MySQL最佳化主要從其邏輯架構的
核心層
-分析、最佳化、執行階段著手。
實驗過程中,我使用的MySQL版本是5.7,關於MySQL的快速安裝,可參考 如何在CentOS7下快速安裝MySQL5.7。
MySQL調優,我們先從效能監控
開始,找點感覺。
show profile
show profile
是MySQL的查詢剖析工具。
mysql> show variables like '%profil%'; +------------------------+-------+| Variable_name | Value | +------------------------+-------+| have_profiling | YES | | profiling | OFF | | profiling_history_size | 15 | +------------------------+-------+3 rows in set (0.00 sec)复制代码
可以看到此工具預設是
OFF
(停用)的,可以透過伺服器變數在目前會話層級動態的修改
mysql> set profiling=1; Query OK, 0 rows affected, 1 warning (0.00 sec)复制代码
當設定完成之後,在伺服器上執行的所有語句,都會測量其耗費的時間和其他一些查詢執行狀態變更相關的資料。
我們來建立一個表格來示範一下:
mysql> create database mytest; Query OK, 1 row affected (0.01 sec) mysql> use mytest; Database changed mysql> create table stu(id int(8), name varchar(20)); Query OK, 0 rows affected (0.07 sec) mysql> insert into stu(id, name) values(1, 'Jack Ma'); Query OK, 1 row affected (0.01 sec) mysql> insert into stu(id, name) values(1, 'Pony'); Query OK, 1 row affected (0.03 sec)复制代码
查詢一下stu
表格:
mysql> select * from stu; +------+---------+| id | name | +------+---------+| 1 | Jack Ma | | 2 | Pony | +------+---------+2 rows in set (0.00 sec)复制代码
這個2 rows in set (0.00 sec)
就是用指令列執行一個sql後給的提示,而且還有執行時間0.00秒。
在mysql的命令列模式下只能顯示兩位小數的時間,可以使用以下命令查看具體的執行時間
mysql> show profile; +----------------------+----------+| Status | Duration | +----------------------+----------+| starting | 0.000061 | | checking permissions | 0.000009 | | Opening tables | 0.000019 | | init | 0.000020 | | System lock | 0.000009 | | optimizing | 0.000005 | | statistics | 0.000014 | | preparing | 0.000011 | | executing | 0.000003 | | Sending data | 0.000121 | | end | 0.000012 | | query end | 0.000011 | | closing tables | 0.000011 | | freeing items | 0.000021 | | cleaning up | 0.000021 | +----------------------+----------+复制代码
這樣我們可以清晰地看到,Starting
、System lock
系統鎖定、Sending data
等等一系列操作分別花了多少時間。
透過上面的示範,我們可以看到,show profile
一執行完就給我們回傳每個項目的執行時間,並沒有顯示是針對哪一條sql的,其實它就是列出了最近一條你執行的sql語句的執行時間。
那一定有可以看之前執行過的多條sql語句的情況吧?是的,show profiles
show profiles
mysql> show profiles; +----------+------------+------------------------------------------------+| Query_ID | Duration | Query | +----------+------------+------------------------------------------------+| 1 | 0.02907350 | create table stu(id int(8), name varchar(20)) | | 2 | 0.00337800 | create database mytest | | 3 | 0.02786850 | SELECT DATABASE() | | 4 | 0.00065300 | show databases | | 5 | 0.00086700 | show tables | | 6 | 0.06554900 | create table stu(id int(8), name varchar(20)) | | 7 | 0.00079850 | insert into stu(1, 'Jack') | | 8 | 0.06901975 | show create table stu | | 9 | 0.00008800 | insert into stu(1, 'Jack') | | 10 | 0.00616000 | insert into stu(id, name) values(1, 'Jack Ma') | | 11 | 0.03119675 | insert into stu(id, name) values(1, 'Pony') | | 12 | 0.03124900 | update stu set id=2 where name='Pony' | | 13 | 0.00036975 | select * from stu | | 14 | 0.00034925 | select * from stu | +----------+------------+------------------------------------------------+14 rows in set, 1 warning (0.00 sec)复制代码
列出了執行過的sql語句執行時間,我們可以挑選一個查詢來單獨看,例如我想看insert into stu(id, name) values(1, 'Pony')
這條語句的情況,可以用下面的操作:
mysql> show profile for query 11; +----------------------+----------+ | Status | Duration | +----------------------+----------+ | starting | 0.000072 | | checking permissions | 0.000009 | | Opening tables | 0.000021 | | init | 0.000018 | | System lock | 0.000008 | | update | 0.000094 | | end | 0.000005 | | query end | 0.030815 | | closing tables | 0.000036 | | freeing items | 0.000081 | | cleaning up | 0.000039 | +----------------------+----------+11 rows in set, 1 warning (0.00 sec)复制代码
