加入收藏 | 设为首页 | 会员中心 | 我要投稿 大连站长网 （https://www.0411zz.cn/）- 科技、建站、经验、云计算、5G、大数据,站长网!
当前位置： 首页 > 站长学院 > MySql教程 > 正文

怎么领略MySQL5.6中的PERFORMANCE_SCHEM

发布时间：2021-12-23 10:41:41 所属栏目：MySql教程 来源：互联网
导读：本篇内容介绍了怎么理解MySQL5.6中的PERFORMANCE_SCHEM的有关知识，在实际案例的操作过程中，不少人都会遇到这样的困境，接下来就让小编带领大家学习一下如何处理这些情况吧！希望大家仔细阅读，能够学有所成！ 通过sql语句找到在经历什么等待事件! Stateme
本篇内容介绍了“怎么理解MySQL5.6中的PERFORMANCE_SCHEM”的有关知识，在实际案例的操作过程中，不少人都会遇到这样的困境，接下来就让小编带领大家学习一下如何处理这些情况吧！希望大家仔细阅读，能够学有所成！
 
通过sql语句找到在经历什么等待事件!
 
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。
 
比如分析包含count(*)的某条SQL语句，具体如下：(类似于oracle的v$sql, v$sqlstat, v$sqlarea)
 
SELECT
 
EVENT_ID,
 
sql_text
 
FROM events_statements_history
 
WHERE sql_text LIKE '%count(*)%';
 
+----------+--------------------------------------+
 
| EVENT_ID | sql_text |
 
+----------+--------------------------------------+
 
| 1690 | select count(*) from chuck.test_slow |
 
+----------+--------------------------------------+
 
a.查看每个阶段的时间消耗：(类似于oracle的时间模型V$SYS_TIME_MODEL  V$SESS_TIME_MODEL)
 
SELECT
 
event_id,
 
EVENT_NAME,
 
SOURCE,
 
TIMER_END - TIMER_START
 
FROM events_stages_history_long
 
WHERE NESTING_EVENT_ID = 1690;
 
+----------+--------------------------------+----------------------+-----------------------+
 
| event_id | EVENT_NAME | SOURCE | TIMER_END-TIMER_START |
 
+----------+--------------------------------+----------------------+-----------------------+
 
……
 
| 2647 | stage/sql/Sending data | sql_executor.cc:192 | 7369072089000 |
 
b.查看某个阶段的锁等待情况 (类似于oracle的v$session_wait)
 
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。
 
SELECT
 
event_id,
 
event_name,
 
source,
 
timer_wait,
 
object_name,
 
index_name,
 
operation,
 
nesting_event_id
 
FROM events_waits_history_long
 
WHERE nesting_event_id = 2647;
 
+----------+---------------------------+-----------------+------------+-------------+------------+-----------+------------------+
 
| event_id | event_name | source | timer_wait | object_name | index_name | operation | nesting_event_id |
 
+----------+---------------------------+-----------------+------------+-------------+------------+-----------+------------------+
 
| 190607 | wait/io/table/sql/handler | handler.cc:2842 | 1845888 | test_slow | idx_c1 | fetch | 2647 |
 
 
 
https://www.cnblogs.com/zhoujinyi/p/5236705.html
 
MySQL5.6 PERFORMANCE_SCHEMA 说明
 
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：
 
[mysqld]
 
performance_schema=ON
 
查看是否开启：
 
mysql>show variables like 'performance_schema';
 
+--------------------+-------+
 
| Variable_name      | Value |
 
+--------------------+-------+
 
| performance_schema | ON    |
 
+--------------------+-------+
 
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：
 
zjy@performance_schema 10:16:56>show tables like '%setup%';
 
+----------------------------------------+
 
| Tables_in_performance_schema (%setup%) |
 
+----------------------------------------+
 
| setup_actors                           |
 
| setup_consumers                        |
 
| setup_instruments                      |
 
| setup_objects                          |
 
| setup_timers                           |
 
+----------------------------------------+
 
1，setup_actors：配置用户纬度的监控，默认监控所有用户。
 
zjy@performance_schema 10:19:11>select * from setup_actors;
 
+------+------+------+
 
| HOST | USER | ROLE |
 
+------+------+------+
 
| %    | %    | %    |
 
+------+------+------+
 
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。
 
zjy@: performance_schema 10:23:35>select * from setup_consumers;
 
+--------------------------------+---------+
 
| NAME                           | ENABLED |
 
+--------------------------------+---------+
 
| events_stages_current          | NO      |
 
| events_stages_history          | NO      |
 
| events_stages_history_long     | NO      |
 
| events_statements_current      | YES     |
 
| events_statements_history      | NO      |
 
| events_statements_history_long | NO      |
 
| events_waits_current           | NO      |
 
| events_waits_history           | NO      |
 
| events_waits_history_long      | NO      |
 
| global_instrumentation         | YES     |
 
| thread_instrumentation         | YES     |
 
| statements_digest              | YES     |
 
+--------------------------------+---------+
 
这里需要说明的是需要查看哪个就更新其ENABLED列为YES。如：
 
zjy@performance_schema 10:25:02>update setup_consumers set ENABLED='YES' where NAME in ('events_stages_current','events_waits_current');
 
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：
 
[mysqld]
 
