[DRILL-2044] Filter not being pushed down when we join tables with wide records

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Fix Version/s: Future
• Component/s: Query Planning & Optimization
• Assignee: Unassigned
• Reporter: Rahul Kumar Challapalli
• Created: 20/Jan/15 19:23
• Updated: 01/Oct/15 16:38

Description

git.commit.id.abbrev=a418af1

The filter is not being pushed down according to the plan. This could either be a bug or expected behavior based on the optimization rules. So someone needs to verify that it is atleast not a bug

explain plan for select count(ws1.str_var) from widestrings_small ws1 INNER JOIN widestrings_small ws2 on ws1.str_fixed_null_empty=ws2.str_var_null_empty where ws1.tinyint_var > 120;

00-00    Screen
00-01      StreamAgg(group=[{}], EXPR$0=[COUNT($0)])
00-02        Project(str_var=[$2])
00-03          SelectionVectorRemover
00-04            Filter(condition=[>($1, 120)])
00-05              HashJoin(condition=[=($0, $3)], joinType=[inner])
00-07                Project(str_fixed_null_empty=[$2], tinyint_var=[$1], str_var=[$0])
00-08                  Scan(groupscan=[ParquetGroupScan [entries=[ReadEntryWithPath [path=maprfs:/drill/testdata/data-shapes/wide-columns/5000/1000rows/parquet/widestrings_small]], selectionRoot=/drill/testdata/data-shapes/wide-columns/5000/1000rows/parquet/widestrings_small, numFiles=1, columns=[`str_fixed_null_empty`, `tinyint_var`, `str_var`]]])
00-06                Scan(groupscan=[ParquetGroupScan [entries=[ReadEntryWithPath [path=maprfs:/drill/testdata/data-shapes/wide-columns/5000/1000rows/parquet/widestrings_small]], selectionRoot=/drill/testdata/data-shapes/wide-columns/5000/1000rows/parquet/widestrings_small, numFiles=1, columns=[`str_var_null_empty`]]])

I attached the data file used. Let me know if you have any questions