也就是說可以根據Query_ID
結合show profile
來看。
MySQL官網上,關於show profile
語句:
SHOW PROFILE [type [, type] ... ] [FOR QUERY n] [LIMIT row_count [OFFSET offset]] type: { ALL | BLOCK IO | CONTEXT SWITCHES | CPU | IPC | MEMORY | PAGE FAULTS | SOURCE | SWAPS }复制代码
show profile
、for query n
前面已經說過了,下面我列一下每個type
是什麼意思。
all
顯示所有效能資訊
block io
顯示區塊io操作的次數
context switches
顯示上下文切換次數,被動和主動
cpu
顯示用戶cpu時間、系統cpu時間
IPC
顯示發送和接受的訊息數量
Memory
暫未實作
#page faults
顯示頁錯誤數量##source
##顯示原始碼中的函數名稱與位置
顯示swap的次數
##實操看看
我們針對mysql> show profiles; +----------+------------+-------------------------------+| Query_ID | Duration | Query | +----------+------------+-------------------------------+| 1 | 0.00009625 | show variable like '%profi%' | | 2 | 0.00373950 | show variables like '%profi%' | | 3 | 0.00025825 | select * from stu | +----------+------------+-------------------------------+3 rows in set, 1 warning (0.00 sec)复制代码登入後複製Query_ID
為3的進行檢視
#mysql> show profile all for query 3\G *************************** 1. row *************************** Status: starting Duration: 0.000070 CPU_user: 0.000000 CPU_system: 0.000064 Context_voluntary: 0Context_involuntary: 0 Block_ops_in: 0 Block_ops_out: 0 Messages_sent: 0 Messages_received: 0 Page_faults_major: 0 Page_faults_minor: 0 Swaps: 0 Source_function: NULL Source_file: NULL Source_line: NULL*************************** 2. row *************************** Status: checking permissions Duration: 0.000011 CPU_user: 0.000000 CPU_system: 0.000009 Context_voluntary: 0Context_involuntary: 0 Block_ops_in: 0 Block_ops_out: 0 Messages_sent: 0 Messages_received: 0 Page_faults_major: 0 Page_faults_minor: 0 Swaps: 0 Source_function: check_access Source_file: sql_authorization.cc Source_line: 809*************************** 3. row *************************** Status: Opening tables Duration: 0.000021 CPU_user: 0.000000 CPU_system: 0.000021 Context_voluntary: 0Context_involuntary: 0 Block_ops_in: 0 Block_ops_out: 0 Messages_sent: 0 Messages_received: 0 Page_faults_major: 0 Page_faults_minor: 0 Swaps: 0 Source_function: open_tables Source_file: sql_base.cc Source_line: 5793*************************** 4. row *************************** Status: init Duration: 0.000019 CPU_user: 0.000000 CPU_system: 0.000019 Context_voluntary: 0Context_involuntary: 0 Block_ops_in: 0 Block_ops_out: 0 Messages_sent: 0 Messages_received: 0 Page_faults_major: 0 Page_faults_minor: 0 Swaps: 0 Source_function: handle_query Source_file: sql_select.cc Source_line: 128*************************** 5. row *************************** Status: System lock Duration: 0.000009 CPU_user: 0.000000 CPU_system: 0.000009 Context_voluntary: 0Context_involuntary: 0 Block_ops_in: 0 Block_ops_out: 0 Messages_sent: 0 Messages_received: 0 Page_faults_major: 0 Page_faults_minor: 0 Swaps: 0 Source_function: mysql_lock_tables Source_file: lock.cc Source_line: 330*************************** 6. row *************************** Status: optimizing Duration: 0.000005 CPU_user: 0.000000 CPU_system: 0.000004 Context_voluntary: 0Context_involuntary: 0 Block_ops_in: 0 Block_ops_out: 0 Messages_sent: 0 Messages_received: 0 Page_faults_major: 0 Page_faults_minor: 0 Swaps: 0 Source_function: optimize Source_file: sql_optimizer.cc Source_line: 158*************************** 7. row *************************** Status: statistics Duration: 0.000014 CPU_user: 0.000000 CPU_system: 