背景:
MySQL 5.5开始新增一个数据库:PERFORMANCE_SCHEMA,主要用于收集数据库服务器性能参数。并且库里表的存储引擎均为PERFORMANCE_SCHEMA,而用户是不能创建存储引擎为PERFORMANCE_SCHEMA的表。MySQL5.5默认是关闭的,需要手动开启,在配置文件里添加:
1.[mysqld]2.performance_schema=ON查看是否开启:
1.mysql>show
variables like ‘performance_schema‘;2.+--------------------+-------+3.|
Variable_name | Value |4.+--------------------+-------+5.|
performance_schema | <strong>ON</strong> |6.+--------------------+-------+从MySQL5.6开始,默认打开,本文就从MySQL5.6来说明,在数据库使用当中PERFORMANCE_SCHEMA的一些比较常用的功能。具体的信息可以查看官方文档。
相关表信息:
一:配置(setup)表:
01.zjy@performance_schema 10:16:56>show
tables like ‘%setup%‘;02.+----------------------------------------+03.|
Tables_in_performance_schema (%setup%) |04.+----------------------------------------+05.|
setup_actors |06.|
setup_consumers |07.|
setup_instruments |08.|
setup_objects |09.|
setup_timers |10.+----------------------------------------+1,setup_actors:配置用户纬度的监控,默认监控所有用户。
1.zjy@performance_schema 10:19:11>select
* from setup_actors;2.+------+------+------+3.|
HOST | USER | ROLE |4.+------+------+------+5.|
% | % | % |6.+------+------+------+2,setup_consumers:配置events的消费者类型,即收集的events写入到哪些统计表中。
01.zjy@:
performance_schema 10:23:35>select
* from setup_consumers;02.+--------------------------------+---------+03.|
NAME | ENABLED |04.+--------------------------------+---------+05.|
events_stages_current | NO |06.|
events_stages_history | NO |07.|
events_stages_history_long | NO |08.|
events_statements_current | YES |09.|
events_statements_history | NO |10.|
events_statements_history_long | NO |11.|
events_waits_current | NO |12.|
events_waits_history | NO |13.|
events_waits_history_long | NO |14.|
global_instrumentation | YES |15.|
thread_instrumentation | YES |16.|
statements_digest | YES |17.+--------------------------------+---------+这里需要说明的是需要查看哪个就更新其ENABLED列为YES。如:
1.zjy@performance_schema 10:25:02>update
setup_consumers set ENABLED=‘YES‘ where
NAME in (‘events_stages_current‘,‘events_waits_current‘);2.Query
OK, 2 rows
affected (0.00 sec)更新完后立即生效,但是服务器重启之后又会变回默认值,要永久生效需要在配置文件里添加:
1.[mysqld]2.#performance_schema3.performance_schema_consumer_events_waits_current=on4.performance_schema_consumer_events_stages_current=on5.performance_schema_consumer_events_statements_current=on6.performance_schema_consumer_events_waits_history=on7.performance_schema_consumer_events_stages_history=on8.performance_schema_consumer_events_statements_history=on即在这些表的前面加上:performance_schema_consumer_xxx。表setup_consumers里面的值有个层级关系:
1.<strong>global_instrumentation</strong>
> <strong>thread_instrumentation</strong> = <strong>statements_digest</strong> > events_stages_<strong>current</strong> = events_statements_current = events_waits_current > events_stages_<strong>history</strong> = events_statements_history = events_waits_history
> events_stages_<strong>history_long</strong> = events_statements_history_long = events_waits_history_long只有上一层次的为YES,才会继续检查该本层为YES or NO。global_instrumentation是最高级别consumer,如果它设置为NO,则所有的consumer都会忽略。其中history和history_long存的是current表的历史记录条数,history表记录了每个线程最近等待的10个事件,而history_long表则记录了最近所有线程产生的10000个事件,这里的10和10000都是可以配置的。这三个表表结构相同,history和history_long表数据都来源于current表。长度通过控制参数:
01.zjy@performance_schema 11:10:03>show
variables like ‘performance_schema%history%size‘;02.+--------------------------------------------------------+-------+03.|
Variable_name | Value |04.+--------------------------------------------------------+-------+05.|
performance_schema_events_stages_history_long_size | 10000 |06.|
performance_schema_events_stages_history_size | 10 |07.|
performance_schema_events_statements_history_long_size | 10000 |08.|
performance_schema_events_statements_history_size | 10 |09.|
performance_schema_events_waits_history_long_size | 10000 |10.|
performance_schema_events_waits_history_size | 10 |11.+--------------------------------------------------------+-------+3,setup_instruments:配置具体的instrument,主要包含4大类:idle、stage/xxx、statement/xxx、wait/xxx:
01.zjy@performance_schema 10:56:35>select
name,count(*) from setup_instruments group by LEFT(name,5);02.+---------------------------------+----------+03.|
name | count(*) |04.+---------------------------------+----------+05.|
idle | 1 |06.|
stage/sql/After create | 111 |07.|
statement/sql/select | 179 |08.|
wait/synch/mutex/sql/PAGE::lock | 296 |09.+---------------------------------+----------+idle表示socket空闲的时间,stage类表示语句的每个执行阶段的统计,statement类统计语句维度的信息,wait类统计各种等待事件,比如IO,mutux,spin_lock,condition等。
4,setup_objects:配置监控对象,默认对mysql,performance_schema和information_schema中的表都不监控,而其它DB的所有表都监控。
01.zjy@performance_schema 11:00:18>select
* from setup_objects;02.+-------------+--------------------+-------------+---------+-------+03.|
OBJECT_TYPE | OBJECT_SCHEMA | OBJECT_NAME | ENABLED | TIMED |04.+-------------+--------------------+-------------+---------+-------+05.|
