Hi all!
I tested the chunk skipping indexes feature in the test database and everything was successful: data for multiple id values is selected very quickly.
In a productive database, data is not selected as quickly when the option is enabled.
I noticed differences in the table "_timescaledb_catalog".chunk_column_stats'
: on the test database the valid
field contains the values false
, and on the production database for chunks the valid
field contains the values true
.
Implementation steps:
- The
timescaledb.enable_chunk_skipping = on
setting is enabled. - A new table has been created and the option
SELECT enable_chunk_skipping('hypertable', 'id') is enabled;
. - The table were populated and its chunks were compressed.
- Execution of a query specifying the id values from different chunks:
SELECT * FROM hypertable WHERE id IN (1, 22222, 333333333);
Everything is fine in the test database, but not on the prod.
3 posts - 2 participants