同样的代码逻辑,同样的表结构,由于数据量的不同导致查询没有走索引,而引起请求超时案例分析
作者:互联网
问题描述:
预发环境目前加载全部变更会超时,线上正常。需要看一下是什么问题避免预发上的问题带到线上
分析解决过程:
EXPLAIN
SELECT i.damand_manager_info_id as damand_manager_id,
title,
i.damand_change_type,
i.customer_cloud_name_id,
i.damand_change_type_l2,
i.gmt_create,
i.pre_begin_time,
i.change_status,
i.category,
i.damand_pri,
i.change_result,
i.pre_end_time,
i.executor,
i.creator,
i.pro_inst_id,
i.damand_id,
i.share_change,
i.creator,
i.change_resaon,
i.actual_begin_time,
i.actual_end_time,
i.cloud_product,
i.cloud_site_id as site_id,
s.aliase,
s.cloud_site_name as site_name,
i.template_id,
i.is_curing,
i.risk_level,
i.sky_eye_id,
(CASE WHEN idbf.damand_id IS NULL THEN 0 WHEN idbf.`is_valid`= 0 THEN 0 ELSE 1 END) AS hasTodo,
(CASE WHEN idbf.`is_valid`= 0 THEN 0 WHEN idbf.status IS NULL THEN 0 WHEN idbf.status= '已处理' THEN 0 WHEN idbf.status= '未处理' THEN 1 ELSE 0 END) AS todoStatus
from damand_manager_info i
LEFT JOIN damand_manager_info_son son ON i.damand_manager_info_id= son.damand_manager_id
LEFT JOIN cloud_site s ON i.cloud_site_id= s.id
LEFT JOIN itom_damand_backfills idbf on idbf.damand_id= i.`damand_manager_info_id`
order by i.gmt_create desc
limit 0,
10;查看这个sql语句的执行计划,发现预发和生产不一样
预发:
生产:
解决方法:是更改sql语句,让它强制走索引,格式如下:force index("索引名");
加上force index(idx_201903081253)和force index(idx_damand_id);如下:
explain SELECT
i.damand_manager_info_id AS damand_manager_id,
title,
i.damand_change_type,
i.customer_cloud_name_id,
i.damand_change_type_l2,
i.gmt_create,
i.pre_begin_time,
i.change_status,
i.category,
i.damand_pri,
i.change_result,
i.pre_end_time,
i.executor,
i.creator,
i.pro_inst_id,
i.damand_id,
i.share_change,
i.creator,
i.change_resaon,
i.actual_begin_time,
i.actual_end_time,
i.cloud_product,
i.cloud_site_id AS site_id,
s.aliase,
s.cloud_site_name AS site_name,
i.template_id,
i.is_curing,
i.risk_level,
i.sky_eye_id,(
CASE
WHEN idbf.damand_id IS NULL THEN
0
WHEN idbf.`is_valid` = 0 THEN
0 ELSE 1
END
) AS hasTodo,(
CASE
WHEN idbf.`is_valid` = 0 THEN
0
WHEN idbf.STATUS IS NULL THEN
0
WHEN idbf.STATUS = '已处理' THEN
0
WHEN idbf.STATUS = '未处理' THEN
1 ELSE 0
END
) AS todoStatus
FROM
damand_manager_info i force index(idx_201903081253)
LEFT JOIN damand_manager_info_son son ON i.damand_manager_info_id = son.damand_manager_id
LEFT JOIN cloud_site s ON i.cloud_site_id = s.id
LEFT JOIN itom_damand_backfills idbf force index(idx_damand_id) ON idbf.damand_id = i.`damand_manager_info_id`
ORDER BY
i.gmt_create DESC
LIMIT 0,10
参考网址见:https://www.cnblogs.com/lipengsheng-javaweb/p/16619073.html
标签:damand,同样,manager,cloud,数据量,超时,id,change,idbf 来源: https://www.cnblogs.com/lipengsheng-javaweb/p/16619067.html