TABLE | mysql | % | NO | NO |06.|
TABLE | performance_schema | % | NO | NO |07.|
TABLE | information_schema | % | NO | NO |08.|
TABLE | % | % | <strong>YES</strong> | <strong>YES</strong> |09.+-------------+--------------------+-------------+---------+-------+5,setup_timers:配置每种类型指令的统计时间单位。MICROSECOND表示统计单位是微妙,CYCLE表示统计单位是时钟周期,时间度量与CPU的主频有关,NANOSECOND表示统计单位是纳秒。但无论采用哪种度量单位,最终统计表中统计的时间都会装换到皮秒。(1秒=1000000000000皮秒)
01.zjy@performance_schema 11:05:12>select
* from setup_timers;02.+-----------+-------------+03.|
NAME | TIMER_NAME |04.+-----------+-------------+05.|
idle | MICROSECOND |06.|
wait | CYCLE |07.|
stage | NANOSECOND |08.|
statement | NANOSECOND |09.+-----------+-------------+二:instance表
1,cond_instances:条件等待对象实例
表中记录了系统中使用的条件变量的对象,OBJECT_INSTANCE_BEGIN为对象的内存地址。
2,file_instances:文件实例
表中记录了系统中打开了文件的对象,包括ibdata文件,redo文件,binlog文件,用户的表文件等,open_count显示当前文件打开的数目,如果重来没有打开过,不会出现在表中。
01.zjy@performance_schema 11:20:04>select
* from file_instances limit 2,5;02.+---------------------------------+--------------------------------------+------------+03.|
FILE_NAME | EVENT_NAME | <strong>OPEN_COUNT</strong> |04.+---------------------------------+--------------------------------------+------------+05.|
/var/lib/mysql/mysql/plugin.frm | wait/io/file/sql/FRM | 0 |06.|
/var/lib/mysql/mysql/plugin.MYI | wait/io/file/myisam/kfile | 1 |07.|
/var/lib/mysql/mysql/plugin.MYD | wait/io/file/myisam/dfile | 1 |08.|
/var/lib/mysql/ibdata1 | wait/io/file/innodb/innodb_data_file | 2 |09.|
/var/lib/mysql/ib_logfile0 | wait/io/file/innodb/innodb_log_file | 2 |10.+---------------------------------+--------------------------------------+------------+3,mutex_instances:互斥同步对象实例
表中记录了系统中使用互斥量对象的所有记录,其中name为:wait/synch/mutex/*。LOCKED_BY_THREAD_ID显示哪个线程正持有mutex,若没有线程持有,则为NULL。
4,rwlock_instances: 读写锁同步对象实例
表中记录了系统中使用读写锁对象的所有记录,其中name为 wait/synch/rwlock/*。WRITE_LOCKED_BY_THREAD_ID为正在持有该对象的thread_id,若没有线程持有,则为NULL。READ_LOCKED_BY_COUNT为记录了同时有多少个读者持有读锁。(通过 events_waits_current 表可以知道,哪个线程在等待锁;通过rwlock_instances知道哪个线程持有锁。rwlock_instances的缺陷是,只能记录持有写锁的线程,对于读锁则无能为力)。
5,socket_instances:活跃会话对象实例
表中记录了thread_id,socket_id,ip和port,其它表可以通过thread_id与socket_instance进行关联,获取IP-PORT信息,能够与应用对接起来。
event_name主要包含3类:
wait/io/socket/sql/server_unix_socket,服务端unix监听socket
wait/io/socket/sql/server_tcpip_socket,服务端tcp监听socket
wait/io/socket/sql/client_connection,客户端socket
三:Wait表
1,events_waits_current:记录了当前线程等待的事件
2,events_waits_history:记录了每个线程最近等待的10个事件
3,events_waits_history_long:记录了最近所有线程产生的10000个事件
表结构定义如下:
01.CREATE
TABLE `events_waits_current` (02.`THREAD_ID`
bigint(20)
unsigned NOT NULL COMMENT ‘线程ID‘,03.`EVENT_ID`
bigint(20)
unsigned NOT NULL COMMENT ‘当前线程的事件ID,和THREAD_ID确定唯一‘,04.`END_EVENT_ID`
bigint(20)
unsigned DEFAULT NULL COMMENT ‘当事件开始时,这一列被设置为NULL。当事件结束时,再更新为当前的事件ID‘,05.`EVENT_NAME`
varchar(128)
NOT NULL COMMENT ‘事件名称‘,06.`SOURCE`
varchar(64)
DEFAULT NULL COMMENT ‘该事件产生时的源码文件‘,07.`TIMER_START`
bigint(20)
unsigned DEFAULT NULL COMMENT ‘事件开始时间(皮秒)‘,08.`TIMER_END`
bigint(20)
unsigned DEFAULT NULL COMMENT ‘事件结束结束时间(皮秒)‘,09.`TIMER_WAIT`
bigint(20)
unsigned DEFAULT NULL COMMENT ‘事件等待时间(皮秒)‘,10.`SPINS` int(10)
unsigned DEFAULT NULL COMMENT ‘‘,11.`OBJECT_SCHEMA`
varchar(64)
DEFAULT NULL COMMENT ‘库名‘,12.`OBJECT_NAME`
varchar(512)
DEFAULT NULL COMMENT ‘文件名、表名、IP:SOCK值‘,13.`OBJECT_TYPE`
varchar(64)
DEFAULT NULL COMMENT ‘FILE、TABLE、TEMPORARY
TABLE‘,14.`INDEX_NAME`
varchar(64)
DEFAULT NULL COMMENT ‘索引名‘,15.`OBJECT_INSTANCE_BEGIN`
bigint(20)
unsigned NOT NULL COMMENT ‘内存地址‘,16.`NESTING_EVENT_ID`
bigint(20)
unsigned DEFAULT NULL COMMENT ‘该事件对应的父事件ID‘,17.`NESTING_EVENT_TYPE` enum(‘STATEMENT‘,‘STAGE‘,‘WAIT‘)
DEFAULT NULL COMMENT ‘父事件类型(STATEMENT,
STAGE, WAIT)‘,18.`OPERATION`
varchar(32)
NOT NULL COMMENT ‘操作类型(lock,
read, write)‘,19.`NUMBER_OF_BYTES`
bigint(20)
DEFAULT NULL COMMENT ‘‘,20.`FLAGS` int(10)
unsigned DEFAULT NULL COMMENT ‘标记‘21.)
ENGINE=PERFORMANCE_SCHEMA DEFAULT CHARSET=utf8四:Stage 表
1,events_stages_current:记录了当前线程所处的执行阶段
2,events_stages_history:记录了当前线程所处的执行阶段10条历史记录
3,events_stages_history_long:记录了当前线程所处的执行阶段10000条历史记录
表结构定义如下:
01.CREATE
TABLE `events_stages_current` (02.`THREAD_ID`
bigint(20)
unsigned NOT NULL COMMENT ‘线程ID‘,03.`EVENT_ID`
bigint(20)
unsigned NOT NULL COMMENT ‘事件ID‘,04.`END_EVENT_ID`
bigint(20)
unsigned DEFAULT NULL COMMENT ‘结束事件ID‘,05.`EVENT_NAME`
varchar(128)
NOT NULL COMMENT ‘事件名称‘,06.`SOURCE`
varchar(64)
DEFAULT NULL COMMENT ‘源码位置‘,07.`TIMER_START`
bigint(20)
unsigned DEFAULT NULL COMMENT ‘事件开始时间(皮秒)‘,08.`TIMER_END`
bigint(20)
unsigned DEFAULT NULL COMMENT ‘事件结束结束时间(皮秒)‘,09.`TIMER_WAIT`
bigint(20)
unsigned DEFAULT NULL COMMENT ‘事件等待时间(皮秒)‘,10.`NESTING_EVENT_ID`
bigint(20)
unsigned DEFAULT NULL COMMENT ‘该事件对应的父事件ID‘,11.`NESTING_EVENT_TYPE` enum(‘STATEMENT‘,‘STAGE‘,‘WAIT‘)
DEFAULT NULL COMMENT ‘父事件类型(STATEMENT,
STAGE, WAIT)‘12.)
