Question

I am running two queries to my database for pagination reasons. As such, each query is nearly identical. My COUNT(*) query is not returning the number of results that the non-count query is. I’m baffled as to why this is the case. The queries are below.

SELECT p.host_id, p.rating_support, p.rating_tech, MAX(p.rating_overall) AS rating_overall, p.publish_rating, h.name, prices.price, prices.term_duration
FROM plans p
INNER JOIN hosts AS h ON h.id = p.host_id
INNER JOIN (SELECT plan_id, price, term_duration FROM prices WHERE price > 0 AND price < 50 AND term_duration = 1) prices ON prices.plan_id = p.id
WHERE p.published = 1 AND h.published = 1
GROUP BY p.host_id
ORDER BY rating_overall desc LIMIT 0, 12

SELECT COUNT(*) AS count
FROM plans p
INNER JOIN hosts AS h ON h.id = p.host_id
INNER JOIN (SELECT plan_id, price, term_duration FROM prices WHERE price > 0 AND price < 50 AND term_duration = 1) prices ON prices.plan_id = p.id
WHERE p.published = 1 AND h.published = 1
GROUP BY p.host_id

I’m not an expert at MySQL. Besides the count not providing the correct number of results, the non-count query works perfectly.

Any light on this problem would be great.

Answer 1

With the help of Dems’ comment (hunt down and upvote him somewhere :), I created this query. Notice that I removed the subquery, because it seemed unnecessary:

SELECT
  COUNT( DISTINCT p.host_id )
FROM       plans p
INNER JOIN hosts h ON h.id = p.host_id
INNER JOIN prices  ON prices.plan_id = p.id
                  AND prices.price > 0
                  AND prices.price < 50
                  AND prices.term_duration = 1
WHERE p.published = 1
  AND h.published = 1

---

My original answer:

To get the number of total row, you have to wrap the GROUP BY query into an outer SELECT:

SELECT COUNT(*)
FROM (
  SELECT NULL -- we are just counting, so we need no actual data -> a bit faster
  FROM       plans p
  INNER JOIN hosts h ON h.id = p.host_id
  INNER JOIN prices  ON prices.plan_id = p.id
                    AND prices.price > 0
                    AND prices.price < 50
                    AND prices.term_duration = 1
  WHERE p.published = 1
    AND h.published = 1
  GROUP BY p.host_id
) AS all_rows_without_data

Or you could use SQL_CALC_FOUND_ROWS + FOUND_ROWS()

http://dev.mysql.com/doc/refman/5.0/en/information-functions.html#function_found-rows

A SELECT statement may include a LIMIT clause to restrict the number
of rows the server returns to the client. In some cases, it is
desirable to know how many rows the statement would have returned
without the LIMIT, but without running the statement again. To obtain
this row count, include a SQL_CALC_FOUND_ROWS option in the SELECT
statement, and then invoke FOUND_ROWS() afterward:

First, simply select the required rows, but add SQL_CALC_FOUND_ROWS:

SELECT SQL_CALC_FOUND_ROWS
  p.host_id, p.rating_support, p.rating_tech,
  MAX(p.rating_overall) AS rating_overall,
  p.publish_rating, h.name, prices.price, prices.term_duration
FROM       plans p
INNER JOIN hosts AS h ON h.id = p.host_id
INNER JOIN prices  ON prices.plan_id = p.id
                  AND prices.price > 0
                  AND prices.price < 50
                  AND prices.term_duration = 1
WHERE p.published = 1 AND h.published = 1
GROUP BY p.host_id
ORDER BY rating_overall desc
LIMIT 0, 12;

Second, get the number of rows that would have been returned if there weren’t a LIMIT statement in the first query:

SELECT FOUND_ROWS();

Update: SQL_CALC_FOUND_ROWS + FOUND_ROWS() doesn’t seem very reliable, always returs zero for unknown reason (not just me: FOUND_ROWS() keeps returning 0 ):

http://sqlfiddle.com/#!2/7304d/8