#performance_schema
 
performance_schema_consumer_events_waits_current=on
 
performance_schema_consumer_events_stages_current=on
 
performance_schema_consumer_events_statements_current=on
 
performance_schema_consumer_events_waits_history=on
 
performance_schema_consumer_events_stages_history=on
 
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：
 
global_instrumentation > thread_instrumentation = statements_digest > events_stages_current = events_statements_current = events_waits_current > events_stages_history = events_statements_history = events_waits_history > events_stages_history_long = 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：
 
zjy@performance_schema 11:10:03>show variables like 'performance_schema%history%size';
 
+--------------------------------------------------------+-------+
 
| Variable_name                                          | Value |
 
+--------------------------------------------------------+-------+
 
| performance_schema_events_stages_history_long_size     | 10000 |
 
| performance_schema_events_stages_history_size          | 10    |
 
| performance_schema_events_statements_history_long_size | 10000 |
 
| performance_schema_events_statements_history_size      | 10    |
 
| performance_schema_events_waits_history_long_size      | 10000 |
 
| performance_schema_events_waits_history_size           | 10    |
 
+--------------------------------------------------------+-------+
 
3，setup_instruments：配置具体的instrument，主要包含4大类：idle、stage/xxx、statement/xxx、wait/xxx：
 
zjy@performance_schema 10:56:35>select name,count(*) from setup_instruments group by LEFT(name,5);
 
+---------------------------------+----------+
 
| name                            | count(*) |
 
+---------------------------------+----------+
 
| idle                            |        1 |
 
| stage/sql/After create          |      111 |
 
| statement/sql/select            |      179 |
 
| wait/synch/mutex/sql/PAGE::lock |      296 |
 
+---------------------------------+----------+
 
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。
 
zjy@performance_schema 11:00:18>select * from setup_objects;
 
+-------------+--------------------+-------------+---------+-------+
 
| OBJECT_TYPE | OBJECT_SCHEMA      | OBJECT_NAME | ENABLED | TIMED |
 
+-------------+--------------------+-------------+---------+-------+
 
| TABLE       | mysql              | %           | NO      | NO    |
 
| TABLE       | performance_schema | %           | NO      | NO    |
 
| TABLE       | information_schema | %           | NO      | NO    |
 
| TABLE       | %                  | %           | YES     | YES   |
 
+-------------+--------------------+-------------+---------+-------+
 
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）
 
zjy@performance_schema 11:05:12>select * from setup_timers;
 
+-----------+-------------+
 
| NAME      | TIMER_NAME  |
 
+-----------+-------------+
 
| idle      | MICROSECOND |
 
| wait      | CYCLE       |
 
| stage     | NANOSECOND  |
 
| statement | NANOSECOND  |
 
+-----------+-------------+
 
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。
 
zjy@performance_schema 11:20:04>select * from file_instances limit 2,5;
 
+---------------------------------+--------------------------------------+------------+
 
| FILE_NAME                       | EVENT_NAME                           | OPEN_COUNT |
 
+---------------------------------+--------------------------------------+------------+
 
| /var/lib/mysql/mysql/plugin.frm | wait/io/file/sql/FRM                 |          |
 
| /var/lib/mysql/mysql/plugin.MYI | wait/io/file/myisam/kfile            |          1 |
 
| /var/lib/mysql/mysql/plugin.MYD | wait/io/file/myisam/dfile            |          1 |
 
| /var/lib/mysql/ibdata1          | wait/io/file/innodb/innodb_data_file |          2 |
 
| /var/lib/mysql/ib_logfile0      | wait/io/file/innodb/innodb_log_file  |          2 |
 
+---------------------------------+--------------------------------------+------------+
 
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：
 
CREATE TABLE `events_waits_current` (
 
  `THREAD_ID` bigint(20) unsigned NOT NULL COMMENT '线程ID',
 
  `EVENT_ID` bigint(20) unsigned NOT NULL COMMENT '当前线程的事件ID，和THREAD_ID确定唯一',
 
  `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',
 
  `EVENT_NAME` varchar(128) NOT NULL COMMENT '事件名称',
 
  `SOURCE` varchar(64) DEFAULT NULL COMMENT '该事件产生时的源码文件',
 
  `TIMER_START` bigint(20) unsigned DEFAULT NULL COMMENT '事件开始时间（皮秒）',
 
  `TIMER_END` bigint(20) unsigned DEFAULT NULL COMMENT '事件结束结束时间（皮秒）',
 
  `TIMER_WAIT` bigint(20) unsigned DEFAULT NULL COMMENT '事件等待时间（皮秒）',
 
  `SPINS` int(10) unsigned DEFAULT NULL COMMENT '',
 
  `OBJECT_SCHEMA` varchar(64) DEFAULT NULL COMMENT '库名',
 
  `OBJECT_NAME` varchar(512) DEFAULT NULL COMMENT '文件名、表名、IP:SOCK值',
 
  `OBJECT_TYPE` varchar(64) DEFAULT NULL COMMENT 'FILE、TABLE、TEMPORARY TABLE',
 
  `INDEX_NAME` varchar(64) DEFAULT NULL COMMENT '索引名',
 
  `OBJECT_INSTANCE_BEGIN` bigint(20) unsigned NOT NULL COMMENT '内存地址',
 
  `NESTING_EVENT_ID` bigint(20) unsigned DEFAULT NULL COMMENT '该事件对应的父事件ID',
 
  `NESTING_EVENT_TYPE` enum('STATEMENT','STAGE','WAIT') DEFAULT NULL COMMENT '父事件类型(STATEMENT, STAGE, WAIT)',
 
  `OPERATION` varchar(32) NOT NULL COMMENT '操作类型（lock, read, write）',
 
  `NUMBER_OF_BYTES` bigint(20) DEFAULT NULL COMMENT '',
 
  `FLAGS` int(10) unsigned DEFAULT NULL COMMENT '标记'
 