ENGINE=PERFORMANCE_SCHEMA DEFAULT CHARSET=utf8五:Statement 表
1,events_statements_current:通过 thread_id+event_id可以唯一确定一条记录。Statments表只记录最顶层的请求,SQL语句或是COMMAND,每条语句一行。event_name形式为statement/sql/*,或statement/com/*
2,events_statements_history
3,events_statements_history_long
表结构定义如下:
01.CREATE
TABLE `events_statements_current` (02.`THREAD_ID`
bigint(20)
unsigned NOT NULL COMMENT ‘线程ID‘,03.`EVENT_ID`
bigint(20)
unsigned NOT NULL COMMENT ‘事件ID‘,04.`END_EVENT_ID`
bigint(20)
unsigned DEFAULT NULL COMMENT ‘结束事件ID‘,05.`EVENT_NAME`
varchar(128)
NOT NULL COMMENT ‘事件名称‘,06.`SOURCE`
varchar(64)
DEFAULT NULL COMMENT ‘源码位置‘,07.`TIMER_START`
bigint(20)
unsigned DEFAULT NULL COMMENT ‘事件开始时间(皮秒)‘,08.`TIMER_END`
bigint(20)
unsigned DEFAULT NULL COMMENT ‘事件结束结束时间(皮秒)‘,09.`TIMER_WAIT`
bigint(20)
unsigned DEFAULT NULL COMMENT ‘事件等待时间(皮秒)‘,10.`LOCK_TIME`
bigint(20)
unsigned NOT NULL COMMENT ‘锁时间‘,11.`SQL_TEXT`
longtext COMMENT ‘记录SQL语句‘,12.`DIGEST`
varchar(32)
DEFAULT NULL COMMENT ‘对SQL_TEXT做MD5产生的32位字符串‘,13.`DIGEST_TEXT`
longtext COMMENT ‘将语句中值部分用问号代替,用于SQL语句归类‘,14.`CURRENT_SCHEMA`
varchar(64)
DEFAULT NULL COMMENT ‘默认的数据库名‘,15.`OBJECT_TYPE`
varchar(64)
DEFAULT NULL COMMENT ‘保留字段‘,16.`OBJECT_SCHEMA`
varchar(64)
DEFAULT NULL COMMENT ‘保留字段‘,17.`OBJECT_NAME`
varchar(64)
DEFAULT NULL COMMENT ‘保留字段‘,18.`OBJECT_INSTANCE_BEGIN`
bigint(20)
unsigned DEFAULT NULL COMMENT ‘内存地址‘,19.`MYSQL_ERRNO` int(11)
DEFAULT NULL COMMENT ‘‘,20.`RETURNED_SQLSTATE`
varchar(5)
DEFAULT NULL COMMENT ‘‘,21.`MESSAGE_TEXT`
varchar(128)
DEFAULT NULL COMMENT ‘信息‘,22.`ERRORS`
bigint(20)
unsigned NOT NULL COMMENT ‘错误数目‘,23.`WARNINGS`
bigint(20)
unsigned NOT NULL COMMENT ‘警告数目‘,24.`ROWS_AFFECTED`
bigint(20)
unsigned NOT NULL COMMENT ‘影响的数目‘,25.`ROWS_SENT`
bigint(20)
unsigned NOT NULL COMMENT ‘返回的记录数‘,26.`ROWS_EXAMINED`
bigint(20)
unsigned NOT NULL COMMENT ‘读取扫描的记录数目‘,27.`CREATED_TMP_DISK_TABLES`
bigint(20)
unsigned NOT NULL COMMENT ‘创建磁盘临时表数目‘,28.`CREATED_TMP_TABLES`
bigint(20)
unsigned NOT NULL COMMENT ‘创建临时表数目‘,29.`SELECT_FULL_JOIN`
bigint(20)
unsigned NOT NULL COMMENT ‘join时,第一个表为全表扫描的数目‘,30.`SELECT_FULL_RANGE_JOIN`
bigint(20)
unsigned NOT NULL COMMENT ‘引用表采用range方式扫描的数目‘,31.`SELECT_RANGE`
bigint(20)
unsigned NOT NULL COMMENT ‘join时,第一个表采用range方式扫描的数目‘,32.`SELECT_RANGE_CHECK`
bigint(20)
unsigned NOT NULL COMMENT ‘‘,33.`SELECT_SCAN`
bigint(20)
unsigned NOT NULL COMMENT ‘join时,第一个表位全表扫描的数目‘,34.`SORT_MERGE_PASSES`
bigint(20)
unsigned NOT NULL COMMENT ‘‘,35.`SORT_RANGE`
bigint(20)
unsigned NOT NULL COMMENT ‘范围排序数目‘,36.`SORT_ROWS`
bigint(20)
unsigned NOT NULL COMMENT ‘排序的记录数目‘,37.`SORT_SCAN`
bigint(20)
unsigned NOT NULL COMMENT ‘全表排序数目‘,38.`NO_INDEX_USED`
bigint(20)
unsigned NOT NULL COMMENT ‘没有使用索引数目‘,39.`NO_GOOD_INDEX_USED`
bigint(20)
unsigned NOT NULL COMMENT ‘‘,40.`NESTING_EVENT_ID`
bigint(20)
unsigned DEFAULT NULL COMMENT ‘该事件对应的父事件ID‘,41.`NESTING_EVENT_TYPE` enum(‘STATEMENT‘,‘STAGE‘,‘WAIT‘)
DEFAULT NULL COMMENT ‘父事件类型(STATEMENT,
STAGE, WAIT)‘42.)