0.000014 Context_voluntary: 0Context_involuntary: 0 Block_ops_in: 0 Block_ops_out: 0 Messages_sent: 0 Messages_received: 0 Page_faults_major: 0 Page_faults_minor: 0 Swaps: 0 Source_function: optimize Source_file: sql_optimizer.cc Source_line: 374*************************** 8. row *************************** Status: preparing Duration: 0.000011 CPU_user: 0.000000 CPU_system: 0.000010 Context_voluntary: 0Context_involuntary: 0 Block_ops_in: 0 Block_ops_out: 0 Messages_sent: 0 Messages_received: 0 Page_faults_major: 0 Page_faults_minor: 0 Swaps: 0 Source_function: optimize Source_file: sql_optimizer.cc Source_line: 482*************************** 9. row *************************** Status: executing Duration: 0.000004 CPU_user: 0.000000 CPU_system: 0.000004 Context_voluntary: 0Context_involuntary: 0 Block_ops_in: 0 Block_ops_out: 0 Messages_sent: 0 Messages_received: 0 Page_faults_major: 0 Page_faults_minor: 0 Swaps: 0 Source_function: exec Source_file: sql_executor.cc Source_line: 126*************************** 10. row *************************** Status: Sending data Duration: 0.000050 CPU_user: 0.000000 CPU_system: 0.000050 Context_voluntary: 0Context_involuntary: 0 Block_ops_in: 0 Block_ops_out: 0 Messages_sent: 0 Messages_received: 0 Page_faults_major: 0 Page_faults_minor: 0 Swaps: 0 Source_function: exec Source_file: sql_executor.cc Source_line: 202*************************** 11. row *************************** Status: end Duration: 0.000005 CPU_user: 0.000000 CPU_system: 0.000005 Context_voluntary: 0Context_involuntary: 0 Block_ops_in: 0 Block_ops_out: 0 Messages_sent: 0 Messages_received: 0 Page_faults_major: 0 Page_faults_minor: 0 Swaps: 0 Source_function: handle_query Source_file: sql_select.cc Source_line: 206*************************** 12. row *************************** Status: query end Duration: 0.000008 CPU_user: 0.000000 CPU_system: 0.000008 Context_voluntary: 0Context_involuntary: 0 Block_ops_in: 0 Block_ops_out: 0 Messages_sent: 0 Messages_received: 0 Page_faults_major: 0 Page_faults_minor: 0 Swaps: 0 Source_function: mysql_execute_command Source_file: sql_parse.cc Source_line: 4956*************************** 13. row *************************** Status: closing tables Duration: 0.000008 CPU_user: 0.000000 CPU_system: 0.000008 Context_voluntary: 0Context_involuntary: 0 Block_ops_in: 0 Block_ops_out: 0 Messages_sent: 0 Messages_received: 0 Page_faults_major: 0 Page_faults_minor: 0 Swaps: 0 Source_function: mysql_execute_command Source_file: sql_parse.cc Source_line: 5009*************************** 14. row *************************** Status: freeing items Duration: 0.000013 CPU_user: 0.000000 CPU_system: 0.000013 Context_voluntary: 0Context_involuntary: 0 Block_ops_in: 0 Block_ops_out: 0 Messages_sent: 0 Messages_received: 0 Page_faults_major: 0 Page_faults_minor: 0 Swaps: 0 Source_function: mysql_parse Source_file: sql_parse.cc Source_line: 5622*************************** 15. row *************************** Status: cleaning up Duration: 0.000014 CPU_user: 0.000000 CPU_system: 0.000013 Context_voluntary: 0Context_involuntary: 0 Block_ops_in: 0 Block_ops_out: 0 Messages_sent: 0 Messages_received: 0 Page_faults_major: 0 Page_faults_minor: 0 Swaps: 0 Source_function: dispatch_command Source_file: sql_parse.cc Source_line: 193115 rows in set, 1 warning (0.00 sec)复制代码
\G的目的是格式化顯示查詢結果。
這裡我們可以看到,
show profile all
可以看到所有的效能資訊。
block io
