I have a table similar to:
domain | file | Number
------------------------------------
aaa.com | aaa.com_1 | 111
bbb.com | bbb.co
As documented under SELECT Syntax:
The
HAVINGclause is applied nearly last, just before items are sent to the client, with no optimization.
In other words, it is applied after the grouping operation has been performed (in contrast with WHERE, which is performed before any grouping operation). See WHERE vs HAVING.
Therefore, your current query first forms the resultset from the following:
SELECT COUNT(domain) AS `sum domains`, file
FROM `table`
GROUP BY Number
See it on sqlfiddle:
| SUM DOMAINS | FILE | --------------------------- | 2 | aaa.com_1 | | 2 | bbb.com_1 | | 1 | eee.com_1 |
As you can see, the values selected for the file column are merely one of the values from each group—as documented under MySQL Extensions to GROUP BY:
The server is free to choose any value from each group, so unless they are the same, the values chosen are indeterminate.
Your current query then proceeds to filter these results according to your HAVING clause:
HAVING COUNT(Number) > 1 AND file LIKE '%\_1'
With the values of file selected above, every single group matches on the second criterion; and the first two groups match on the first criterion. Therefore the results of the complete query are:
| SUM DOMAINS | FILE | --------------------------- | 2 | aaa.com_1 | | 2 | bbb.com_1 |
Following your comments above, you want to filter the records on file before grouping and then filter the resulting groups for those containing more than one match. Therefore use WHERE and HAVING respectively (and select Number instead of file to identify each group):
SELECT Number, COUNT(*) AS `sum domains`
FROM `table`
WHERE file LIKE '%\_1'
GROUP BY Number
HAVING `sum domains` > 1
See it on sqlfiddle:
| NUMBER | SUM DOMAINS | ------------------------ | 222 | 2 |