ENGINE=PERFORMANCE_SCHEMA DEFAULT CHARSET=utf8六:Connection 表
1,users:记录用户连接数信息
2,hosts:记录了主机连接数信息
3,accounts:记录了用户主机连接数信息

01.zjy@performance_schema 12:03:27>select
* from users;02.+------------------+---------------------+-------------------+03.|
USER | CURRENT_CONNECTIONS | TOTAL_CONNECTIONS |04.+------------------+---------------------+-------------------+05.|
debian-sys-maint | 0 | 36 |06.|
zjy | 1 | 22285 |07.|
dchat_php | 0 | 37864 |08.|
dxyslave | 2 | 9 |09.|
nagios | 0 | 10770 |10.|
dchat_data | 140 | 2233023 |11.|
NULL | 0 | 15866 |12.|
dchat_api | 160 | 2754212 |13.|
mha_data | 1 | 36 |14.|
backup | 0 | 15 |15.|
cacti | 0 | 4312 |16.|
kol | 10 | 172414 |17.+------------------+---------------------+-------------------+18.12 rows
in set (0.00 sec)19. 20.zjy@performance_schema 12:03:34>select
* from hosts;21.+-----------------+---------------------+-------------------+22.|
HOST | CURRENT_CONNECTIONS | TOTAL_CONNECTIONS |23.+-----------------+---------------------+-------------------+24.| 192.168.100.218 | 150 | 2499422 |25.| 192.168.100.240 | 10 | 172429 |26.| 192.168.100.139 | 0 | 698 |27.| 192.168.100.21 | 0 | 2 |28.| 192.168.100.220 | 150 | 2526136 |29.| 192.168.100.25 | 1 | 7 |30.|
NULL | 0 | 15867 |31.| 192.168.100.241 | 0 | 21558 |32.| 192.168.100.191 | 1 | 34 |33.|
localhost | 0 | 10807 |34.| 192.168.100.118 | 1 | 2 |35.| 192.168.100.251 | 0 | 4312 |36.| 192.168.100.23 | 1 | 31 |37.| 192.168.100.193 | 0 | 15 |38.+-----------------+---------------------+-------------------+39.14 rows
in set (0.01 sec)40. 41.zjy@performance_schema 12:05:21>select
* from accounts;42.+------------------+-----------------+---------------------+-------------------+43.|
USER | HOST | CURRENT_CONNECTIONS | TOTAL_CONNECTIONS |44.+------------------+-----------------+---------------------+-------------------+45.|
cacti | 192.168.100.251 | 0 | 4313 |46.|
debian-sys-maint | localhost | 0 | 36 |47.|
backup | 192.168.100.193 | 0 | 15 |48.|
dchat_api | 192.168.100.220 | 80 | 1382585 |49.|
dchat_php | 192.168.100.220 | 0 | 20292 |50.|
zjy | 192.168.100.139 | 0 | 698 |51.|
zjy | 192.168.100.241 | 0 | 21558 |52.|
mha_data | 192.168.100.191 | 1 | 34 |53.|
dxyslave | 192.168.100.118 | 1 | 2 |54.|
kol | 192.168.100.240 | 10 | 172431 |55.|
dxyslave | 192.168.100.25 | 1 | 7 |56.|
dchat_data | 192.168.100.218 | 70 | 1109974 |57.|
zjy | 192.168.100.23 | 1 | 31 |58.|
dchat_php | 192.168.100.218 | 0 | 17572 |59.|
dchat_data | 192.168.100.220 | 70 | 1123306 |60.|
NULL | NULL | 0 | 15868 |61.|
mha_data | 192.168.100.21 | 0 | 2 |62.|
dchat_api | 192.168.100.218 | 80 | 1371918 |63.|
nagios | localhost | 0 | 10771 |64.+------------------+-----------------+---------------------+-------------------+七:Summary 表: Summary表聚集了各个维度的统计信息包括表维度,索引维度,会话维度,语句维度和锁维度的统计信息
1,events_waits_summary_global_by_event_name:按等待事件类型聚合,每个事件一条记录
1.CREATE
TABLE `events_waits_summary_global_by_event_name` (2.`EVENT_NAME`
varchar(128)
NOT NULL COMMENT ‘事件名称‘,3.`COUNT_STAR`
bigint(20)
unsigned NOT NULL COMMENT ‘事件计数‘,4.`SUM_TIMER_WAIT`
bigint(20)
unsigned NOT NULL COMMENT ‘总的等待时间‘,5.`MIN_TIMER_WAIT`
bigint(20)
unsigned NOT NULL COMMENT ‘最小等待时间‘,6.`AVG_TIMER_WAIT`
bigint(20)
unsigned NOT NULL COMMENT ‘平均等待时间‘,7.`MAX_TIMER_WAIT`
bigint(20)
unsigned NOT NULL COMMENT ‘最大等待时间‘8.)
ENGINE=PERFORMANCE_SCHEMA DEFAULT CHARSET=utf8
2,events_waits_summary_by_instance:按等待事件对象聚合,同一种等待事件,可能有多个实例,每个实例有不同的内存地址,因此
event_name+object_instance_begin唯一确定一条记录。
01.CREATE
TABLE `events_waits_summary_by_instance` (02.`EVENT_NAME`
varchar(128)
NOT NULL COMMENT ‘事件名称‘,03.`OBJECT_INSTANCE_BEGIN`
bigint(20)
unsigned NOT NULL COMMENT ‘内存地址‘,04.`COUNT_STAR`
bigint(20)
unsigned NOT NULL COMMENT ‘事件计数‘,05.`SUM_TIMER_WAIT`
bigint(20)
unsigned NOT NULL COMMENT ‘总的等待时间‘,06.`MIN_TIMER_WAIT`
bigint(20)
unsigned NOT NULL COMMENT ‘最小等待时间‘,07.`AVG_TIMER_WAIT`
bigint(20)
unsigned NOT NULL COMMENT ‘平均等待时间‘,08.`MAX_TIMER_WAIT`
bigint(20)
unsigned NOT NULL COMMENT ‘最大等待时间‘09.)
ENGINE=PERFORMANCE_SCHEMA DEFAULT CHARSET=utf83,events_waits_summary_by_thread_by_event_name:按每个线程和事件来统计,thread_id+event_name唯一确定一条记录。
01.CREATE
TABLE `events_waits_summary_by_thread_by_event_name` (02.`THREAD_ID`
bigint(20)
unsigned NOT NULL COMMENT ‘线程ID‘,03.`EVENT_NAME`
varchar(128)
NOT NULL COMMENT ‘事件名称‘,04.`COUNT_STAR`
bigint(20)
unsigned NOT NULL COMMENT ‘事件计数‘,05.`SUM_TIMER_WAIT`
bigint(20)
unsigned NOT NULL COMMENT ‘总的等待时间‘,06.`MIN_TIMER_WAIT`
bigint(20)
unsigned NOT NULL COMMENT ‘最小等待时间‘,07.`AVG_TIMER_WAIT`
bigint(20)
unsigned NOT NULL COMMENT ‘平均等待时间‘,08.`MAX_TIMER_WAIT`
bigint(20)
unsigned NOT NULL COMMENT ‘最大等待时间‘09.)
