Rahul Challapalli created DRILL-4999:
----------------------------------------
Summary: Metadata Cache Pruning not taking place when we have both
directory partitioning and auto partitioned parquet files
Key: DRILL-4999
URL: https://issues.apache.org/jira/browse/DRILL-4999
Project: Apache Drill
Issue Type: Bug
Components: Metadata, Query Planning & Optimization
Reporter: Rahul Challapalli
git.commit.id.abbrev=17b9648
Metadata cache pruning is not taking place in the below scenario
{code}
0: jdbc:drill:zk=10.10.100.190:5181> explain plan for select distinct
i_rec_end_date from
dfs.`/drill/testdata/parquet_date/auto_partition/item_single` where dir0='1.9'
and i_rec_start_date = date '1997-10-27';
+------+------+
| text | json |
+------+------+
| 00-00 Screen
00-01 Project(i_rec_end_date=[$0])
00-02 HashAgg(group=[{0}])
00-03 Project(i_rec_end_date=[$2])
00-04 SelectionVectorRemover
00-05 Filter(condition=[AND(=($0, '1.9'), =($1, 1997-10-27))])
00-06 Scan(groupscan=[ParquetGroupScan
[entries=[ReadEntryWithPath
[path=/drill/testdata/parquet_date/auto_partition/item_single/1.9/0_0_1.parquet]],
selectionRoot=/drill/testdata/parquet_date/auto_partition/item_single,
numFiles=1, usedMetadataFile=true,
cacheFileRoot=/drill/testdata/parquet_date/auto_partition/item_single,
columns=[`dir0`, `i_rec_start_date`, `i_rec_end_date`]]])
{code}
I attached the data set and the log file
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
