使用慢日志查询的 PHP Mysql 优化
Posted
技术标签:
【中文标题】使用慢日志查询的 PHP Mysql 优化【英文标题】：PHp Mysql OPtimization using slow-log-query 【发布时间】：2011-02-11 07:48:35 【问题描述】：您好，我仍在研究 mysql 性能，目前我开始检查 mysql 服务器上的 -log-slow-query 命令。
它会返回下一个结果：
/usr/sbin/mysqld，版本：5.0.81-community-log（MySQL 社区版（GPL））。开始于：
Tcp port: 0 Unix socket: (null)
Time Id Command Argument
110211 0:28:16 1 Connect promls_sysdba@localhost on
1 Query SET NAMES utf8
1 Init DB promls_box
1 Query select id , fullname from users where company_id = 7
1 Query SELECT FOUND_ROWS() AS total_results
1 Query select id , name from groups where company_id = 7 and context = "agency"
1 Query SELECT FOUND_ROWS() AS total_results
110211 0:28:26 1 Query SET NAMES utf8
1 Init DB promls_box
1 Query SELECT SQL_CALC_FOUND_ROWS property_id, property_type,property_name, bathrooms, bedrooms, for_sale, for_rent, for_vacational, units,
city,estate,country, rent_price, sale_price, min_price,mid_price, hig_price, max_sale,min_sale, min_rent, max_rent, min_vrental,max_vrental,
negotiation,property_status,
category_es category, image, dev_logo, inventory,company_id,broker_id FROM property_listing where property_status in('active','active-rented','active-sold') group by parent order by min_price asc, mid_price asc, hig_price asc, rent_price asc, sale_price asc limit 10, 10
1 Query SELECT FOUND_ROWS() AS total_results
1 Query SET NAMES utf8
1 Init DB promls_box
1 Query select st.id estate_id,st.description estate from active_locations group by estate_id
1 Query SELECT FOUND_ROWS() AS total_results
1 Query select city_id , city from active_locations where estate_id =272 group by city_id
1 Query SELECT FOUND_ROWS() AS total_results
1 Query SET NAMES utf8
1 Init DB promls_box
1 Query select f.id, f.name_es as name from families f inner join category_types ct on ct.category = f.id where type='category' group by f.id
1 Query SELECT FOUND_ROWS() AS total_results
1 Query select t.id, t.name_es as name from property_types t inner join category_types ct on ct.property_type = t.id where ct.category =12
1 Query SELECT FOUND_ROWS() AS total_results
110211 0:28:56 2 Connect root@localhost on
2 Query select id_area , description from areas where parent_area = 2
2 Quit
110211 0:28:58 1 Quit
110211 0:29:01 3 Connect promls_sysdba@localhost on
3 Query SET NAMES utf8
3 Init DB promls_box
3 Query select id , fullname from users where company_id = 7
3 Query SELECT FOUND_ROWS() AS total_results
3 Query select id , name from groups where company_id = 7 and context = "agency"
3 Query SELECT FOUND_ROWS() AS total_results
110211 0:29:43 4 Connect viva_webjun@localhost on
4 Init DB viva_blogen
4 Query SET NAMES utf8
4 Query select name, val from txp_prefs where prefs_id=1 AND user_name=''
4 Query select name, data from txp_lang where lang='en-gb' AND ( event='public' OR event='common')
4 Query select name, code, version from txp_plugin where status = 1 AND type IN (0,1) order by load_order
4 Query select page, css from txp_section where name = 'default' limit 1
4 Query select unix_timestamp(val) from txp_prefs where `name` = 'lastmod' limit 1
4 Query select name from txp_section where in_rss != '1'
4 Query select *, unix_timestamp(Posted) as uPosted, unix_timestamp(LastMod) as uLastMod, unix_timestamp(Expires) as uExpires, ID as thisid from textpattern where Status = 4 and Section != 'about' and Posted < now() and (now() <= Expires or Expires = '0000-00-00 00:00:00') order by Posted desc limit 5
4 Query select RealName from txp_users where `name` = 'webjunkie01' limit 1
4 Query select unix_timestamp(Posted) from textpattern where Posted <= now() and Status >= 4 order by Posted desc limit 1
4 Quit
110211 0:29:44 5 Connect viva_webjun@localhost on
5 Init DB viva_blogen
5 Query SET NAMES utf8
5 Query select name, val from txp_prefs where prefs_id=1 AND user_name=''
5 Query select name, data from txp_lang where lang='en-gb' AND ( event='public' OR event='common')
5 Query select name, code, version from txp_plugin where status = 1 AND type IN (0,1) order by load_order
5 Query select page, css from txp_section where name = 'default' limit 1
