| set hive.stats.column.autogather=false; |
| set hive.stats.fetch.column.stats=true; |
| set hive.compute.query.using.stats=true; |
| |
| |
| drop table calendar; |
| |
| CREATE TABLE calendar (year int, month int) clustered by (month) into 2 buckets stored as orc; |
| |
| insert into calendar values (2010, 10), (2011, 11), (2012, 12); |
| |
| desc formatted calendar; |
| |
| analyze table calendar compute statistics for columns year; |
| |
| desc formatted calendar; |
| |
| explain select max(year) from calendar; |
| |
| select max(year) from calendar; |
| |
| explain select count(1) from calendar; |
| |
| select count(1) from calendar; |
| |
| ALTER TABLE calendar CHANGE year year1 INT; |
| |
| --after patch, should be old stats rather than -1 |
| |
| desc formatted calendar; |
| |
| --but basic/column stats can not be used by optimizer |
| |
| explain select max(month) from calendar; |
| |
| select max(month) from calendar; |
| |
| explain select count(1) from calendar; |
| |
| select count(1) from calendar; |
| |
| truncate table calendar; |
| |
| --after patch, should be 0 |
| |
| desc formatted calendar; |
| |
| --but column stats can not be used by optimizer |
| |
| explain select max(month) from calendar; |
| |
| select max(month) from calendar; |
| |
| --basic stats can be used by optimizer |
| |
| explain select count(1) from calendar; |
| |
| select count(1) from calendar; |