ENGINE=PERFORMANCE_SCHEMA DEFAULT CHARSET=utf84,events_stages_summary_global_by_event_name:按事件阶段类型聚合,每个事件一条记录,表结构同上。
5,events_stages_summary_by_thread_by_event_name:按每个线程和事件来阶段统计,表结构同上。
6,events_statements_summary_by_digest:按照事件的语句进行聚合。
01.CREATE
TABLE `events_statements_summary_by_digest` (02.`SCHEMA_NAME`
varchar(64)
DEFAULT NULL COMMENT ‘库名‘,03.`DIGEST`
varchar(32)
DEFAULT NULL COMMENT ‘对SQL_TEXT做MD5产生的32位字符串。如果为consumer表中没有打开statement_digest选项,则为NULL‘,04.`DIGEST_TEXT`
longtext COMMENT ‘将语句中值部分用问号代替,用于SQL语句归类。如果为consumer表中没有打开statement_digest选项,则为NULL。‘,05.`COUNT_STAR`
bigint(20)
unsigned NOT NULL COMMENT ‘事件计数‘,06.`SUM_TIMER_WAIT`
bigint(20)
unsigned NOT NULL COMMENT ‘总的等待时间‘,07.`MIN_TIMER_WAIT`
bigint(20)
unsigned NOT NULL COMMENT ‘最小等待时间‘,08.`AVG_TIMER_WAIT`
bigint(20)
unsigned NOT NULL COMMENT ‘平均等待时间‘,09.`MAX_TIMER_WAIT`
bigint(20)
unsigned NOT NULL COMMENT ‘最大等待时间‘,10.`SUM_LOCK_TIME`
bigint(20)
unsigned NOT NULL COMMENT ‘锁时间总时长‘,11.`SUM_ERRORS`
bigint(20)
unsigned NOT NULL COMMENT ‘错误数的总‘,12.`SUM_WARNINGS`
bigint(20)
unsigned NOT NULL COMMENT ‘警告的总数‘,13.`SUM_ROWS_AFFECTED`
bigint(20)
unsigned NOT NULL COMMENT ‘影响的总数目‘,14.`SUM_ROWS_SENT`
bigint(20)
unsigned NOT NULL COMMENT ‘返回总数目‘,15.`SUM_ROWS_EXAMINED`
bigint(20)
unsigned NOT NULL COMMENT ‘总的扫描的数目‘,16.`SUM_CREATED_TMP_DISK_TABLES`
bigint(20)
unsigned NOT NULL COMMENT ‘创建磁盘临时表的总数目‘,17.`SUM_CREATED_TMP_TABLES`
bigint(20)
unsigned NOT NULL COMMENT ‘创建临时表的总数目‘,18.`SUM_SELECT_FULL_JOIN`
bigint(20)
unsigned NOT NULL COMMENT ‘第一个表全表扫描的总数目‘,19.`SUM_SELECT_FULL_RANGE_JOIN`
bigint(20)
unsigned NOT NULL COMMENT ‘总的采用range方式扫描的数目‘,20.`SUM_SELECT_RANGE`
bigint(20)
unsigned NOT NULL COMMENT ‘第一个表采用range方式扫描的总数目‘,21.`SUM_SELECT_RANGE_CHECK`
bigint(20)
unsigned NOT NULL COMMENT ‘‘,22.`SUM_SELECT_SCAN`
bigint(20)
unsigned NOT NULL COMMENT ‘第一个表位全表扫描的总数目‘,23.`SUM_SORT_MERGE_PASSES`
bigint(20)
unsigned NOT NULL COMMENT ‘‘,24.`SUM_SORT_RANGE`
bigint(20)
unsigned NOT NULL COMMENT ‘范围排序总数‘,25.`SUM_SORT_ROWS`
bigint(20)
unsigned NOT NULL COMMENT ‘排序的记录总数目‘,26.`SUM_SORT_SCAN`
bigint(20)
unsigned NOT NULL COMMENT ‘第一个表排序扫描总数目‘,27.`SUM_NO_INDEX_USED`
bigint(20)
unsigned NOT NULL COMMENT ‘没有使用索引总数‘,28.`SUM_NO_GOOD_INDEX_USED`
bigint(20)
unsigned NOT NULL COMMENT ‘‘,29.`FIRST_SEEN`
timestamp NOT NULL DEFAULT ‘0000-00-00
00:00:00‘ COMMENT ‘第一次执行时间‘,30.`LAST_SEEN`
timestamp NOT NULL DEFAULT ‘0000-00-00
00:00:00‘ COMMENT ‘最后一次执行时间‘31.)
ENGINE=PERFORMANCE_SCHEMA DEFAULT CHARSET=utf87,events_statements_summary_global_by_event_name:按照事件的语句进行聚合。表结构同上。
8,events_statements_summary_by_thread_by_event_name:按照线程和事件的语句进行聚合,表结构同上。
9,file_summary_by_instance:按事件类型统计(物理IO维度)
10,file_summary_by_event_name:具体文件统计(物理IO维度)
9和10一起说明:
统计IO操作:COUNT_STAR,SUM_TIMER_WAIT,MIN_TIMER_WAIT,AVG_TIMER_WAIT,MAX_TIMER_WAIT
统计读 :COUNT_READ,SUM_TIMER_READ,MIN_TIMER_READ,AVG_TIMER_READ,MAX_TIMER_READ, SUM_NUMBER_OF_BYTES_READ
统计写 :COUNT_WRITE,SUM_TIMER_WRITE,MIN_TIMER_WRITE,AVG_TIMER_WRITE,MAX_TIMER_WRITE, SUM_NUMBER_OF_BYTES_WRITE
统计其他IO事件,比如create,delete,open,close等:COUNT_MISC,SUM_TIMER_MISC,MIN_TIMER_MISC,AVG_TIMER_MISC,MAX_TIMER_MISC
11,table_io_waits_summary_by_table:根据wait/io/table/sql/handler,聚合每个表的I/O操作(逻辑IO纬度)
统计IO操作:COUNT_STAR,SUM_TIMER_WAIT,MIN_TIMER_WAIT,AVG_TIMER_WAIT,MAX_TIMER_WAIT
统计读 :COUNT_READ,SUM_TIMER_READ,MIN_TIMER_READ,AVG_TIMER_READ,MAX_TIMER_READ
:COUNT_FETCH,SUM_TIMER_FETCH,MIN_TIMER_FETCH,AVG_TIMER_FETCH, MAX_TIMER_FETCH
统计写 :COUNT_WRITE,SUM_TIMER_WRITE,MIN_TIMER_WRITE,AVG_TIMER_WRITE,MAX_TIMER_WRITE
INSERT统计,相应的还有DELETE和UPDATE统计:COUNT_INSERT,SUM_TIMER_INSERT,MIN_TIMER_INSERT,AVG_TIMER_INSERT,MAX_TIMER_INSERT
12,table_io_waits_summary_by_index_usage:与table_io_waits_summary_by_table类似,按索引维度统计
13,table_lock_waits_summary_by_table:聚合了表锁等待事件,包括internal lock 和 external lock
internal lock通过SQL层函数thr_lock调用,OPERATION值为:
read normal、read with shared locks、read high priority、read no insert、write allow write、write concurrent insert、write delayed、write low priority、write normal
external lock则通过接口函数handler::external_lock调用存储引擎层,OPERATION列的值为:read external、write external
14,Connection Summaries表:account、user、host
events_waits_summary_by_account_by_event_name
events_waits_summary_by_user_by_event_name
events_waits_summary_by_host_by_event_name
events_stages_summary_by_account_by_event_name
events_stages_summary_by_user_by_event_name
events_stages_summary_by_host_by_event_name
events_statements_summary_by_account_by_event_name
events_statements_summary_by_user_by_event_name
events_statements_summary_by_host_by_event_name
15,socket_summary_by_instance、socket_summary_by_event_name:socket聚合统计表。
八:其他相关表
1,performance_timers:系统支持的统计时间单位
2,threads:监视服务端的当前运行的线程
统计应用:
关于SQL维度的统计信息主要集中在events_statements_summary_by_digest表中,通过将SQL语句抽象出digest,可以统计某类SQL语句在各个维度的统计信息
1,哪个SQL执行最多:
01.zjy@performance_schema 11:36:22><strong>SELECT
SCHEMA_NAME,DIGEST_TEXT,COUNT_STAR,SUM_ROWS_SENT,SUM_ROWS_EXAMINED,FIRST_SEEN,LAST_SEEN FROM events_statements_summary_by_digest ORDER BY COUNT_STAR desc LIMIT 1G02.</strong>*************************** 1.
