MySQL 数据库监控
可用性监控
1.网络监控
使用mysqladmin：
[root@mysql10 mysql]# mysqladmin -uroot -p ping
Enter password:
mysqld is alive
使用telnet：
成功则返回：
[root@mysql10 mysql]# telnet localhost 3306
Trying 127.0.0.1...
Connected to localhost.
Escape character is '^]'.
N
失败则返回：
[root@mysql10 mysql]# telnet 10.104.103.2 3306
Trying 10.104.103.2...
telnet: connect to address 10.104.103.2: No route to host
2.连接数
查看设置最大连接数：
[root@mysql10 mysql]# mysql -uroot -p -e "show variables like 'max_connections'"
Enter password:
+-----------------+-------+
| Variable_name | Value |
+-----------------+-------+
| max_connections | 1000 |
+-----------------+-------+
查看当前连接数：
[root@mysql10 mysql]# mysql -uroot -p -e "show global status like 'Threads_connected'"
Enter password:
+-------------------+-------+
| Variable_name | Value |
+-------------------+-------+
| Threads_connected | 54 |
+-------------------+-------+
性能监控
1.QPS
QPS，即每秒数据库执行的请求数量。
公式为：
QPS=(Queries2-Queries1)/(Uptime_since_flush_status2-Uptime_since_flush_status1)
2.TPS
TPS，即每秒数据库处理的事务数量。
公式为：
TPS=((Com_insert2+Com_update2+Com_delete2)-(Com_insert1+Com_update1+Com_delete1))/(Uptime_since_flush_status2-Uptime_since_flush_status1)
3.并发请求
查看数据库并发请求数量：
[root@mysql10 mysql]# mysql -uroot -p -e "show global status like 'Threads_running'"
Enter password:
+-----------------+-------+
| Variable_name | Value |
+-----------------+-------+
| Threads_running | 1 |
+-----------------+-------+
4.Innodb阻塞
可使用下面脚本查看锁住的线程，脚本会返回所有阻塞时间大于10s的线程：
SELECT b.trx_mysql_thread_id AS '被阻塞线程'
,b.trx_query AS '被阻塞SQL'
,c.trx_mysql_thread_id AS '阻塞线程'
,c.trx_query AS '阻塞SQL'
,(UNIX_TIMESTAMP()-UNIX_TIMESTAMP(c.trx_started)) AS '阻塞时间'
FROM information_schema.innodb_lock_waits a
JOIN information_schema.innodb_trx b ON a.requesting_trx_id=b.trx_id
JOIN information_schema.innodb_trx c ON a.blocking_trx_id=c.trx_id
WHERE (UNIX_TIMESTAMP()-UNIX_TIMESTAMP(c.trx_started))>10
查看连接号：
mysql> select connection_id();
+-----------------+
| connection_id() |
+-----------------+
| 382666 |
+-----------------+
1 row in set (0.00 sec)
主从复制监控
1.链路状态
使用show slave status命令查看：
Slave_IO_Running: Yes
Slave_SQL_Running: Yes
2.主从复制延迟
使用show slave status命令查看：
Seconds_Behind_Master: 0
再检查主从gtid是否同步：
主库：
mysql> show master status\G
*************************** 1. row ***************************
File: mysqlbin.000875
Position: 194
Binlog_Do_DB:
Binlog_Ignore_DB:
Executed_Gtid_Set: c94e3c22-72eb-11e9-b569-005056a86449:1-78241
1 row in set (0.00 sec)
从库：
Master_Log_File: mysqlbin.000875
Read_Master_Log_Pos: 194
Master_UUID: c94e3c22-72eb-11e9-b569-005056a86449
Retrieved_Gtid_Set: c94e3c22-72eb-11e9-b569-005056a86449:9-78241
Executed_Gtid_Set: c94e3c22-72eb-11e9-b569-005056a86449:8-78241
说明主从同步
3.主从一致性检测
使用pt-table-checksum工具，在主库上运行：
pt-table-checksum u=root,p='xxx' \
--databases mysql \
--replicate test.checksums