) 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：
 
CREATE TABLE `events_stages_current` (
 
  `THREAD_ID` bigint(20) unsigned NOT NULL COMMENT '线程ID',
 
  `EVENT_ID` bigint(20) unsigned NOT NULL COMMENT '事件ID',
 
  `END_EVENT_ID` bigint(20) unsigned DEFAULT NULL COMMENT '结束事件ID',
 
  `EVENT_NAME` varchar(128) NOT NULL COMMENT '事件名称',
 
  `SOURCE` varchar(64) DEFAULT NULL COMMENT '源码位置',
 
  `TIMER_START` bigint(20) unsigned DEFAULT NULL COMMENT '事件开始时间（皮秒）',
 
  `TIMER_END` bigint(20) unsigned DEFAULT NULL COMMENT '事件结束结束时间（皮秒）',
 
  `TIMER_WAIT` bigint(20) unsigned DEFAULT NULL COMMENT '事件等待时间（皮秒）',
 
  `NESTING_EVENT_ID` bigint(20) unsigned DEFAULT NULL COMMENT '该事件对应的父事件ID',
 
  `NESTING_EVENT_TYPE` enum('STATEMENT','STAGE','WAIT') DEFAULT NULL COMMENT '父事件类型(STATEMENT, STAGE, WAIT)'
 
) 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：
 
CREATE TABLE `events_statements_current` (
 
  `THREAD_ID` bigint(20) unsigned NOT NULL COMMENT '线程ID',
 
  `EVENT_ID` bigint(20) unsigned NOT NULL COMMENT '事件ID',
 
  `END_EVENT_ID` bigint(20) unsigned DEFAULT NULL COMMENT '结束事件ID',
 
  `EVENT_NAME` varchar(128) NOT NULL COMMENT '事件名称',
 
  `SOURCE` varchar(64) DEFAULT NULL COMMENT '源码位置',
 
  `TIMER_START` bigint(20) unsigned DEFAULT NULL COMMENT '事件开始时间（皮秒）',
 
  `TIMER_END` bigint(20) unsigned DEFAULT NULL COMMENT '事件结束结束时间（皮秒）',
 
  `TIMER_WAIT` bigint(20) unsigned DEFAULT NULL COMMENT '事件等待时间（皮秒）',
 
  `LOCK_TIME` bigint(20) unsigned NOT NULL COMMENT '锁时间',
 
  `SQL_TEXT` longtext COMMENT '记录SQL语句',
 
  `DIGEST` varchar(32) DEFAULT NULL COMMENT '对SQL_TEXT做MD5产生的32位字符串',
 
  `DIGEST_TEXT` longtext COMMENT '将语句中值部分用问号代替，用于SQL语句归类',
 
  `CURRENT_SCHEMA` varchar(64) DEFAULT NULL COMMENT '默认的数据库名',
 
  `OBJECT_TYPE` varchar(64) DEFAULT NULL COMMENT '保留字段',
 
  `OBJECT_SCHEMA` varchar(64) DEFAULT NULL COMMENT '保留字段',
 
  `OBJECT_NAME` varchar(64) DEFAULT NULL COMMENT '保留字段',
 
  `OBJECT_INSTANCE_BEGIN` bigint(20) unsigned DEFAULT NULL COMMENT '内存地址',
 
  `MYSQL_ERRNO` int(11) DEFAULT NULL COMMENT '',
 
  `RETURNED_SQLSTATE` varchar(5) DEFAULT NULL COMMENT '',
 
  `MESSAGE_TEXT` varchar(128) DEFAULT NULL COMMENT '信息',
 
  `ERRORS` bigint(20) unsigned NOT NULL COMMENT '错误数目',
 
  `WARNINGS` bigint(20) unsigned NOT NULL COMMENT '警告数目',
 
  `ROWS_AFFECTED` bigint(20) unsigned NOT NULL COMMENT '影响的数目',
 
  `ROWS_SENT` bigint(20) unsigned NOT NULL COMMENT '返回的记录数',
 
  `ROWS_EXAMINED` bigint(20) unsigned NOT NULL COMMENT '读取扫描的记录数目',
 
  `CREATED_TMP_DISK_TABLES` bigint(20) unsigned NOT NULL COMMENT '创建磁盘临时表数目',
 
  `CREATED_TMP_TABLES` bigint(20) unsigned NOT NULL COMMENT '创建临时表数目',
 
  `SELECT_FULL_JOIN` bigint(20) unsigned NOT NULL COMMENT 'join时，第一个表为全表扫描的数目',
 
  `SELECT_FULL_RANGE_JOIN` bigint(20) unsigned NOT NULL COMMENT '引用表采用range方式扫描的数目',
 