5 Query select unix_timestamp(val) from txp_prefs where `name` = 'lastmod' limit 1
5 Query select name from txp_section where in_rss != '1'
5 Query select *, unix_timestamp(Posted) as uPosted, unix_timestamp(LastMod) as uLastMod, unix_timestamp(Expires) as uExpires, ID as thisid from textpattern where Status = 4 and Section != 'about' and Posted < now() and (now() <= Expires or Expires = '0000-00-00 00:00:00') order by Posted desc limit 5
5 Query select RealName from txp_users where `name` = 'webjunkie01' limit 1
5 Query select unix_timestamp(Posted) from textpattern where Posted <= now() and Status >= 4 order by Posted desc limit 1
5 Quit
110211 0:30:02 6 Connect root@localhost on
6 Query show processlist
6 Quit
110211 0:32:28 3 Quit
7 Connect promls_sysdba@localhost on
7 Query SET NAMES utf8
7 Init DB promls_box
7 Query select id , fullname from users where company_id = 7
7 Query SELECT FOUND_ROWS() AS total_results
7 Query select id , name from groups where company_id = 7 and context = "agency"
7 Query SELECT FOUND_ROWS() AS total_results
110211 0:33:02 7 Query SET NAMES utf8
7 Init DB promls_box
7 Query SET NAMES utf8
7 Init DB promls_box
7 Query SET NAMES utf8
7 Init DB promls_box
7 Query select * from message_listing where (company_id = 7 or agency_id =7 ) and ( broker_id =2 ) and readed !=1 limit 0,5
7 Query SELECT FOUND_ROWS() AS total_results
7 Query SET NAMES utf8
7 Init DB promls_box
7 Query select * from message_listing where message_type = 'networking' and (category_message ='all' or (category_message='city' and area_id = 588 ) or (category_message='estate' and area_id = 272 ) or (category_message='agency' and company_id = 7 )) limit 0,5
7 Query SELECT FOUND_ROWS() AS total_results
7 Query SET NAMES utf8
7 Init DB promls_box
7 Query select SQL_CALC_FOUND_ROWS id , name, contact, email_contact, phone_contact, address, phone,
fax, email, website, creation_date, last_modification, area_id, area, status , logo, type, owner_id, users,
creator, creator_id
from companies_listing where status = 'active' and type in( 'sponsor' )
7 Query SELECT FOUND_ROWS() AS total_results
110211 0:34:20 7 Quit
110211 0:34:44 8 Connect promls_sysdba@localhost on
8 Query SET NAMES utf8
8 Init DB promls_box
8 Query select id , fullname from users where company_id = 7
8 Query SELECT FOUND_ROWS() AS total_results
8 Query select id , name from groups where company_id = 7 and context = "agency"
8 Query SELECT FOUND_ROWS() AS total_results
110211 0:35:01 9 Connect root@localhost on
9 Query show processlist
9 Quit
110211 0:35:16 8 Query SET NAMES utf8
8 Init DB promls_box
8 Query SELECT SQL_CALC_FOUND_ROWS property_id, property_type,property_name, bathrooms, bedrooms, for_sale, for_rent, for_vacational, units,
city,estate,country, rent_price, sale_price, min_price,mid_price, hig_price, max_sale,min_sale, min_rent, max_rent, min_vrental,max_vrental,
negotiation,property_status,
category_es category, image, dev_logo, inventory,company_id,broker_id FROM property_listing where property_status in('active','active-rented','active-sold') group by parent order by min_price asc, mid_price asc, hig_price asc, rent_price asc, sale_price asc limit 0, 10
110211 0:35:17 8 Query SELECT FOUND_ROWS() AS total_results
8 Query SET NAMES utf8
8 Init DB promls_box
8 Query select st.id estate_id,st.description estate from active_locations group by estate_id
8 Query SELECT FOUND_ROWS() AS total_results
8 Query select city_id , city from active_locations where estate_id =272 group by city_id
8 Query SELECT FOUND_ROWS() AS total_results
8 Query SET NAMES utf8
8 Init DB promls_box
8 Query select f.id, f.name_es as name from families f inner join category_types ct on ct.category = f.id where type='category' group by f.id
8 Query SELECT FOUND_ROWS() AS total_results
8 Query select t.id, t.name_es as name from property_types t inner join category_types ct on ct.property_type = t.id where ct.category =12
8 Query SELECT FOUND_ROWS() AS total_results
110211 0:35:43 10 Connect promls_sysdba@localhost on
10 Query SET NAMES utf8
10 Init DB promls_box
10 Query select id , fullname from users where company_id = 7