row ***************************<strong>03.SCHEMA_NAME</strong>:
dchat04.<strong>DIGEST_TEXT</strong>:
SELECT ...05.<strong>COUNT_STAR</strong>: 116121010206.SUM_ROWS_SENT: 116120784207.SUM_ROWS_EXAMINED: 0<strong>08.FIRST_SEEN</strong>: 2016-02-17 00:36:46<strong>09.LAST_SEEN</strong>: 2016-03-07 11:36:29各个字段的注释可以看上面的表结构说明:从2月17号到3月7号该SQL执行了1161210102次。
2,哪个SQL平均响应时间最多:
01.zjy@performance_schema 11:36:28><strong>SELECT
SCHEMA_NAME,DIGEST_TEXT,COUNT_STAR,AVG_TIMER_WAIT,SUM_ROWS_SENT,SUM_ROWS_EXAMINED,FIRST_SEEN,LAST_SEEN FROM events_statements_summary_by_digest ORDER BY AVG_TIMER_WAIT desc LIMIT 1G02.</strong>*************************** 1.
row ***************************<strong>03.SCHEMA_NAME</strong>:
dchat04.<strong>DIGEST_TEXT</strong>:
SELECT ...05.COUNT_STAR: 1<strong>06.AVG_TIMER_WAIT</strong>: 27323818396400007.SUM_ROWS_SENT: 5020808.SUM_ROWS_EXAMINED: 5565651<strong>09.FIRST_SEEN</strong>: 2016-02-22 13:27:33<strong>10.LAST_SEEN</strong>: 2016-02-22 13:27:33各个字段的注释可以看上面的表结构说明:从2月17号到3月7号该SQL平均响应时间273238183964000皮秒(1000000000000皮秒=1秒)
3,哪个SQL扫描的行数最多:
SUM_ROWS_EXAMINED
4,哪个SQL使用的临时表最多:
SUM_CREATED_TMP_DISK_TABLES、SUM_CREATED_TMP_TABLES
5,哪个SQL返回的结果集最多:
SUM_ROWS_SENT
6,哪个SQL排序数最多:
SUM_SORT_ROWS
通过上述指标我们可以间接获得某类SQL的逻辑IO(SUM_ROWS_EXAMINED),CPU消耗(SUM_SORT_ROWS),网络带宽(SUM_ROWS_SENT)的对比。
通过file_summary_by_instance表,可以获得系统运行到现在,哪个文件(表)物理IO最多,这可能意味着这个表经常需要访问磁盘IO。
7,哪个表、文件逻辑IO最多(热数据):
01.zjy@performance_schema 12:16:18><strong>SELECT
FILE_NAME,EVENT_NAME,COUNT_READ,SUM_NUMBER_OF_BYTES_READ,COUNT_WRITE,SUM_NUMBER_OF_BYTES_WRITE FROM file_summary_by_instance ORDER BY SUM_NUMBER_OF_BYTES_READ+SUM_NUMBER_OF_BYTES_WRITE DESC LIMIT 2G02.</strong>*************************** 1.
row ***************************03.FILE_NAME:
/var/lib/mysql/<strong>ibdata1 #文件</strong>04.EVENT_NAME:
wait/io/file/innodb/innodb_data_file05.COUNT_READ: 54406.SUM_NUMBER_OF_BYTES_READ: 1097728007.COUNT_WRITE: 370072908.SUM_NUMBER_OF_BYTES_WRITE: 143373421772809.*************************** 2.
row ***************************10.FILE_NAME:
/var/lib/mysql/dchat/<strong>fans.ibd #表</strong>11.EVENT_NAME:
wait/io/file/innodb/innodb_data_file12.COUNT_READ: 937068013.SUM_NUMBER_OF_BYTES_READ: 15352918835214.COUNT_WRITE: 6757637615.SUM_NUMBER_OF_BYTES_WRITE: 11078154321928,哪个索引使用最多:
1.zjy@performance_schema 12:18:42><strong>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 limit 1;2.</strong>+-------------+------------+-------------+--------------+--------------+--------------+3.|
OBJECT_NAME | INDEX_NAME | COUNT_FETCH | COUNT_INSERT | COUNT_UPDATE | COUNT_DELETE |4.+-------------+------------+-------------+--------------+--------------+--------------+5.|
<strong>fans</strong> | <strong>PRIMARY</strong> | 29002695158 | 0| 296373434 | 0 |6.+-------------+------------+-------------+--------------+--------------+--------------+7.1 row
in set (0.29 sec)通过table_io_waits_summary_by_index_usage表,可以获得系统运行到现在,哪个表的具体哪个索引(包括主键索引,二级索引)使用最多。
9,哪个索引没有使用过:
1.zjy@performance_schema 12:23:22><strong>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;</strong>10,哪个等待事件消耗的时间最多:
1.zjy@performance_schema 12:25:22><strong>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 LIMIT 1;</strong>11,类似profiling功能:
分析具体某条SQL,该SQL在执行各个阶段的时间消耗,通过events_statements_xxx表和events_stages_xxx表,就可以达到目的。两个表通过event_id与nesting_event_id关联,stages表的nesting_event_id为对应statements表的event_id;针对每个stage可能出现的锁等待,一个stage会对应一个或多个wait,通过stage_xxx表的event_id字段与waits_xxx表的nesting_event_id进行关联。如:

001.比如分析包含count(*)的某条SQL语句,具体如下:002. 003.SELECT004.EVENT_ID,005.sql_text006.FROM
events_statements_history007.WHERE
sql_text LIKE ‘%count(*)%‘;008.+----------+--------------------------------------+009.|
EVENT_ID | sql_text |010.+----------+--------------------------------------+011.| 1690 |
select count(*) from chuck.test_slow |012.+----------+--------------------------------------+013.首先得到了语句的event_id为1690,通过查找events_stages_xxx中nesting_event_id为1690的记录,可以达到目的。014. 015.a.查看每个阶段的时间消耗:016.SELECT017.event_id,018.EVENT_NAME,019.SOURCE,020.TIMER_END
- TIMER_START021.FROM
events_stages_history_long022.WHERE
NESTING_EVENT_ID = 1690;023.+----------+--------------------------------+----------------------+-----------------------+024.|
event_id | EVENT_NAME | SOURCE | TIMER_END-TIMER_START |025.+----------+--------------------------------+----------------------+-----------------------+026.| 1691 |
stage/sql/init | mysqld.cc:990 | 316945000 |027.| 1693 |
stage/sql/checking permissions | sql_parse.cc:5776 | 26774000 |028.| 1695 |
stage/sql/Opening tables | sql_base.cc:4970 | 41436934000 |029.| 2638 |
stage/sql/init | sql_select.cc:1050 | 85757000 |030.| 2639 |
stage/sql/System lock | lock.cc:303 | 40017000 |031.| 2643 |
stage/sql/optimizing | sql_optimizer.cc:138 | 38562000 |032.| 2644 |
stage/sql/statistics | sql_optimizer.cc:362 | 52845000 |033.| 2645 |