  `SELECT_RANGE` bigint(20) unsigned NOT NULL COMMENT 'join时，第一个表采用range方式扫描的数目',
 
  `SELECT_RANGE_CHECK` bigint(20) unsigned NOT NULL COMMENT '',
 
  `SELECT_SCAN` bigint(20) unsigned NOT NULL COMMENT 'join时，第一个表位全表扫描的数目',
 
  `SORT_MERGE_PASSES` bigint(20) unsigned NOT NULL COMMENT '',
 
  `SORT_RANGE` bigint(20) unsigned NOT NULL COMMENT '范围排序数目',
 
  `SORT_ROWS` bigint(20) unsigned NOT NULL COMMENT '排序的记录数目',
 
  `SORT_SCAN` bigint(20) unsigned NOT NULL COMMENT '全表排序数目',
 
  `NO_INDEX_USED` bigint(20) unsigned NOT NULL COMMENT '没有使用索引数目',
 
  `NO_GOOD_INDEX_USED` bigint(20) unsigned NOT NULL COMMENT '',
 
  `NESTING_EVENT_ID` bigint(20) unsigned DEFAULT NULL COMMENT '该事件对应的父事件ID',
 
  `NESTING_EVENT_TYPE` enum('STATEMENT','STAGE','WAIT') DEFAULT NULL COMMENT '父事件类型(STATEMENT, STAGE, WAIT)'
 
) 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
 
zjy@performance_schema 12:03:27>select * from users;
 
+------------------+---------------------+-------------------+
 
| USER             | CURRENT_CONNECTIONS | TOTAL_CONNECTIONS |
 
+------------------+---------------------+-------------------+
 
| debian-sys-maint |                   |                36 |
 
| zjy              |                   1 |             22285 |
 
| dchat_php        |                   |             37864 |
 
| dxyslave         |                   2 |                 9 |
 
| nagios           |                   |             10770 |
 
| dchat_data       |                 140 |           2233023 |
 
| NULL             |                   |             15866 |
 
| dchat_api        |                 160 |           2754212 |
 
| mha_data         |                   1 |                36 |
 
| backup           |                   |                15 |
 
| cacti            |                   |              4312 |
 
| kol              |                  10 |            172414 |
 
+------------------+---------------------+-------------------+
 
12 rows in set (0.00 sec)
 
zjy@performance_schema 12:03:34>select * from hosts;
 
+-----------------+---------------------+-------------------+
 
| HOST            | CURRENT_CONNECTIONS | TOTAL_CONNECTIONS |
 
+-----------------+---------------------+-------------------+
 
| 192.168.100.218 |                 150 |           2499422 |
 
| 192.168.100.240 |                  10 |            172429 |
 
| 192.168.100.139 |                    |               698 |
 
| 192.168.100.21  |                   |                 2 |
 
| 192.168.100.220 |                 150 |           2526136 |
 
| 192.168.100.25  |                   1 |                 7 |
 
| NULL            |                    |             15867 |
 
| 192.168.100.241 |                   |             21558 |
 
| 192.168.100.191 |                   1 |                34 |
 
| localhost       |                   |             10807 |
 
| 192.168.100.118 |                   1 |                 2 |
 
| 192.168.100.251 |                   |              4312 |
 
| 192.168.100.23  |                   1 |                31 |
 
| 192.168.100.193 |                   |                15 |
 
+-----------------+---------------------+-------------------+
 
14 rows in set (0.01 sec)
 
zjy@performance_schema 12:05:21>select * from accounts;
 
+------------------+-----------------+---------------------+-------------------+
 
| USER             | HOST            | CURRENT_CONNECTIONS | TOTAL_CONNECTIONS |
 
+------------------+-----------------+---------------------+-------------------+
 
| cacti            | 192.168.100.251 |                   |              4313 |
 
| debian-sys-maint | localhost       |                   |                36 |
 
| backup           | 192.168.100.193 |                   |                15 |
 
| dchat_api        | 192.168.100.220 |                  80 |           1382585 |
 
| dchat_php        | 192.168.100.220 |                   |             20292 |
 
| zjy              | 192.168.100.139 |                   |               698 |
 
| zjy              | 192.168.100.241 |                   |             21558 |
 
| mha_data         | 192.168.100.191 |                   1 |                34 |
 
| dxyslave         | 192.168.100.118 |                   1 |                 2 |
 
| kol              | 192.168.100.240 |                  10 |            172431 |
 
| dxyslave         | 192.168.100.25  |                   1 |                 7 |
 
| dchat_data       | 192.168.100.218 |                  70 |           1109974 |
 
| zjy              | 192.168.100.23  |                   1 |                31 |
 
| dchat_php        | 192.168.100.218 |                   |             17572 |
 
| dchat_data       | 192.168.100.220 |                  70 |           1123306 |
 
| NULL             | NULL            |                   |             15868 |
 
| mha_data         | 192.168.100.21  |                   |                 2 |
 
| dchat_api        | 192.168.100.218 |                  80 |           1371918 |
 
| nagios           | localhost       |                   |             10771 |
 
+------------------+-----------------+---------------------+-------------------+
 
View Code
 
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
 
CREATE TABLE `events_waits_summary_global_by_event_name` (
 
  `EVENT_NAME` varchar(128) NOT NULL COMMENT '事件名称',
 
  `COUNT_STAR` bigint(20) unsigned NOT NULL COMMENT '事件计数',
 
  `SUM_TIMER_WAIT` bigint(20) unsigned NOT NULL COMMENT '总的等待时间',
 
  `MIN_TIMER_WAIT` bigint(20) unsigned NOT NULL COMMENT '最小等待时间',
 
  `AVG_TIMER_WAIT` bigint(20) unsigned NOT NULL COMMENT '平均等待时间',
 
  `MAX_TIMER_WAIT` bigint(20) unsigned NOT NULL COMMENT '最大等待时间'
 