mysql> show profile block io for query 3; +----------------------+----------+--------------+---------------+| Status | Duration | Block_ops_in | Block_ops_out | +----------------------+----------+--------------+---------------+| starting | 0.000070 | 0 | 0 | | checking permissions | 0.000011 | 0 | 0 | | Opening tables | 0.000021 | 0 | 0 | | init | 0.000019 | 0 | 0 | | System lock | 0.000009 | 0 | 0 | | optimizing | 0.000005 | 0 | 0 | | statistics | 0.000014 | 0 | 0 | | preparing | 0.000011 | 0 | 0 | | executing | 0.000004 | 0 | 0 | | Sending data | 0.000050 | 0 | 0 | | end | 0.000005 | 0 | 0 | | query end | 0.000008 | 0 | 0 | | closing tables | 0.000008 | 0 | 0 | | freeing items | 0.000013 | 0 | 0 | | cleaning up | 0.000014 | 0 | 0 | +----------------------+----------+--------------+---------------+15 rows in set, 1 warning (0.00 sec)复制代码
最後兩列就表示區塊io操作的次數。 type
的語法列出來,示範效果不在此贅述了。
show profile page faults for query 3
show profile source for query 3 #show profile swaps for query 3#show profile
用起來感覺很酷啊,他能幫我們看到sql語句各部分的執行時間,這對我們定位線上sql執行慢的問題很有幫助啊,but,官網上有這麼一段話:
The SHOW PROFILE and SHOW PROFILES statements are deprecated and will be removed in a future MySQL release. Use the Performance Schema instead;
MySQL的performance schema用于监控MySQL server在一个较低级别的运行过程中的资源消耗、资源等待等情况。使用performance schema可以更加容易的监控MySQL。
在MySQL的5.7版本中,性能模式是默认开启的,如果想要显式的关闭的话需要修改配置文件,不能直接进行修改,会报错Variable 'performance_schema' is a read only variable。
mysql> show variables like 'performance_schema'; +--------------------+-------+| Variable_name | Value | +--------------------+-------+| performance_schema | ON | +--------------------+-------+mysql> set performance_schema=0; ERROR 1238 (HY000): Variable 'performance_schema' is a read only variable复制代码
在配置文件中修改performance_schema的属性值,on表示开启,off表示关闭
[mysqld] performance_schema=ON复制代码
看下performance_schema
数据库有哪些功能表:
mysql> use performance_schema; mysql> show tables;复制代码
一共有87张表,分别存储着相关的信息。
可以通过show create table tablename来查看创建表的时候的表结构:
mysql> show create table threads\G *************************** 1. row *************************** Table: threadsCreate Table: CREATE TABLE `threads` ( `THREAD_ID` bigint(20) unsigned NOT NULL, `NAME` varchar(128) NOT NULL, `TYPE` varchar(10) NOT NULL, `PROCESSLIST_ID` bigint(20) unsigned DEFAULT NULL, `PROCESSLIST_USER` varchar(32) DEFAULT NULL, `PROCESSLIST_HOST` varchar(60) DEFAULT NULL, `PROCESSLIST_DB` varchar(64) DEFAULT NULL, `PROCESSLIST_COMMAND` varchar(16) DEFAULT NULL, `PROCESSLIST_TIME` bigint(20) DEFAULT NULL, `PROCESSLIST_STATE` varchar(64) DEFAULT NULL, `PROCESSLIST_INFO` longtext, `PARENT_THREAD_ID` bigint(20) unsigned DEFAULT NULL, `ROLE` varchar(64) DEFAULT NULL, `INSTRUMENTED` enum('YES','NO') NOT NULL, `HISTORY` enum('YES','NO') NOT NULL, `CONNECTION_TYPE` varchar(16) DEFAULT NULL, `THREAD_OS_ID` bigint(20) unsigned DEFAULT NULL) ENGINE=PERFORMANCE_SCHEMA DEFAULT CHARSET=utf81 row in set (0.00 sec)复制代码
我一看到threads
这个表名,还有THREAD_ID
、PROCESSLIST_ID
等等字段,就感觉这些比较重要啊有木有!
再来看一下setup_consumers
这个表:
mysql> show create table setup_consumers; +-----------------+--------------------------------------------------------------------------------------------------------------------------------------------------------+| Table | Create Table | +-----------------+--------------------------------------------------------------------------------------------------------------------------------------------------------+| setup_consumers | CREATE TABLE `setup_consumers` ( `NAME` varchar(64) NOT NULL, `ENABLED` enum('YES','NO') NOT NULL) ENGINE=PERFORMANCE_SCHEMA DEFAULT CHARSET=utf8 | +-----------------+--------------------------------------------------------------------------------------------------------------------------------------------------------+1 row in set (0.00 sec)复制代码
