Details
-
Bug
-
Status: Closed (View Workflow)
-
Critical
-
Resolution: Incomplete
-
10.4.19
-
CentOS
Description
After making optimizer_use_condition_selectivity=1 from 4 and disabling the below optimizer_switch options that weren't enabled in 10.3.21 we still get a different 'key' selected. As a result query that took ~0.5 sec in 10.3 takes ~1.5 min in 10.4
- optimize_join_buffer_size
- condition_pushdown_for_subquery
- rowid_filter
- condition_pushdown_from_having
FORCE INDEX (key used in 10.3) however worked in 10.4 and execution time was same like 10.3
EXPLAIN IN 10.4
id: 1 |
select_type: SIMPLE
|
table: table_name
|
type: range
|
possible_keys: index_table_col1, index_table_col2, ... ,index_table_col6
|
key: index_table_col6
|
key_len: 4 |
ref: NULL
|
rows: 80196 |
Extra: Using index condition; Using where
|
1 row in set (0.003 sec) |
Same Query in 10.3
id: 1 |
select_type: SIMPLE
|
table: table_name
|
type: range
|
possible_keys: index_table_col1, index_table_col2, ... ,index_table_col6
|
key: index_table_col5
|
key_len: 7 |
ref: NULL
|
rows: 65841 |
Extra: Using index condition; Using where
|
1 row in set (0.017 sec) |