) 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。
 
CREATE TABLE `events_waits_summary_by_instance` (
 
  `EVENT_NAME` varchar(128) NOT NULL COMMENT '事件名称',
 
  `OBJECT_INSTANCE_BEGIN` bigint(20) unsigned NOT NULL COMMENT '内存地址',
 
  `COUNT_STAR` bigint(20) unsigned NOT NULL COMMENT '事件计数',
 
  `SUM_TIMER_WAIT` bigint(20) unsigned NOT NULL COMMENT '总的等待时间',
 
  `MIN_TIMER_WAIT` bigint(20) unsigned NOT NULL COMMENT '最小等待时间',
 
  `AVG_TIMER_WAIT` bigint(20) unsigned NOT NULL COMMENT '平均等待时间',
 
  `MAX_TIMER_WAIT` bigint(20) unsigned NOT NULL COMMENT '最大等待时间'
 
) ENGINE=PERFORMANCE_SCHEMA DEFAULT CHARSET=utf8
 
3，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。
 
CREATE TABLE `events_waits_summary_by_thread_by_event_name` (
 
  `THREAD_ID` bigint(20) unsigned NOT NULL COMMENT '线程ID',
 
  `EVENT_NAME` varchar(128) NOT NULL COMMENT '事件名称',
 
  `COUNT_STAR` bigint(20) unsigned NOT NULL COMMENT '事件计数',
 
  `SUM_TIMER_WAIT` bigint(20) unsigned NOT NULL COMMENT '总的等待时间',
 
  `MIN_TIMER_WAIT` bigint(20) unsigned NOT NULL COMMENT '最小等待时间',
 
  `AVG_TIMER_WAIT` bigint(20) unsigned NOT NULL COMMENT '平均等待时间',
 
  `MAX_TIMER_WAIT` bigint(20) unsigned NOT NULL COMMENT '最大等待时间'
 
) ENGINE=PERFORMANCE_SCHEMA DEFAULT CHARSET=utf8
 
4，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。
 
CREATE TABLE `events_statements_summary_by_digest` (
 
  `SCHEMA_NAME` varchar(64) DEFAULT NULL COMMENT '库名',
 
  `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',
 
  `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。',
 
  `COUNT_STAR` bigint(20) unsigned NOT NULL COMMENT '事件计数',
 
  `SUM_TIMER_WAIT` bigint(20) unsigned NOT NULL COMMENT '总的等待时间',
 
  `MIN_TIMER_WAIT` bigint(20) unsigned NOT NULL COMMENT '最小等待时间',
 
  `AVG_TIMER_WAIT` bigint(20) unsigned NOT NULL COMMENT '平均等待时间',
 
  `MAX_TIMER_WAIT` bigint(20) unsigned NOT NULL COMMENT '最大等待时间',
 
  `SUM_LOCK_TIME` bigint(20) unsigned NOT NULL COMMENT '锁时间总时长',
 
  `SUM_ERRORS` bigint(20) unsigned NOT NULL COMMENT '错误数的总',
 
  `SUM_WARNINGS` bigint(20) unsigned NOT NULL COMMENT '警告的总数',
 
  `SUM_ROWS_AFFECTED` bigint(20) unsigned NOT NULL COMMENT '影响的总数目',
 
  `SUM_ROWS_SENT` bigint(20) unsigned NOT NULL COMMENT '返回总数目',
 
  `SUM_ROWS_EXAMINED` bigint(20) unsigned NOT NULL COMMENT '总的扫描的数目',
 
  `SUM_CREATED_TMP_DISK_TABLES` bigint(20) unsigned NOT NULL COMMENT '创建磁盘临时表的总数目',
 
  `SUM_CREATED_TMP_TABLES` bigint(20) unsigned NOT NULL COMMENT '创建临时表的总数目',
 
  `SUM_SELECT_FULL_JOIN` bigint(20) unsigned NOT NULL COMMENT '第一个表全表扫描的总数目',
 
  `SUM_SELECT_FULL_RANGE_JOIN` bigint(20) unsigned NOT NULL COMMENT '总的采用range方式扫描的数目',
 
  `SUM_SELECT_RANGE` bigint(20) unsigned NOT NULL COMMENT '第一个表采用range方式扫描的总数目',
 
  `SUM_SELECT_RANGE_CHECK` bigint(20) unsigned NOT NULL COMMENT '',
 
  `SUM_SELECT_SCAN` bigint(20) unsigned NOT NULL COMMENT '第一个表位全表扫描的总数目',
 
  `SUM_SORT_MERGE_PASSES` bigint(20) unsigned NOT NULL COMMENT '',
 
  `SUM_SORT_RANGE` bigint(20) unsigned NOT NULL COMMENT '范围排序总数',
 
  `SUM_SORT_ROWS` bigint(20) unsigned NOT NULL COMMENT '排序的记录总数目',
 
  `SUM_SORT_SCAN` bigint(20) unsigned NOT NULL COMMENT '第一个表排序扫描总数目',
 
  `SUM_NO_INDEX_USED` bigint(20) unsigned NOT NULL COMMENT '没有使用索引总数',
 
  `SUM_NO_GOOD_INDEX_USED` bigint(20) unsigned NOT NULL COMMENT '',
 
  `FIRST_SEEN` timestamp NOT NULL DEFAULT '0000-00-00 00:00:00' COMMENT '第一次执行时间',
 
  `LAST_SEEN` timestamp NOT NULL DEFAULT '0000-00-00 00:00:00' COMMENT '最后一次执行时间'
 
