有问题的查询：

SELECT np.name AS name_trips, 
        s.station_name AS station_name, 
        np.id AS id, 
        COUNT(*) AS num_trips, 
        COUNT(NULLIF(np.start_end, 'Start')) AS num_start, 
        COUNT(NULLIF(np.start_end, 'End')) AS num_end
    FROM name_id_pairs np
    JOIN stations s ON np.id=s.station_id
    WHERE id IN(2498,3794)
    GROUP BY name, id
    ORDER BY id, num_trips DESC;

我的另一个每次都有效的查询是

SELECT np.name AS name_trips, 
        s.station_name AS station_name, 
        np.id AS id, 
        COUNT(*) AS num_trips, 
        COUNT(NULLIF(np.start_end, 'Start')) AS num_start, 
        COUNT(NULLIF(np.start_end, 'End')) AS num_end
    FROM name_id_pairs np
    JOIN stations s ON np.id=s.station_id
    WHERE id IN(
        SELECT id
        FROM name_id_pairs 
        GROUP BY id
        HAVING COUNT(DISTINCT name)>1)
    GROUP BY name, id
    ORDER BY id, num_trips DESC;

这两个查询之间唯一的区别是WHERE子句，那么为什么带有子查询的查询总是在几秒钟内完成，而WHERE id IN(2498,3794)将继续执行，直到我重新启动服务器。
两个查询都从同一视图中进行选择

CREATE VIEW name_id_pairs AS
    SELECT checkout_station_name AS name, checkout_station_id AS id, "Start" AS start_end
    FROM trips
    UNION ALL
    SELECT return_station_name AS name, return_station_id AS id, "End" AS start_end
    FROM trips;

我意识到视图可能会对MySQL的性能造成影响，但如果两个查询都使用同一个视图，我不认为它只会影响其中一个的性能。
所有的查询都是通过Python和MySQL连接器执行的。
由于查询之间的唯一区别是where子句，所以我尝试了几个不同版本的查询，但都没有正确执行。我尝试用WHERE id = 2498 OR id = 3794替换列表，但没有任何改变。我还尝试用等效的s.station_id或www.example.com替换idnp.id，但查询仍然没有执行。