这个consumer
总感觉似曾相识啊!
其实,想要搞明白后续的内容,我们需要理解两个基本概念:
performance_schema库下的表可以按照监视不同的纬度就行分组。
--语句事件记录表,这些表记录了语句事件信息,当前语句事件表events_statements_current、历史语句事件表events_statements_history和长语句历史事件表events_statements_history_long、以及聚合后的摘要表summary,其中,summary表还可以根据帐号(account),主机(host),程序(program),线程(thread),用户(user)和全局(global)再进行细分)show tables like '%statement%';--等待事件记录表,与语句事件类型的相关记录表类似:show tables like '%wait%';--阶段事件记录表,记录语句执行的阶段事件的表show tables like '%stage%';--事务事件记录表,记录事务相关的事件的表show tables like '%transaction%';--监控文件系统层调用的表show tables like '%file%';--监视内存使用的表show tables like '%memory%';--动态对performance_schema进行配置的配置表show tables like '%setup%';复制代码
来,那一个事务
相关的来瞅一眼:
mysql> show tables like '%transaction%'; +------------------------------------------------------+| Tables_in_performance_schema (%transaction%) | +------------------------------------------------------+| events_transactions_current | | events_transactions_history | | events_transactions_history_long | | events_transactions_summary_by_account_by_event_name | | events_transactions_summary_by_host_by_event_name | | events_transactions_summary_by_thread_by_event_name | | events_transactions_summary_by_user_by_event_name | | events_transactions_summary_global_by_event_name | +------------------------------------------------------+8 rows in set (0.00 sec)复制代码
数据库刚刚初始化并启动时,并非所有instruments(事件采集项,在采集项的配置表中每一项都有一个开关字段,或为YES,或为NO)和consumers(与采集项类似,也有一个对应的事件类型保存表配置项,为YES就表示对应的表保存性能数据,为NO就表示对应的表不保存性能数据)都启用了,所以默认不会收集所有的事件。
可能你需要检测的事件并没有打开,需要进行设置,可以使用如下两个语句打开对应的instruments和consumers(行计数可能会因MySQL版本而异)。
打开等待事件
的采集器配置项开关,需要修改setup_instruments
配置表中对应的采集器配置项
mysql> UPDATE setup_instruments SET ENABLED = 'YES', TIMED = 'YES' where name like 'wait%'; Query OK, 267 rows affected (0.03 sec) Rows matched: 321 Changed: 267 Warnings: 0复制代码
打开等待事件
的保存表配置开关,修改setup_consumers
配置表中对应的配置项
mysql> UPDATE setup_consumers SET ENABLED = 'YES' where name like '%wait%'; Query OK, 3 rows affected (0.04 sec) Rows matched: 3 Changed: 3 Warnings: 0复制代码
当配置完成之后可以查看当前server正在做什么。
events_waits_current
表来得知,该表中每个线程只包含一行数据,用于显示每个线程的最新监视事件:释义:
该信息表示线程id为3的线程正在等待buf_dblwr_mutex锁,等待时间为69917。 属性说明: id:事件来自哪个线程,事件编号是多少 event_name:表示检测到的具体的内容 source:表示这个检测代码在哪个源文件中以及行号 timer_start:表示该事件的开始时间 timer_end:表示该事件的结束时间 timer_wait:表示该事件总的花费时间 注意:_current表中每个线程只保留一条记录,一旦线程完成工作,该表中不会再记录该线程的事件信息。复制代码
_history
表中记录每个线程应该执行完成的事件信息,但每个线程的事件信息只会记录10条,再多就会被覆盖,(下面的演示可以看到每个线程只有10条记录)。_history_long
表中记录所有线程的事件信息,但总记录数量是10000,超过就会被覆盖掉mysql> mysql> select thread_id,event_id,event_name,timer_wait from events_waits_history order by thread_id limit 21; +-----------+----------+------------------------------------------+------------+| thread_id | event_id | event_name | timer_wait | +-----------+----------+------------------------------------------+------------+| 3 | 4042 | wait/synch/mutex/innodb/buf_pool_mutex | 47785 | | 3 | 4043 | wait/synch/mutex/innodb/flush_list_mutex | 43761 | | 3 | 4044 | wait/synch/mutex/innodb/buf_pool_mutex | 58851 | | 3 | 4045 | wait/synch/mutex/innodb/buf_dblwr_mutex | 73438 | | 3 | 4046 | wait/synch/mutex/innodb/buf_pool_mutex | 222829 | | 3 | 4047 | wait/synch/mutex/innodb/buf_pool_mutex | 42755 | | 3 | 4048 | wait/synch/mutex/innodb/flush_list_mutex | 44767 | | 3 | 4049 | wait/synch/mutex/innodb/buf_pool_mutex | 59857 | | 3 | 4050 | wait/synch/mutex/innodb/buf_dblwr_mutex | 72432 | | 3 | 4041 | wait/synch/mutex/innodb/buf_pool_mutex | 243452 | | 17 | 2442 | wait/synch/mutex/innodb/sync_array_mutex | 74947 | | 17 | 2440 | wait/synch/mutex/innodb/sync_array_mutex | 82492 | | 17 | 2439 | wait/synch/mutex/innodb/sync_array_mutex | 458233 | | 17 | 2438 | wait/synch/mutex/innodb/log_sys_mutex | 239428 | | 17 | 2437 | wait/synch/mutex/innodb/sync_array_mutex | 85510 | | 17 | 2446 | wait/synch/mutex/innodb/sync_array_mutex | 84504 | | 17 | 2445 | wait/synch/mutex/innodb/sync_array_mutex | 77462 | | 17 | 2444 | wait/synch/mutex/innodb/log_sys_mutex | 238422 | | 17 | 2443 | wait/synch/mutex/innodb/sync_array_mutex | 83498 | | 17 | 2441 | wait/synch/mutex/innodb/log_sys_mutex | 229368 | | 19 | 7281 | wait/synch/mutex/innodb/flush_list_mutex | 43761 | +-----------+----------+------------------------------------------+------------+21 rows in set (0.00 sec)复制代码
summary
表提供所有事件的汇总信息,该组中的表以不同的方式汇总事件数据(如:按用户,按主机,按线程等等)。例如:要查看哪些instruments占用最多的时间,可以通过对events_waits_summary_global_by_event_name表的COUNT_STAR或SUM_TIMER_WAIT列进行查询(这两列是对事件的记录数执行COUNT(*)、事件记录的TIMER_WAIT列执行SUM(TIMER_WAIT)统计而来)
mysql> SELECT EVENT_NAME,COUNT_STAR FROM events_waits_summary_global_by_event_name ORDER BY COUNT_STAR DESC LIMIT 10; +-----------------------------------------------+------------+| EVENT_NAME | COUNT_STAR | +-----------------------------------------------+------------+| wait/synch/mutex/innodb/buf_pool_mutex | 3357 | | wait/synch/mutex/innodb/log_sys_mutex | 3357 | | wait/synch/mutex/innodb/sync_array_mutex | 2241 | | wait/synch/mutex/innodb/flush_list_mutex | 2238 | | wait/io/file/sql/FRM | 1590 | | wait/synch/mutex/innodb/log_flush_order_mutex | 1119 | | wait/synch/mutex/innodb/log_sys_write_mutex | 1119 | | wait/synch/mutex/innodb/fil_system_mutex | 1119 | | wait/synch/mutex/innodb/dict_sys_mutex | 1119 | | wait/synch/mutex/innodb/buf_dblwr_mutex | 1119 | +-----------------------------------------------+------------+10 rows in set (0.10 sec)复制代码
instance
表记录了哪些类型的对象会被检测。这些对象在被server使用时,在该表中将会产生一条事件记录。例如,file_instances表列出了文件I/O操作及其关联文件名
mysql> select * from file_instances limit 10; +----------------------------------------+--------------------------------------+------------+| FILE_NAME | EVENT_NAME | OPEN_COUNT | +----------------------------------------+--------------------------------------+------------+| /usr/share/mysql/english/errmsg.sys | wait/io/file/sql/ERRMSG | 0 | | /usr/share/mysql/charsets/Index.xml | wait/io/file/mysys/charset | 0 | | /var/lib/mysql/ibdata1 | wait/io/file/innodb/innodb_data_file | 3 | | /var/lib/mysql/ib_logfile0 | wait/io/file/innodb/innodb_log_file | 2 | | /var/lib/mysql/ib_logfile1 | wait/io/file/innodb/innodb_log_file | 2 | | /var/lib/mysql/mysql/engine_cost.ibd | wait/io/file/innodb/innodb_data_file | 3 | | /var/lib/mysql/mysql/gtid_executed.ibd | wait/io/file/innodb/innodb_data_file | 3 | | /var/lib/mysql/mysql/help_category.ibd | wait/io/file/innodb/innodb_data_file | 3 | | /var/lib/mysql/mysql/help_keyword.ibd | wait/io/file/innodb/innodb_data_file | 3 | | /var/lib/mysql/mysql/help_relation.ibd | wait/io/file/innodb/innodb_data_file | 3 | +----------------------------------------+--------------------------------------+------------+10 rows in set (0.00 sec)复制代码
