[IMPALA-9529] Predicates on nested types may not be applied correctly on table masking view

• Type: Bug
• Status: Resolved
• Priority: Critical
• Resolution: Fixed
• Fix Version/s: Impala 3.4.0
• Labels: correctness
• Epic Link: Ranger Column Masking
• Epic Color: ghx-label-7
• Assignee: Quanlong Huang
• Reporter: Quanlong Huang
• Created: 17/Mar/20 13:16
• Updated: 14/Apr/20 22:32
• Resolved: 09/Apr/20 01:54

Description

When column masking is enabled, predicates that reference both primitive and nested types will be resolved to "multi-tuple predicates" (not bound to any single tuple id). This makes the predicates not being picked up correctly.

Reproduce
Create a column masking policy on functional_parquet.complextypestbl table: id => 100 * id. The following query has incorrect results:

select id, nested_struct.a from functional_parquet.complextypestbl t
where id = 100 or nested_struct.a = 1;
+-----+-----------------+
| id  | nested_struct.a |
+-----+-----------------+
| 100 | 1               |
| 200 | NULL            |
| 300 | NULL            |
| 400 | NULL            |
| 500 | NULL            |
| 600 | NULL            |
| 700 | 7               |
| 800 | -1              |
+-----+-----------------+

Explaining the query shows the predicate are not assigned:

Query: explain select id, nested_struct.a from functional_parquet.complextypestbl t
where id = 100 or nested_struct.a = 1
+---------------------------------------------------------------------------------------+
| Explain String                                                                        |
+---------------------------------------------------------------------------------------+
| Max Per-Host Resource Reservation: Memory=16.00KB Threads=3                           |
| Per-Host Resource Estimates: Memory=32MB                                              |
| WARNING: The following tables are missing relevant table and/or column statistics.    |
| functional_parquet.complextypestbl                                                    |
| Analyzed query: SELECT id, nested_struct.a FROM (SELECT CAST(CAST(100 AS BIGINT)      |
| * id AS BIGINT) id FROM functional_parquet.complextypestbl t) WHERE id =              |
| CAST(100 AS BIGINT) OR nested_struct.a = CAST(1 AS INT)                               |
|                                                                                       |
| F01:PLAN FRAGMENT [UNPARTITIONED] hosts=1 instances=1                                 |
| Per-Host Resources: mem-estimate=57.78KB mem-reservation=0B thread-reservation=1      |
|   PLAN-ROOT SINK                                                                      |
|   |  output exprs: CAST(CAST(100 AS BIGINT) * id AS BIGINT), nested_struct.a          |
|   |  mem-estimate=0B mem-reservation=0B thread-reservation=0                          |
|   |                                                                                   |
|   01:EXCHANGE [UNPARTITIONED]                                                         |
|      mem-estimate=57.78KB mem-reservation=0B thread-reservation=0                     |
|      tuple-ids=0 row-size=12B cardinality=4.40K                                       |
|      in pipelines: 00(GETNEXT)                                                        |
|                                                                                       |
| F00:PLAN FRAGMENT [RANDOM] hosts=2 instances=2                                        |
| Per-Host Resources: mem-estimate=32.00MB mem-reservation=16.00KB thread-reservation=2 |
|   DATASTREAM SINK [FRAGMENT=F01, EXCHANGE=01, UNPARTITIONED]                          |
|   |  mem-estimate=0B mem-reservation=0B thread-reservation=0                          |
|   00:SCAN HDFS [functional_parquet.complextypestbl t, RANDOM]                         |
|      HDFS partitions=1/1 files=2 size=6.92KB                                          |
|      stored statistics:                                                               |
|        table: rows=unavailable size=unavailable                                       |
|        columns missing stats: id                                                      |
|      extrapolated-rows=disabled max-scan-range-rows=unavailable                       |
|      mem-estimate=32.00MB mem-reservation=16.00KB thread-reservation=1                |
|      tuple-ids=0 row-size=12B cardinality=4.40K                                       |
|      in pipelines: 00(GETNEXT)                                                        |
+---------------------------------------------------------------------------------------+

Issue Links

• links to: Code Review