stage/sql/preparing | sql_optimizer.cc:485 | 53196000 |034.| 2646 |
stage/sql/executing | sql_executor.cc:112 | 3153000 |035.| 2647 |
stage/sql/Sending data | sql_executor.cc:192 | 7369072089000 |036.| 4304138 |
stage/sql/end | sql_select.cc:1105 | 19920000 |037.| 4304139 |
stage/sql/query end | sql_parse.cc:5463 | 44721000 |038.| 4304145 |
stage/sql/closing tables | sql_parse.cc:5524 | 61723000 |039.| 4304152 |
stage/sql/freeing items | sql_parse.cc:6838 | 455678000 |040.| 4304155 |
stage/sql/logging slow query | sql_parse.cc:2258 | 83348000 |041.| 4304159 |
stage/sql/cleaning up | sql_parse.cc:2163 | 4433000 |042.+----------+--------------------------------+----------------------+-----------------------+043.通过间接关联,我们能分析得到SQL语句在每个阶段的时间消耗,时间单位以皮秒表示。这里展示的结果很类似profiling功能,有了performance
schema,就不再需要profiling这个功能了。另外需要注意的是,由于默认情况下events_stages_history表中只为每个连接记录了最近10条记录,为了确保获取所有记录,需要访问events_stages_history_long表044. 045.b.查看某个阶段的锁等待情况046.针对每个stage可能出现的锁等待,一个stage会对应一个或多个wait,events_waits_history_long这个表容易爆满[默认阀值10000]。由于select
count(*)需要IO(逻辑IO或者物理IO),所以在stage/sql/Sending data阶段会有io等待的统计。通过stage_xxx表的event_id字段与waits_xxx表的nesting_event_id进行关联。047.SELECT048.event_id,049.event_name,050.source,051.timer_wait,052.object_name,053.index_name,054.operation,055.nesting_event_id056.FROM
events_waits_history_long057.WHERE
nesting_event_id = 2647;058.+----------+---------------------------+-----------------+------------+-------------+------------+-----------+------------------+059.|
event_id | event_name | source | timer_wait | object_name | index_name | operation | nesting_event_id |060.+----------+---------------------------+-----------------+------------+-------------+------------+-----------+------------------+061.| 190607 |
wait/io/table/sql/handler | handler.cc:2842 | 1845888 |
test_slow | idx_c1 | fetch | 2647 |062.| 190608 |
wait/io/table/sql/handler | handler.cc:2842 | 1955328 |
test_slow | idx_c1 | fetch | 2647 |063.| 190609 |
wait/io/table/sql/handler | handler.cc:2842 | 1929792 |
test_slow | idx_c1 | fetch | 2647 |064.| 190610 |
wait/io/table/sql/handler | handler.cc:2842 | 1869600 |
test_slow | idx_c1 | fetch | 2647 |065.| 190611 |
wait/io/table/sql/handler | handler.cc:2842 | 1922496 |
test_slow | idx_c1 | fetch | 2647 |066.+----------+---------------------------+-----------------+------------+-------------+------------+-----------+------------------+067.通过上面的实验,我们知道了statement,stage,wait的三级结构,通过nesting_event_id进行关联,它表示某个事件的父event_id。068. 069.(2).模拟innodb行锁等待的例子070.会话A执行语句update
test_icp set y=y+1 where
x=1(x为primary
key),不commit;会话B执行同样的语句update test_icp set y=y+1 where
x=1,会话B堵塞,并最终报错。通过连接连接查询events_statements_history_long和events_stages_history_long,可以看到在updating阶段花了大约60s的时间。这主要因为实例上的innodb_lock_wait_timeout设置为60,等待60s后超时报错了。071. 072.SELECT073.statement.EVENT_ID,074.stages.event_id,075.statement.sql_text,076.stages.event_name,077.stages.timer_wait078.FROM
events_statements_history_long statement079.join
events_stages_history_long stages080.on
statement.event_id=stages.nesting_event_id081.WHERE
statement.sql_text = ‘update
test_icp set y=y+1 where x=1‘;082.+----------+----------+-------------------------------------+--------------------------------+----------------+083.|
EVENT_ID | event_id | sql_text | event_name | timer_wait |084.+----------+----------+-------------------------------------+--------------------------------+----------------+085.| 5816 | 5817 |
update test_icp set y=y+1 where
x=1 |
stage/sql/init | 195543000 |086.| 5816 | 5819 |
update test_icp set y=y+1 where
x=1 |
stage/sql/checking permissions |22730000 |087.| 5816 | 5821 |
update test_icp set y=y+1 where
x=1 |
stage/sql/Opening tables | 66079000 |088.| 5816 | 5827 |
update test_icp set y=y+1 where
x=1 |
stage/sql/init | 89116000 |089.| 5816 | 5828 |
update test_icp set y=y+1 where
x=1 |
stage/sql/System lock | 218744000 |090.| 5816 | 5832 |
update test_icp set y=y+1 where
x=1 |
stage/sql/updating | 6001362045000 |091.| 5816 | 5968 |
update test_icp set y=y+1 where
x=1 |
stage/sql/end | 10435000 |092.| 5816 | 5969 |
update test_icp set y=y+1 where
x=1 |
stage/sql/query end | 85979000 |093.| 5816 | 5983 |
update test_icp set y=y+1 where
x=1 |
stage/sql/closing tables | 56562000 |094.| 5816 | 5990 |
update test_icp set y=y+1 where
x=1 |
stage/sql/freeing items | 83563000 |095.| 5816 | 5992 |
update test_icp set y=y+1 where
x=1 |
stage/sql/cleaning up | 4589000 |096.+----------+----------+-------------------------------------+--------------------------------+----------------+097.查看wait事件:098.SELECT099.event_id,100.event_name,101.source,102.timer_wait,103.object_name,104.index_name,105.operation,106.nesting_event_id107.FROM
events_waits_history_long108.WHERE
nesting_event_id = 5832;109.*************************** 1.