performance_schema_consumer_events_statements_current=TRUE#是否在mysql server启动时就开启events_statements_current表的记录功能(该表记录当前的语句事件信息),启动之后也可以在setup_consumers表中使用UPDATE语句进行动态更新setup_consumers配置表中的events_statements_current配置项,默认值为TRUEperformance_schema_consumer_events_statements_history=TRUE#与performance_schema_consumer_events_statements_current选项类似,但该选项是用于配置是否记录语句事件短历史信息,默认为TRUEperformance_schema_consumer_events_stages_history_long=FALSE#与performance_schema_consumer_events_statements_current选项类似,但该选项是用于配置是否记录语句事件长历史信息,默认为FALSE除了statement(语句)事件之外,还支持:wait(等待)事件、state(阶段)事件、transaction(事务)事件,他们与statement事件一样都有三个启动项分别进行配置,但这些等待事件默认未启用,如果需要在MySQL Server启动时一同启动,则通常需要写进my.cnf配置文件中 performance_schema_consumer_global_instrumentation=TRUE 是否在MySQL Server启动时就开启全局表(如:mutex_instances、rwlock_instances、cond_instances、file_instances、users、hostsaccounts、socket_summary_by_event_name、file_summary_by_instance等大部分的全局对象计数统计和事件汇总统计信息表 )的记录功能,启动之后也可以在setup_consumers表中使用UPDATE语句进行动态更新全局配置项 默认值为TRUEperformance_schema_consumer_statements_digest=TRUE是否在MySQL Server启动时就开启events_statements_summary_by_digest 表的记录功能,启动之后也可以在setup_consumers表中使用UPDATE语句进行动态更新digest配置项 默认值为TRUEperformance_schema_consumer_thread_instrumentation=TRUE是否在MySQL Server启动时就开启 events_xxx_summary_by_yyy_by_event_name表的记录功能,启动之后也可以在setup_consumers表中使用UPDATE语句进行动态更新线程配置项 默认值为TRUEperformance_schema_instrument[=name] 是否在MySQL Server启动时就启用某些采集器,由于instruments配置项多达数千个,所以该配置项支持key-value模式,还支持%号进行通配等,如下:# [=name]可以指定为具体的Instruments名称(但是这样如果有多个需要指定的时候,就需要使用该选项多次),也可以使用通配符,可以指定instruments相同的前缀+通配符,也可以使用%代表所有的instruments## 指定开启单个instruments--performance-schema-instrument= 'instrument_name=value'## 使用通配符指定开启多个instruments--performance-schema-instrument= 'wait/synch/cond/%=COUNTED'## 开关所有的instruments--performance-schema-instrument= '%=ON'--performance-schema-instrument= '%=OFF'注意,这些启动选项要生效的前提是,需要设置performance_schema=ON。另外,这些启动选项虽然无法使用show variables语句查看,但我们可以通过setup_instruments和setup_consumers表查询这些选项指定的值。复制代码
show variables like '%performance_schema%'; --重要的属性解释 performance_schema=ON /* 控制performance_schema功能的开关,要使用MySQL的performance_schema,需要在mysqld启动时启用,以启用事件收集功能 该参数在5.7.x之前支持performance_schema的版本中默认关闭,5.7.x版本开始默认开启 注意:如果mysqld在初始化performance_schema时发现无法分配任何相关的内部缓冲区,则performance_schema将自动禁用,并将performance_schema设置为OFF */ performance_schema_digests_size=10000 /* 控制events_statements_summary_by_digest表中的最大行数。如果产生的语句摘要信息超过此最大值,便无法继续存入该表,此时performance_schema会增加状态变量 */ performance_schema_events_statements_history_long_size=10000 /* 控制events_statements_history_long表中的最大行数,该参数控制所有会话在events_statements_history_long表中能够存放的总事件记录数,超过这个限制之后,最早的记录将被覆盖 全局变量,只读变量,整型值,5.6.3版本引入 * 5.6.x版本中,5.6.5及其之前的版本默认为10000,5.6.6及其之后的版本默认值为-1,通常情况下,自动计算的值都是10000 * 5.7.x版本中,默认值为-1,通常情况下,自动计算的值都是10000 */ performance_schema_events_statements_history_size=10 /* 控制events_statements_history表中单个线程(会话)的最大行数,该参数控制单个会话在events_statements_history表中能够存放的事件记录数,超过这个限制之后,单个会话最早的记录将被覆盖 全局变量,只读变量,整型值,5.6.3版本引入 * 5.6.x版本中,5.6.5及其之前的版本默认为10,5.6.6及其之后的版本默认值为-1,通常情况下,自动计算的值都是10 * 5.7.x版本中,默认值为-1,通常情况下,自动计算的值都是10 除了statement(语句)事件之外,wait(等待)事件、state(阶段)事件、transaction(事务)事件,他们与statement事件一样都有三个参数分别进行存储限制配置,有兴趣的同学自行研究,这里不再赘述 */ performance_schema_max_digest_length=1024 /* 用于控制标准化形式的SQL语句文本在存入performance_schema时的限制长度,该变量与max_digest_length变量相关(max_digest_length变量含义请自行查阅相关资料) 全局变量,只读变量,默认值1024字节,整型值,取值范围0~1048576 */ performance_schema_max_sql_text_length=1024 /* 控制存入events_statements_current,events_statements_history和events_statements_history_long语句事件表中的SQL_TEXT列的最大SQL长度字节数。 超出系统变量performance_schema_max_sql_text_length的部分将被丢弃,不会记录,一般情况下不需要调整该参数,除非被截断的部分与其他SQL比起来有很大差异 全局变量,只读变量,整型值,默认值为1024字节,取值范围为0~1048576,5.7.6版本引入 降低系统变量performance_schema_max_sql_text_length值可以减少内存使用,但如果汇总的SQL中,被截断部分有较大差异,会导致没有办法再对这些有较大差异的SQL进行区分。 增加该系统变量值会增加内存使用,但对于汇总SQL来讲可以更精准地区分不同的部分。 */复制代码
