Details
-
Type:
Bug
-
Status: Closed (View Workflow)
-
Priority:
Critical
-
Resolution: Fixed
-
Affects Version/s: N/A
-
Fix Version/s: 10.7.1
-
Component/s: Optimizer
-
Labels:None
Description
One of points from our earlier slack discussion was that there can not be more buckets in a JSON histogram than histogram_size. The below example proves the contrary.
It is possible that the conclusion was wrong or too generic and there are allowed exceptions from the rule. These exception will need to be a part of histogram_size description in the KB. The description has to be adjust anyway, as now it says that histogram_size means the number of buckets for JSON histograms, and it's certainly not the case.
--source include/have_sequence.inc
|
|
create or replace table t (a int); |
insert into t values (1),(3),(5),(7); |
insert into t select 2 from seq_1_to_25; |
insert into t select 4 from seq_1_to_25; |
insert into t select 6 from seq_1_to_25; |
|
set histogram_size=4, histogram_type=JSON_HB; |
analyze table t persistent for all; |
|
select histogram from mysql.column_stats where table_name = 't'; |
|
drop table t; |
preview-10.7-MDEV-26519-json-histograms da8bb4b470 |
select histogram from mysql.column_stats where table_name = 't'; |
histogram
|
{
|
"histogram_hb_v2": [ |
{
|
"start": "1", |
"size": 0.240506329, |
"ndv": 2 |
},
|
{
|
"start": "2", |
"size": 0.240506329, |
"ndv": 3 |
},
|
{
|
"start": "4", |
"size": 0.189873418, |
"ndv": 2 |
},
|
{
|
"start": "6", |
"size": 0.316455696, |
"ndv": 1 |
},
|
{
|
"start": "7", |
"end": "7", |
"size": 0.012658228, |
"ndv": 1 |
}
|
]
|
}
|
Attachments
Issue Links
- is caused by
-
MDEV-21130 Histograms: use JSON as on-disk format
-
- Closed
-
-
MDEV-26519 JSON Histograms: improve histogram collection
-
- Closed
-