) ENGINE=PERFORMANCE_SCHEMA DEFAULT CHARSET=utf8
 
7，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：
 
zjy@performance_schema 11:36:22>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 1G
 
*************************** 1. row ***************************
 
      SCHEMA_NAME: dchat
 
      DIGEST_TEXT: SELECT ...
 
       COUNT_STAR: 1161210102
 
    SUM_ROWS_SENT: 1161207842
 
SUM_ROWS_EXAMINED:
 
       FIRST_SEEN: 2016-02-17 00:36:46
 
        LAST_SEEN: 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：
 
zjy@performance_schema 11:36:28>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 1G
 
*************************** 1. row ***************************
 
      SCHEMA_NAME: dchat
 
      DIGEST_TEXT: SELECT ...
 
       COUNT_STAR: 1
 
   AVG_TIMER_WAIT: 273238183964000
 
    SUM_ROWS_SENT: 50208
 
SUM_ROWS_EXAMINED: 5565651
 
       FIRST_SEEN: 2016-02-22 13:27:33
 
        LAST_SEEN: 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）：
 
zjy@performance_schema 12:16:18>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 2G
 
*************************** 1. row ***************************
 
                FILE_NAME: /var/lib/mysql/ibdata1  #文件
 
               EVENT_NAME: wait/io/file/innodb/innodb_data_file
 
               COUNT_READ: 544
 
 SUM_NUMBER_OF_BYTES_READ: 10977280
 
              COUNT_WRITE: 3700729
 
SUM_NUMBER_OF_BYTES_WRITE: 1433734217728
 
*************************** 2. row ***************************
 
                FILE_NAME: /var/lib/mysql/dchat/fans.ibd   #表
 
               EVENT_NAME: wait/io/file/innodb/innodb_data_file
 
               COUNT_READ: 9370680
 
 SUM_NUMBER_OF_BYTES_READ: 153529188352
 
              COUNT_WRITE: 67576376
 
SUM_NUMBER_OF_BYTES_WRITE: 1107815432192
 
8，哪个索引使用最多：
 
zjy@performance_schema 12:18:42>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;
 
+-------------+------------+-------------+--------------+--------------+--------------+
 
| OBJECT_NAME | INDEX_NAME | COUNT_FETCH | COUNT_INSERT | COUNT_UPDATE | COUNT_DELETE |
 
+-------------+------------+-------------+--------------+--------------+--------------+
 
| fans        | PRIMARY    | 29002695158 |            |    296373434 |            |
 
+-------------+------------+-------------+--------------+--------------+--------------+
 
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：
 
zjy@performance_schema 12:23:22>SELECT OBJECT_SCHEMA, OBJECT_NAME, INDEX_NAME FROM table_io_waits_summary_by_index_usage WHERE INDEX_NAME IS NOT NULL AND COUNT_STAR = AND OBJECT_SCHEMA <> 'mysql' ORDER BY OBJECT_SCHEMA,OBJECT_NAME;
 
10，哪个等待事件消耗的时间最多：
 
zjy@performance_schema 12:25:22>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;
 
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：
 
比如分析包含count(*)的某条SQL语句，具体如下：
 
SELECT
 
EVENT_ID,
 
sql_text
 
FROM events_statements_history
 
WHERE sql_text LIKE '%count(*)%';
 
+----------+--------------------------------------+
 
| EVENT_ID | sql_text |
 
+----------+--------------------------------------+
 
| 1690 | select count(*) from chuck.test_slow |
 
+----------+--------------------------------------+
 
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。
 
a.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 = 1690;
 
+----------+--------------------------------+----------------------+-----------------------+
 
| event_id | EVENT_NAME | SOURCE | TIMER_END-TIMER_START |
 
+----------+--------------------------------+----------------------+-----------------------+
 
| 1691 | stage/sql/init | mysqld.cc:990 | 316945000 |
 
| 1693 | stage/sql/checking permissions | sql_parse.cc:5776 | 26774000 |
 
| 1695 | stage/sql/Opening tables | sql_base.cc:4970 | 41436934000 |
 
| 2638 | stage/sql/init | sql_select.cc:1050 | 85757000 |
 
| 2639 | stage/sql/System lock | lock.cc:303 | 40017000 |
 
| 2643 | stage/sql/optimizing | sql_optimizer.cc:138 | 38562000 |
 
| 2644 | stage/sql/statistics | sql_optimizer.cc:362 | 52845000 |
 
| 2645 | stage/sql/preparing | sql_optimizer.cc:485 | 53196000 |
 
| 2646 | stage/sql/executing | sql_executor.cc:112 | 3153000 |
 
| 2647 | stage/sql/Sending data | sql_executor.cc:192 | 7369072089000 |
 
| 4304138 | stage/sql/end | sql_select.cc:1105 | 19920000 |
 
| 4304139 | stage/sql/query end | sql_parse.cc:5463 | 44721000 |
 
| 4304145 | stage/sql/closing tables | sql_parse.cc:5524 | 61723000 |
 
| 4304152 | stage/sql/freeing items | sql_parse.cc:6838 | 455678000 |
 
| 4304155 | stage/sql/logging slow query | sql_parse.cc:2258 | 83348000 |
 
| 4304159 | stage/sql/cleaning up | sql_parse.cc:2163 | 4433000 |
 
+----------+--------------------------------+----------------------+-----------------------+
 