基本了解了表的相关信息之后,可以通过这些表进行实际的查询操作来进行实际的分析。
--1、哪类的SQL执行最多?SELECT DIGEST_TEXT,COUNT_STAR,FIRST_SEEN,LAST_SEEN FROM events_statements_summary_by_digest ORDER BY COUNT_STAR DESC--2、哪类SQL的平均响应时间最多?SELECT DIGEST_TEXT,AVG_TIMER_WAIT FROM events_statements_summary_by_digest ORDER BY COUNT_STAR DESC--3、哪类SQL排序记录数最多?SELECT DIGEST_TEXT,SUM_SORT_ROWS FROM events_statements_summary_by_digest ORDER BY COUNT_STAR DESC--4、哪类SQL扫描记录数最多?SELECT DIGEST_TEXT,SUM_ROWS_EXAMINED FROM events_statements_summary_by_digest ORDER BY COUNT_STAR DESC--5、哪类SQL使用临时表最多?SELECT DIGEST_TEXT,SUM_CREATED_TMP_TABLES,SUM_CREATED_TMP_DISK_TABLES FROM events_statements_summary_by_digest ORDER BY COUNT_STAR DESC--6、哪类SQL返回结果集最多?SELECT DIGEST_TEXT,SUM_ROWS_SENT FROM events_statements_summary_by_digest ORDER BY COUNT_STAR DESC--7、哪个表物理IO最多?SELECT file_name,event_name,SUM_NUMBER_OF_BYTES_READ,SUM_NUMBER_OF_BYTES_WRITE FROM file_summary_by_instance ORDER BY SUM_NUMBER_OF_BYTES_READ + SUM_NUMBER_OF_BYTES_WRITE DESC--8、哪个表逻辑IO最多?SELECT object_name,COUNT_READ,COUNT_WRITE,COUNT_FETCH,SUM_TIMER_WAIT FROM table_io_waits_summary_by_table ORDER BY sum_timer_wait DESC--9、哪个索引访问最多?SELECT OBJECT_NAME,INDEX_NAME,COUNT_FETCH,COUNT_INSERT,COUNT_UPDATE,COUNT_DELETE FROM table_io_waits_summary_by_index_usage ORDER BY SUM_TIMER_WAIT DESC--10、哪个索引从来没有用过?SELECT OBJECT_SCHEMA,OBJECT_NAME,INDEX_NAME FROM table_io_waits_summary_by_index_usage WHERE INDEX_NAME IS NOT NULL AND COUNT_STAR = 0 AND OBJECT_SCHEMA <> 'mysql' ORDER BY OBJECT_SCHEMA,OBJECT_NAME;--11、哪个等待事件消耗时间最多?SELECT EVENT_NAME,COUNT_STAR,SUM_TIMER_WAIT,AVG_TIMER_WAIT FROM events_waits_summary_global_by_event_name WHERE event_name != 'idle' ORDER BY SUM_TIMER_WAIT DESC--12-1、剖析某条SQL的执行情况,包括statement信息,stege信息,wait信息SELECT EVENT_ID,sql_text FROM events_statements_history WHERE sql_text LIKE '%count(*)%';--12-2、查看每个阶段的时间消耗SELECT event_id,EVENT_NAME,SOURCE,TIMER_END - TIMER_START FROM events_stages_history_long WHERE NESTING_EVENT_ID = 1553;--12-3、查看每个阶段的锁等待情况SELECT event_id,event_name,source,timer_wait,object_name,index_name,operation,nesting_event_id FROM events_waits_history_longWHERE nesting_event_id = 1553;复制代码
关于Performance Schema的东西,我们还可以参考官网继续进行详细的了解。
我们来思考一下,否可以做一个系统出来,通过查询Performance Schema
的表的一些信息,并将其展示到web端,这样我们就可以进行可视化监控了?
使用show processlist查看连接的线程个数,来观察是否有大量线程处于不正常的状态或者其他不正常的特征。
mysql> show processlist; +----+------+-----------+--------------------+---------+------+----------+------------------+| Id | User | Host | db | Command | Time | State | Info | +----+------+-----------+--------------------+---------+------+----------+------------------+| 6 | root | localhost | performance_schema | Query | 0 | starting | show processlist | +----+------+-----------+--------------------+---------+------+----------+------------------+1 row in set (0.00 sec)复制代码
属性说明:
id
session id
user
操作的用户
host
操作的主机
db
操作的数据库
command
当前状态
- sleep:线程正在等待客户端发送新的请求
- query:线程正在执行查询或正在将结果发送给客户端
- locked:在mysql的服务层,该线程正在等待表锁
- analyzing and statistics:线程正在收集存储引擎的统计信息,并生成查询的执行计划
- Copying to tmp table:线程正在执行查询,并且将其结果集都复制到一个临时表中
- sorting result:线程正在对结果集进行排序
- sending data:线程可能在多个状态之间传送数据,或者在生成结果集或者向客户端返回数据
info
详细的sql语句
time
相应命令执行时间
state
命令执行状态
MySQL调优之性能监控:
show profile
使用show profile查询剖析工具,可以指定具体的typeperformance schema
使用performance schema来更加容易的监控mysqlshow processlist
使用show processlist查看连接的线程个数,来观察是否有大量线程处于不正常的状态或者其他不正常的特征
更多相关免费学习推荐:mysql教程(视频)
以上是初識MySQL調優之效能監控的詳細內容。更多資訊請關注PHP中文網其他相關文章!