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
> NullPointerException error when reading parquet files
> -----------------------------------------------------
>
> Key: DRILL-7934
> URL: https://issues.apache.org/jira/browse/DRILL-7934
> Project: Apache Drill
> Issue Type: Bug
> Components: Storage - Parquet
> Affects Versions: 1.18.0
> Environment: Drill 1.18
> Ambari 2.7.4
> Spark 3.0.2
> Reporter: cdmikechen
> Priority: Critical
> Fix For: 1.19.0
>
> Attachments:
> part-00000-e849bed7-5cc2-480c-96d8-3fe5f9b4294a-c000.snappy.parquet,
> part-00001-e849bed7-5cc2-480c-96d8-3fe5f9b4294a-c000.snappy.parquet
>
>
> I create a dataset using spark ml, when I use drill 1.18 to query this
> dataset folder, it report error this:
> {code:java}
> [Error Id: 92d3f331-ffca-46b5-a64c-87453b88a108 on xxx.xxx.xxx:31010]
> at
> org.apache.drill.common.exceptions.UserException$Builder.build(UserException.java:657)
> at
> org.apache.drill.exec.work.foreman.Foreman$ForemanResult.close(Foreman.java:788)
> at
> org.apache.drill.exec.work.foreman.QueryStateProcessor.checkCommonStates(QueryStateProcessor.java:322)
> at
> org.apache.drill.exec.work.foreman.QueryStateProcessor.planning(QueryStateProcessor.java:216)
> at
> org.apache.drill.exec.work.foreman.QueryStateProcessor.moveToState(QueryStateProcessor.java:76)
> at org.apache.drill.exec.work.foreman.Foreman.run(Foreman.java:300)
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
> at java.lang.Thread.run(Thread.java:748)
> Caused by: org.apache.drill.exec.work.foreman.ForemanException: Unexpected
> exception during fragment initialization: Error while applying rule
> DrillPushProjectIntoScanRule:enumerable, args
> [rel#478:LogicalProject.NONE.ANY([]).[](input=RelSubset#477,label=$1),
> rel#452:EnumerableTableScan.ENUMERABLE.ANY([]).[](table=[hdfs_dataset.default,
> /home/spark/dataset/default/test2/*.parquet])]
> at org.apache.drill.exec.work.foreman.Foreman.run(Foreman.java:301)
> ... 3 common frames omitted
> Caused by: java.lang.RuntimeException: Error while applying rule
> DrillPushProjectIntoScanRule:enumerable, args
> [rel#478:LogicalProject.NONE.ANY([]).[](input=RelSubset#477,label=$1),
> rel#452:EnumerableTableScan.ENUMERABLE.ANY([]).[](table=[hdfs_dataset.default,
> /home/spark/dataset/default/test2/*.parquet])]
> at
> org.apache.calcite.plan.volcano.VolcanoRuleCall.onMatch(VolcanoRuleCall.java:235)
> at
> org.apache.calcite.plan.volcano.VolcanoPlanner.findBestExp(VolcanoPlanner.java:633)
> at
> org.apache.calcite.tools.Programs$RuleSetProgram.run(Programs.java:327)
> at
> org.apache.drill.exec.planner.sql.handlers.DefaultSqlHandler.transform(DefaultSqlHandler.java:405)
> at
> org.apache.drill.exec.planner.sql.handlers.DefaultSqlHandler.transform(DefaultSqlHandler.java:351)
> at
> org.apache.drill.exec.planner.sql.handlers.DefaultSqlHandler.convertToRawDrel(DefaultSqlHandler.java:245)
> at
> org.apache.drill.exec.planner.sql.handlers.DefaultSqlHandler.convertToDrel(DefaultSqlHandler.java:308)
> at
> org.apache.drill.exec.planner.sql.handlers.DefaultSqlHandler.getPlan(DefaultSqlHandler.java:173)
> at
> org.apache.drill.exec.planner.sql.DrillSqlWorker.getQueryPlan(DrillSqlWorker.java:283)
> at
> org.apache.drill.exec.planner.sql.DrillSqlWorker.getPhysicalPlan(DrillSqlWorker.java:163)
> at
> org.apache.drill.exec.planner.sql.DrillSqlWorker.convertPlan(DrillSqlWorker.java:140)
> at
> org.apache.drill.exec.planner.sql.DrillSqlWorker.getPlan(DrillSqlWorker.java:93)
> at org.apache.drill.exec.work.foreman.Foreman.runSQL(Foreman.java:593)
> at org.apache.drill.exec.work.foreman.Foreman.run(Foreman.java:274)
> ... 3 common frames omitted
> Caused by: java.lang.NullPointerException: null
> at
> org.apache.drill.exec.store.parquet.ParquetGroupScanStatistics.checkForPartitionColumn(ParquetGroupScanStatistics.java:186)
> at
> org.apache.drill.exec.store.parquet.ParquetGroupScanStatistics.collect(ParquetGroupScanStatistics.java:119)