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
 
b.查看某个阶段的锁等待情况
 
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。
 
SELECT
 
event_id,
 
event_name,
 
source,
 
timer_wait,
 
object_name,
 
index_name,
 
operation,
 
nesting_event_id
 
FROM events_waits_history_long
 
WHERE nesting_event_id = 2647;
 
+----------+---------------------------+-----------------+------------+-------------+------------+-----------+------------------+
 
| event_id | event_name | source | timer_wait | object_name | index_name | operation | nesting_event_id |
 
+----------+---------------------------+-----------------+------------+-------------+------------+-----------+------------------+
 
| 190607 | wait/io/table/sql/handler | handler.cc:2842 | 1845888 | test_slow | idx_c1 | fetch | 2647 |
 
| 190608 | wait/io/table/sql/handler | handler.cc:2842 | 1955328 | test_slow | idx_c1 | fetch | 2647 |
 
| 190609 | wait/io/table/sql/handler | handler.cc:2842 | 1929792 | test_slow | idx_c1 | fetch | 2647 |
 
| 190610 | wait/io/table/sql/handler | handler.cc:2842 | 1869600 | test_slow | idx_c1 | fetch | 2647 |
 
| 190611 | wait/io/table/sql/handler | handler.cc:2842 | 1922496 | test_slow | idx_c1 | fetch | 2647 |
 
+----------+---------------------------+-----------------+------------+-------------+------------+-----------+------------------+
 
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。
 
(2).模拟innodb行锁等待的例子
 
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。
 
SELECT
 
statement.EVENT_ID,
 
stages.event_id,
 
statement.sql_text,
 
stages.event_name,
 
stages.timer_wait
 
FROM events_statements_history_long statement
 
join events_stages_history_long stages
 
on statement.event_id=stages.nesting_event_id
 
WHERE statement.sql_text = 'update test_icp set y=y+1 where x=1';
 
+----------+----------+-------------------------------------+--------------------------------+----------------+
 
| EVENT_ID | event_id | sql_text | event_name | timer_wait |
 
+----------+----------+-------------------------------------+--------------------------------+----------------+
 
| 5816 | 5817 | update test_icp set y=y+1 where x=1 | stage/sql/init | 195543000 |
 
| 5816 | 5819 | update test_icp set y=y+1 where x=1 | stage/sql/checking permissions | 22730000 |
 
| 5816 | 5821 | update test_icp set y=y+1 where x=1 | stage/sql/Opening tables | 66079000 |
 
| 5816 | 5827 | update test_icp set y=y+1 where x=1 | stage/sql/init | 89116000 |
 
| 5816 | 5828 | update test_icp set y=y+1 where x=1 | stage/sql/System lock | 218744000 |
 
| 5816 | 5832 | update test_icp set y=y+1 where x=1 | stage/sql/updating | 6001362045000 |
 
| 5816 | 5968 | update test_icp set y=y+1 where x=1 | stage/sql/end | 10435000 |
 
| 5816 | 5969 | update test_icp set y=y+1 where x=1 | stage/sql/query end | 85979000 |
 
| 5816 | 5983 | update test_icp set y=y+1 where x=1 | stage/sql/closing tables | 56562000 |
 
| 5816 | 5990 | update test_icp set y=y+1 where x=1 | stage/sql/freeing items | 83563000 |
 
| 5816 | 5992 | update test_icp set y=y+1 where x=1 | stage/sql/cleaning up | 4589000 |
 
+----------+----------+-------------------------------------+--------------------------------+----------------+
 
查看wait事件：
 
SELECT
 
event_id,
 
event_name,
 
source,
 
timer_wait,
 
object_name,
 
index_name,
 
operation,
 
nesting_event_id
 
FROM events_waits_history_long
 
WHERE nesting_event_id = 5832;
 
*************************** 1. row ***************************
 
event_id: 5832
 
event_name: wait/io/table/sql/handler
 
source: handler.cc:2782
 
timer_wait: 6005946156624
 
object_name: test_icp
 
index_name: PRIMARY
 
operation: fetch
 
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。
 
(3).模拟MDL锁等待的例子
 
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。
 
SELECT
 
statement.EVENT_ID,
 
stages.event_id,
 
statement.sql_text,
 
stages.event_name as stage_name,
 
stages.timer_wait as stage_time
 
FROM events_statements_history_long statement
 
left join events_stages_history_long stages
 
on statement.event_id=stages.nesting_event_id
 
WHERE statement.sql_text = 'alter table test_slow modify c2 varchar(152)';
 