10 Query SELECT FOUND_ROWS() AS total_results
10 Query select id , name from groups where company_id = 7 and context = "agency"
10 Query SELECT FOUND_ROWS() AS total_results
8 Quit
110211 0:36:08 10 Query SET NAMES utf8
10 Init DB promls_box
10 Query SELECT SQL_CALC_FOUND_ROWS property_id, property_type,property_name, bathrooms, bedrooms, for_sale, for_rent, for_vacational, units,
city,estate,country, rent_price, sale_price, min_price,mid_price, hig_price, max_sale,min_sale, min_rent, max_rent, min_vrental,max_vrental,
negotiation,property_status,
category_es category, image, dev_logo, inventory,company_id,broker_id FROM property_listing where property_status in('active','active-rented','active-sold') group by parent order by min_price asc, mid_price asc, hig_price asc, rent_price asc, sale_price asc limit 0, 10
10 Query SELECT FOUND_ROWS() AS total_results
10 Query SET NAMES utf8
10 Init DB promls_box
10 Query select st.id estate_id,st.description estate from active_locations group by estate_id
10 Query SELECT FOUND_ROWS() AS total_results
10 Query select city_id , city from active_locations where estate_id =272 group by city_id
10 Query SELECT FOUND_ROWS() AS total_results
10 Query SET NAMES utf8
10 Init DB promls_box
10 Query select f.id, f.name_es as name from families f inner join category_types ct on ct.category = f.id where type='category' group by f.id
10 Query SELECT FOUND_ROWS() AS total_results
10 Query select t.id, t.name_es as name from property_types t inner join category_types ct on ct.property_type = t.id where ct.category =12
10 Query SELECT FOUND_ROWS() AS total_results
110211 0:36:21 10 Quit
110211 0:36:22 11 Connect promls_sysdba@localhost on
11 Query SET NAMES utf8
11 Init DB promls_box
11 Query select id , fullname from users where company_id = 7
11 Query SELECT FOUND_ROWS() AS total_results
11 Query select id , name from groups where company_id = 7 and context = "agency"
11 Query SELECT FOUND_ROWS() AS total_results
从 php 到 mysql 的连接似乎需要很长时间，我该如何解决？
【问题讨论】：
你能以更易读的方式格式化日志文件吗？ （这里，在***上） 这不是慢查询日志，而是查询日志 我不知道如何格式化。顺便说一句，我不知道它是否是一个日志，只有当我将下一个选项传递给 mysql restart 命令时才会创建这个文件：-log-slow-query 它为我创建了一个名为：og-slow-query 准备好修复日志慢查询的格式 @Angel Estrada：不是慢查询日志。 【参考方案1】：我发现您的日志更令人困惑而不是有用；这些查询看起来都不是特别复杂，但话又说回来，我不知道你的表结构是什么样的，你有多少服务器负载，或者你的表中有多少记录。如果查询本身实际上运行得很慢，您可能需要查看您的表是否已正确索引。如果您在 google 上搜索 mysql 索引，您可以找到大量信息，如果您真的想深入了解，请查看这本电子书： http://sql-performance-explained.com/
我还注意到有一些查询同时使用 ORDER BY 和 LIMIT 语句。在这种情况下，您应该查看这两个命令之间的关系，以及如何优化这些类型的查询。查看此网站了解更多信息：
http://www.mysqlperformanceblog.com/2006/09/01/order-by-limit-performance-optimization/
但如果不是查询，也许是因为您建议 MySQL 很难创建与数据库的连接。如果您有很多用户，可能是您的可用连接太少，您需要增加 MySQL 允许的并发连接数。看看这里如何做到这一点：
http://rackerhacker.com/2007/01/24/increase-mysql-connection-limit/
另一方面，如果连接数据库需要很长时间，您可能需要考虑在 PHP 应用程序中使用持久连接。
http://php.net/manual/en/features.persistent-connections.php
这些可以减少 PHP 连接数据库所花费的时间。或者，如果您已经在使用持久连接，那么可能是您的 PHP 代码中存在错误，该错误正在创建连接、启动事务，但随后保持它并占用连接，没有留下可用的连接其他用户。确保在 PHP 应用程序与数据库交互时使用正确的错误处理。
再次重申，这些只是我根据您展示的内容做出的一些最佳猜测。我希望它能让你朝着正确的方向前进。
【讨论】：
【参考方案2】：我继承了一个糟糕的 PHP 后端，它有数百次查询数据库的错误（站点索引页面最多 600 次）。客户不会为必要的后端重写付费，但我想加快速度。
查询都很小，在一个小型数据库上，但后端连接并为每个查询选择一个数据库。
我编写了以下快速代码来最小化这部分流量：
function DBConnectOnce($dbhost, $dbname, $dbuser, $dbpasswd)
$GLOBALS['dbrecord_Debug'] = ($dbDebug == 'true'); // @TODO better way to handle this within the class?
if(isset($GLOBALS['DBRECORDCONFIGSTATUS'])&&$GLOBALS['DBRECORDCONFIGSTATUS']==$dbuser.$dbname.$dbhost)//check to see if the connectionon the proper type has been made.
$dblink=unserialize($GLOBALS['DBRECORDCONNECTION']);
else
$GLOBALS['DBRECORDCONFIGSTATUS']=$dbuser.$dbname.$dbhost;
$GLOBALS['DBRECORDCONNECTION']=serialize($dblink = mysql_connect($dbhost, $dbuser, $dbpasswd));
if($dblink === false)
trigger_error("DBConnectOnce: Unable to connect to database server - ".mysql_error(), E_USER_ERROR);
if(!mysql_select_db($dbname))
trigger_error("DBConnectOnce: Unable to select database - $dbname at $dbhost", E_USER_ERROR);
return $dblink;
它使用全局变量来检查您已经建立的数据库连接类型，如果它与您需要建立的数据库连接类型匹配，则使用它。
【讨论】：
以上是关于使用慢日志查询的 PHP Mysql 优化的主要内容，如果未能解决你的问题，请参考以下文章