> at
> org.apache.drill.exec.store.parquet.ParquetGroupScanStatistics.<init>(ParquetGroupScanStatistics.java:59)
> at
> org.apache.drill.exec.store.parquet.BaseParquetMetadataProvider.getParquetGroupScanStatistics(BaseParquetMetadataProvider.java:293)
> at
> org.apache.drill.exec.store.parquet.BaseParquetMetadataProvider.getTableMetadata(BaseParquetMetadataProvider.java:249)
> at
> org.apache.drill.exec.store.parquet.BaseParquetMetadataProvider.initializeMetadata(BaseParquetMetadataProvider.java:203)
> at
> org.apache.drill.exec.store.parquet.BaseParquetMetadataProvider.init(BaseParquetMetadataProvider.java:170)
> at
> org.apache.drill.exec.metastore.store.parquet.ParquetTableMetadataProviderImpl.<init>(ParquetTableMetadataProviderImpl.java:95)
> at
> org.apache.drill.exec.metastore.store.parquet.ParquetTableMetadataProviderImpl.<init>(ParquetTableMetadataProviderImpl.java:48)
> at
> org.apache.drill.exec.metastore.store.parquet.ParquetTableMetadataProviderImpl$Builder.build(ParquetTableMetadataProviderImpl.java:415)
> at
> org.apache.drill.exec.store.parquet.ParquetGroupScan.<init>(ParquetGroupScan.java:150)
> at
> org.apache.drill.exec.store.parquet.ParquetGroupScan.<init>(ParquetGroupScan.java:120)
> at
> org.apache.drill.exec.store.parquet.ParquetFormatPlugin.getGroupScan(ParquetFormatPlugin.java:202)
> at
> org.apache.drill.exec.store.parquet.ParquetFormatPlugin.getGroupScan(ParquetFormatPlugin.java:79)
> at
> org.apache.drill.exec.store.dfs.FileSystemPlugin.getPhysicalScan(FileSystemPlugin.java:226)
> at
> org.apache.drill.exec.store.dfs.FileSystemPlugin.getPhysicalScan(FileSystemPlugin.java:209)
> at
> org.apache.drill.exec.planner.logical.DrillTable.getGroupScan(DrillTable.java:119)
> at
> org.apache.drill.exec.planner.logical.DrillPushProjectIntoScanRule.canPushProjectIntoScan(DrillPushProjectIntoScanRule.java:190)
> at
> org.apache.drill.exec.planner.logical.DrillPushProjectIntoScanRule.onMatch(DrillPushProjectIntoScanRule.java:107)
> at
> org.apache.calcite.plan.volcano.VolcanoRuleCall.onMatch(VolcanoRuleCall.java:208)
> ... 16 common frames omitted
> {code}
> It is same like issue https://issues.apache.org/jira/browse/DRILL-7769.
> I add some log information and found this:
> {code:java}
> TRACE o.a.d.e.s.p.ParquetGroupScanStatistics - check schema path
> `features`.`values`.`list`.`element` with major type null
> current partitionColTypeMap = {`features`.`indices`.`list`.`element`=null,
> `features`.`type`=minor_type: TINYINT
> mode: REQUIRED
> , `features`.`size`=minor_type: INT
> mode: OPTIONAL
> }
> 2021-05-25 15:39:21,066 [1f535658-f840-9f0e-1a7b-21080514bb7b:foreman] TRACE
> o.a.d.e.s.p.ParquetGroupScanStatistics - check schema path `label` with major
> type minor_type: FLOAT8
> mode: REQUIRED
> current partitionColTypeMap = {`features`.`indices`.`list`.`element`=null,
> `features`.`type`=minor_type: TINYINT
> mode: REQUIRED
> , `features`.`size`=minor_type: INT
> mode: OPTIONAL
> }
> 2021-05-25 15:39:21,066 [1f535658-f840-9f0e-1a7b-21080514bb7b:foreman] TRACE
> o.a.d.e.s.p.ParquetGroupScanStatistics - check schema path `features`.`size`
> with major type minor_type: INT
> mode: OPTIONAL
> current partitionColTypeMap = {`features`.`indices`.`list`.`element`=null,
> `features`.`type`=minor_type: TINYINT
> mode: REQUIRED
> , `features`.`size`=minor_type: INT
> mode: OPTIONAL
> }
> {code}
> So that there is some condition major type is null, if drill use this code,
> it will catch NullPointerException error:
> {code:java}
> TypeProtos.MajorType majorType = columnMetadata != null ?
> columnMetadata.majorType() : null; # 121
> !partitionColTypeMap.get(schemaPath).equals(type) # 189
> {code}
> we need to change null to *org.apache.drill.common.types.Types.NULL* to avoid
> NullPointerException error
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
