很小的MySQL表会忽略索引吗？

Question

After turning on log_queries_not_using_indexes , I noticed that one query was rapidly filling up the slow query log:

SELECT abc.* FROM abc WHERE abc.id NOT IN ( SELECT DISTINCT abc_id FROM zyx WHERE id = 12345 );

abc is very small, just 3 rows of data. zyx is relatively large with over 100,000 rows of data.

abc.id has an index, but when I EXPLAIN the query, the index isn't listed under either key nor possible_keys . This explains why the query is showing in the slow log, but my question is, why isn't it using the index?

In short, I have two questions:

1. Do very small tables ignore indexes? I could see why, it doesn't save much time to use an index on 3 rows of data.
2. If so, how can I prevent this query from flooding my slow query log?

Thank you for your time! :)

Additional information, if needed:

I have run ANALYZE TABLE abc as I've read sometimes fixes the issue. I have also restarted MariaDB since adding the index.

More of the EXPLAIN : select_type=PRIMARY, table=abc, type=ALL, possible_keys=NULL, key=NULL, key_len=NULL, ref=NULL, rows=3, Extra=Using where

Answer 1

Do very small tables ignore indexes?

Yes. When the entire table can be read in a single disk access, there's no point in performing a separate disk access to read the index.

If so, how can I prevent this query from flooding my slow query log?

Turn off log_queries_not_using_indexes . This is one of the reasons why it isn't on by default.

Answer 2

NOT IN ( SELECT ... ) is very poorly optimized, especially in older versions.

Change to this:

SELECT  abc.*
    FROM  abc
    LEFT JOIN  zyx  ON zyx.abc_id = abc.id
    WHERE  zyx.abc_id IS NULL;
AND  zyx.id = 12345 ;

For zyx, have either INDEX(id, abc_id) or INDEX(abc_id, id)

If zyx.id is the PRIMARY KEY , your query does not make much sense -- why test for a single row (12345)?