What is the exact behaviour of an INSERT IGNORE… SELECT… ORDER BY… LIMIT… in MySql?

Question

In MySql 5.6 I have a query similar to the following:

INSERT IGNORE INTO TABLE_1 (field_a,field_b)
SELECT field_a,field_b
FROM TABLE_2
WHERE
...
ORDER BY field_a
LIMIT 0,10

TABLE_1 is a temporary table that is used to store some tuples and then emptied. It doesn't have a PRIMARY KEY.

Since the process fills the table in multiple times, it may happen that it may contain tuples found throught the SELECT . Considering this, I thought about increasing the limit, so that I would be able to insert all the needed data.

Here is an example of the content of TABLE_1 and of the result of the SELECT query.

TABLE_1 :

+---------+---------+
| field_a | field_b |
+---------+---------+
| foo     | 1       |
| foo     | 2       |
| foo     | 3       |
| foo     | 4       |
| bar     | 2       |
| bar     | 3       |
| bar     | 4       |
| bar     | 5       |
+---------+---------+

SELECT result (ignoring LIMIT ):

+---------+---------+
| field_a | field_b |
+---------+---------+
| foo     | 4       |
| foo     | 5       |
| foo     | 6       |
| foo     | 7       |
| foo     | 8       |
| foo     | 9       |
| foo     | 10      |
| foo     | 11      |
| foo     | 12      |
| foo     | 13      |
| bar     | 5       |
| bar     | 6       |
| bar     | 7       |
| bar     | 8       |
+---------+---------+

So, considering that there are two duplicate tuples, (foo,4) and (bar,5) , I expected 8 elements to be inserted... while, I actually found that the query inserted 10 elements, ignoring the two duplicates.

The point is that I can't find the reason of this behavior in the docs. I found here that LIMIT works in INSERT...SELECT only if there is an ORDER BY , which I have. But I was not able to find a precise description of this behavior, having together INSERT IGNORE , INSERT...SELECT , ORDER BY and LIMIT .

Can somebody help me to explain this unexpected behavior?

Answer 1

Manual for INSERT IGNORE :

If you use the IGNORE keyword, errors that occur while executing the INSERT statement are ignored.

It is likely that there is no unique constraint on ( field_a, field_b) on TABLE_1 . No "duplicate" error can occur without such a constraint. TABLE_2 's primary key (which is a special case of unique constraint) has no effect in your query.

Also, LIMIT n without ORDER BY "works", it just returns the first n random rows that the query happens to generate.