row ***************************110.event_id: 5832111.event_name:
wait/io/table/sql/handler112.source:
handler.cc:2782113.timer_wait: 6005946156624114.object_name:
test_icp115.index_name:
PRIMARY116.operation:
fetch117.从结果来看,waits表中记录了一个fetch等待事件,但并没有更细的innodb行锁等待事件统计。118. 119.(3).模拟MDL锁等待的例子120.会话A执行一个大查询select
count(*) from test_slow,会话B执行表结构变更alter table test_slow modify c2 varchar(152);通过如下语句可以得到alter语句的执行过程,重点关注“stage/sql/Waiting for table
metadata lock”阶段。121. 122.SELECT123.statement.EVENT_ID,124.stages.event_id,125.statement.sql_text,126.stages.event_name
as stage_name,127.stages.timer_wait
as stage_time128.FROM
events_statements_history_long statement129.left
join events_stages_history_long stages130.on
statement.event_id=stages.nesting_event_id131.WHERE
statement.sql_text = ‘alter
table test_slow modify c2 varchar(152)‘;132.+-----------+-----------+----------------------------------------------+----------------------------------------------------+---------------+133.|
EVENT_ID | event_id | sql_text | stage_name | stage_time |134.+-----------+-----------+----------------------------------------------+----------------------------------------------------+---------------+135.| 326526744 | 326526745 |
alter table test_slow modify c2 varchar(152)
| stage/sql/init |216662000 |136.| 326526744 | 326526747 |
alter table test_slow modify c2 varchar(152)
| stage/sql/checking permissions | 18183000 |137.| 326526744 | 326526748 |
alter table test_slow modify c2 varchar(152)
| stage/sql/checking permissions | 10294000 |138.| 326526744 | 326526750 |
alter table test_slow modify c2 varchar(152)
| stage/sql/init |4783000 |139.| 326526744 | 326526751 |
alter table test_slow modify c2 varchar(152)
| stage/sql/Opening tables | 140172000 |140.| 326526744 | 326526760 |
alter table test_slow modify c2 varchar(152)
| stage/sql/setup |157643000 |141.| 326526744 | 326526769 |
alter table test_slow modify c2 varchar(152)
| stage/sql/creating table | 8723217000 |142.| 326526744 | 326526803 |
alter table test_slow modify c2 varchar(152)
| stage/sql/After create | 257332000 |143.| 326526744 | 326526832 |
alter table test_slow modify c2 varchar(152)
| stage/sql/Waitingfor table
metadata lock | 1000181831000 |144.| 326526744 | 326526835 |
alter table test_slow modify c2 varchar(152)
| stage/sql/After create | 33483000 |145.| 326526744 | 326526838 |
alter table test_slow modify c2 varchar(152)
| stage/sql/Waitingfor table
metadata lock | 1000091810000 |146.| 326526744 | 326526841 |
alter table test_slow modify c2 varchar(152)
| stage/sql/After create | 17187000 |147.| 326526744 | 326526844 |
alter table test_slow modify c2 varchar(152)
| stage/sql/Waitingfor table
metadata lock | 1000126464000 |148.| 326526744 | 326526847 |
alter table test_slow modify c2 varchar(152)
| stage/sql/After create | 27472000 |149.| 326526744 | 326526850 |
alter table test_slow modify c2 varchar(152)
| stage/sql/Waitingfor table
metadata lock | 561996133000 |150.| 326526744 | 326526853 |
alter table test_slow modify c2 varchar(152)
| stage/sql/After create | 124876000 |151.| 326526744 | 326526877 |
alter table test_slow modify c2 varchar(152)
| stage/sql/System lock | 30659000 |152.| 326526744 | 326526881 |
alter table test_slow modify c2 varchar(152)
| stage/sql/preparingfor alter
table | 40246000 |153.| 326526744 | 326526889 |
alter table test_slow modify c2 varchar(152)
| stage/sql/altering table | 36628000 |154.| 326526744 | 326528280 |
alter table test_slow modify c2 varchar(152)
| stage/sql/end |43824000 |155.| 326526744 | 326528281 |
alter table test_slow modify c2 varchar(152)
| stage/sql/query end | 112557000 |156.| 326526744 | 326528299 |
alter table test_slow modify c2 varchar(152)
| stage/sql/closing tables | 27707000 |157.| 326526744 | 326528305 |
alter table test_slow modify c2 varchar(152)
| stage/sql/freeing items | 201614000 |158.| 326526744 | 326528308 |
alter table test_slow modify c2 varchar(152)
| stage/sql/cleaning up | 3584000 |159.+-----------+-----------+----------------------------------------------+----------------------------------------------------+---------------+160.从结果可以看到,出现了多次stage/sql/Waiting for table
metadata lock阶段,并且间隔1s,说明每隔1s钟会重试判断。找一个该阶段的event_id,通过nesting_event_id关联,确定到底在等待哪个wait事件。161.SELECT162.event_id,163.event_name,164.source,165.timer_wait,166.object_name,167.index_name,168.operation,169.nesting_event_id170.FROM
events_waits_history_long171.WHERE
nesting_event_id = 326526850;172.+-----------+---------------------------------------------------+------------------+--------------+-------------+------------+------------+------------------+173.|
event_id | event_name | source | timer_wait | object_name | index_name | operation | nesting_event_id |174.+-----------+---------------------------------------------------+------------------+--------------+-------------+------------+------------+------------------+175.| 326526851 |
wait/synch/cond/sql/MDL_context::COND_wait_status | mdl.cc:1327 | 562417991328|
NULL | NULL | timed_wait | 326526850 |176.| 326526852 |
wait/synch/mutex/mysys/my_thread_var::mutex | sql_class.h:3481 | 733248 |
NULL | NULL | lock | 326526850 |177.+-----------+---------------------------------------------------+------------------+--------------+-------------+------------+------------+------------------+178.通过结果可以知道,产生阻塞的是条件变量MDL_context::COND_wait_status,并且显示了代码的位置。总结:
本文通过对Performance Schema数据库的介绍,主要用于收集数据库服务器性能参数:①提供进程等待的详细信息,包括锁、互斥变量、文件信息;②保存历史的事件汇总信息,为提供MySQL服务器性能做出详细的判断;③对于新增和删除监控事件点都非常容易,并可以改变mysql服务器的监控周期,例如(CYCLE、MICROSECOND)。通过该库得到数据库运行的统计信息,更好分析定位问题和完善监控信息。类似的监控还有:
1.打开标准的innodb监控:2.CREATE
TABLE innodb_monitor (a INT) ENGINE=INNODB;3.打开innodb的锁监控:4.CREATE
TABLE innodb_lock_monitor (a INT) ENGINE=INNODB;5.打开innodb表空间监控:6.CREATE
TABLE innodb_tablespace_monitor (a INT) ENGINE=INNODB;7.打开innodb表监控:8.CREATE
TABLE innodb_table_monitor (a INT) ENGINE=INNODB;MySQL5.6 PERFORMANCE_SCHEMA 说明
原文:http://blog.csdn.net/isoleo/article/details/51180593