Re: performance issue

[Rick James]

select count(id_text_meta_data), text_topic_text_medium_id_text_medium
without a GROUP BY does not make sense. Rethink the query.

IN ( SELECT ... )
has terrible performance characteristics. Try to turn it into a JOIN. However, you will then run into troubles with join-group-inflation. That is where a JOIN creates more rows than are desired for the GROUP BY, hence leading to a COUNT() or SUM() with an inflated value.

Can you describe what the query is trying to do?
Can you describe it from the inside out?

(Sorry, I don't see a simple transformation to make the query faster; perhaps your reply will help.)

What kind of stuff is in flag1..5? 5 Boolean values? More than than? I am looking for shrinking the fields and/or moving them to a separate table.

> PRIMARY KEY (`id_file_control`,`text_meta_data_id_text_meta_data`),

* A PRIMARY KEY is a UNIQUE key
* id_file_control is declared AUTO_INCREMENT, which usually indicates you want unique values.
These are fighting each other. Can you explain your choice of PK?

The other is similarly puzzling.

(Your column names are painfully verbose.)

You have a redundant index on `text_meta_data_id_text_meta_data`.