+-----------+-----------+----------------------------------------------+----------------------------------------------------+---------------+
 
| EVENT_ID | event_id | sql_text | stage_name | stage_time |
 
+-----------+-----------+----------------------------------------------+----------------------------------------------------+---------------+
 
| 326526744 | 326526745 | alter table test_slow modify c2 varchar(152) | stage/sql/init | 216662000 |
 
| 326526744 | 326526747 | alter table test_slow modify c2 varchar(152) | stage/sql/checking permissions | 18183000 |
 
| 326526744 | 326526748 | alter table test_slow modify c2 varchar(152) | stage/sql/checking permissions | 10294000 |
 
| 326526744 | 326526750 | alter table test_slow modify c2 varchar(152) | stage/sql/init | 4783000 |
 
| 326526744 | 326526751 | alter table test_slow modify c2 varchar(152) | stage/sql/Opening tables | 140172000 |
 
| 326526744 | 326526760 | alter table test_slow modify c2 varchar(152) | stage/sql/setup | 157643000 |
 
| 326526744 | 326526769 | alter table test_slow modify c2 varchar(152) | stage/sql/creating table | 8723217000 |
 
| 326526744 | 326526803 | alter table test_slow modify c2 varchar(152) | stage/sql/After create | 257332000 |
 
| 326526744 | 326526832 | alter table test_slow modify c2 varchar(152) | stage/sql/Waiting for table metadata lock | 1000181831000 |
 
| 326526744 | 326526835 | alter table test_slow modify c2 varchar(152) | stage/sql/After create | 33483000 |
 
| 326526744 | 326526838 | alter table test_slow modify c2 varchar(152) | stage/sql/Waiting for table metadata lock | 1000091810000 |
 
| 326526744 | 326526841 | alter table test_slow modify c2 varchar(152) | stage/sql/After create | 17187000 |
 
| 326526744 | 326526844 | alter table test_slow modify c2 varchar(152) | stage/sql/Waiting for table metadata lock | 1000126464000 |
 
| 326526744 | 326526847 | alter table test_slow modify c2 varchar(152) | stage/sql/After create | 27472000 |
 
| 326526744 | 326526850 | alter table test_slow modify c2 varchar(152) | stage/sql/Waiting for table metadata lock | 561996133000 |
 
| 326526744 | 326526853 | alter table test_slow modify c2 varchar(152) | stage/sql/After create | 124876000 |
 
| 326526744 | 326526877 | alter table test_slow modify c2 varchar(152) | stage/sql/System lock | 30659000 |
 
| 326526744 | 326526881 | alter table test_slow modify c2 varchar(152) | stage/sql/preparing for alter table | 40246000 |
 
| 326526744 | 326526889 | alter table test_slow modify c2 varchar(152) | stage/sql/altering table | 36628000 |
 
| 326526744 | 326528280 | alter table test_slow modify c2 varchar(152) | stage/sql/end | 43824000 |
 
| 326526744 | 326528281 | alter table test_slow modify c2 varchar(152) | stage/sql/query end | 112557000 |
 
| 326526744 | 326528299 | alter table test_slow modify c2 varchar(152) | stage/sql/closing tables | 27707000 |
 
| 326526744 | 326528305 | alter table test_slow modify c2 varchar(152) | stage/sql/freeing items | 201614000 |
 
| 326526744 | 326528308 | alter table test_slow modify c2 varchar(152) | stage/sql/cleaning up | 3584000 |
 
+-----------+-----------+----------------------------------------------+----------------------------------------------------+---------------+
 
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。
 
SELECT
 
event_id,
 
event_name,
 
source,
 
timer_wait,
 
object_name,
 
index_name,
 
operation,
 
nesting_event_id
 
FROM events_waits_history_long
 
WHERE nesting_event_id = 326526850;
 
+-----------+---------------------------------------------------+------------------+--------------+-------------+------------+------------+------------------+
 
| event_id | event_name | source | timer_wait | object_name | index_name | operation | nesting_event_id |
 
+-----------+---------------------------------------------------+------------------+--------------+-------------+------------+------------+------------------+
 
| 326526851 | wait/synch/cond/sql/MDL_context::COND_wait_status | mdl.cc:1327 | 562417991328 | NULL | NULL | timed_wait | 326526850 |
 
| 326526852 | wait/synch/mutex/mysys/my_thread_var::mutex | sql_class.h:3481 | 733248 | NULL | NULL | lock | 326526850 |
 
+-----------+---------------------------------------------------+------------------+--------------+-------------+------------+------------+------------------+
 
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。
 
View Code
 
“怎么理解MySQL5.6中的PERFORMANCE_SCHEM”的内容就介绍到这里了，感谢大家的阅读。

（编辑：大连站长网）

【声明】本站内容均来自网络，其相关言论仅代表作者个人观点，不代表本站立场。若无意侵犯到您的权利，请及时与联系站长删除相关内容!
    相关内容
      推荐文章
      站长推荐
      热点阅读

      【免责声明】本站内容转载自互联网，其发布内容言论不代表本站观点，如果其链接、内容的侵犯您的权益，烦请提交相关链接至邮箱bqsm@foxmail.com我们将及时予以处理。

      建议您使用1920×1080分辨率、谷歌浏览器Google Chrome、Microsoft Edge以获得本站的最佳浏览效果

      Copygight © 2008-2022 https://www.0411zz.cn/ All Rights Reserved. 大连站长网