MySQL慢查询日志(二)

fgew2

慢查询日志用来优化Query语句，以下是有关的几个参数，5.5版本以后可以到微妙（ms）
mysql> show variables like '%slow%';
+---------------------------+-------------------------------+
| Variable_name             | Value                         |
+---------------------------+-------------------------------+
| log_slow_admin_statements | OFF                           |
| log_slow_slave_statements | OFF                           |
| slow_launch_time          | 2                             |
| slow_query_log            | ON                            |
| slow_query_log_file       | /opt/mysql/data/hack-slow.log |
+---------------------------+-------------------------------+
5 rows in set (0.45 sec)

mysql> show variables like '%long%';
+--------------------------------------------------------+----------+
| Variable_name                                          | Value    |
+--------------------------------------------------------+----------+
| long_query_time                                        | 0.050000 |
| performance_schema_events_stages_history_long_size     | 10000    |
| performance_schema_events_statements_history_long_size | 10000    |
| performance_schema_events_waits_history_long_size      | 10000    |
+--------------------------------------------------------+----------+
4 rows in set (0.03 sec)

mysql> 通过tail或者mysqldumslow查看慢查询日志

[iyunv@hack data]# tail -f hack-slow.log
Tcp port: 3306  Unix socket: /usr/local/mysql/mysql.sock
Time                 Id Command    Argument
/usr/local/mysql/bin/mysqld, Version: 5.6.14-log (Source distribution). started with:
Tcp port: 3306  Unix socket: /usr/local/mysql/mysql.sock
Time                 Id Command    Argument
# Time: 150417  1:23:49
# User@Host: root[root] @ localhost []  Id:     1
# Query_time: 0.118892  Lock_time: 0.000000 Rows_sent: 1  Rows_examined: 0
SET timestamp=1429205029;
select version();

# Time: 150417  1:24:20
# User@Host: root[root] @ localhost []  Id:     1
# Query_time: 0.426234  Lock_time: 0.139966 Rows_sent: 5  Rows_examined: 5
SET timestamp=1429205060;
show variables like '%slow%';
^C
[iyunv@hack data]# mysqldumpslow hack-slow.log

Reading mysql slow query log from hack-slow.log
Count: 1  Time=0.29s (0s)  Lock=0.14s (0s)  Rows=5.0 (5), root[root]@localhost
  show variables like 'S'

Count: 1  Time=0.12s (0s)  Lock=0.00s (0s)  Rows=1.0 (1), root[root]@localhost
  select version()

[iyunv@hack data]#
[iyunv@hack data]# mysqldumpslow --help
Usage: mysqldumpslow [ OPTS... ] [ LOGS... ]

Parse and summarize the MySQL slow query log. Options are

  --verbose    verbose
  --debug      debug
  --help       write this text to standard output

  -v           verbose
  -d           debug
  -s ORDER     what to sort by (al, at, ar, c, l, r, t), 'at' is default
                al: average lock time
                ar: average rows sent
                at: average query time
                 c: count
                 l: lock time
                 r: rows sent
                 t: query time  
  -r           reverse the sort order (largest last instead of first)
  -t NUM       just show the top n queries
  -a           don't abstract all numbers to N and strings to 'S'
  -n NUM       abstract numbers with at least n digits within names
  -g PATTERN   grep: only consider stmts that include this string
  -h HOSTNAME  hostname of db server for *-slow.log filename (can be wildcard),
               default is '*', i.e. match all
  -i NAME      name of server instance (if using mysql.server startup script)
  -l           don't subtract lock time from total time

[iyunv@hack data]#
另外从5.1版本之后，开始慢查询日志可以记录到数据库中，在mysql中存在一个slow_log的表
mysql> show create table mysql.slow_log \G
*************************** 1. row ***************************
       Table: slow_log
Create Table: CREATE TABLE `slow_log` (
  `start_time` timestamp NOT NULL DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP,
  `user_host` mediumtext NOT NULL,
  `query_time` time NOT NULL,
  `lock_time` time NOT NULL,
  `rows_sent` int(11) NOT NULL,
  `rows_examined` int(11) NOT NULL,
  `db` varchar(512) NOT NULL,
  `last_insert_id` int(11) NOT NULL,
  `insert_id` int(11) NOT NULL,
  `server_id` int(10) unsigned NOT NULL,
  `sql_text` mediumtext NOT NULL,
  `thread_id` bigint(21) unsigned NOT NULL
) ENGINE=CSV DEFAULT CHARSET=utf8 COMMENT='Slow log'
1 row in set (0.07 sec)

mysql> 可以看到该表的引擎为CSV

mysql> show variables like '%log_output%';
+---------------+-------+
| Variable_name | Value |
+---------------+-------+
| log_output    | FILE  |
+---------------+-------+
1 row in set (0.03 sec)

mysql> set global log_output='FILE,TABLE';更改日志输出到文件和表中
Query OK, 0 rows affected (0.07 sec)

mysql> show variables like '%log_output%';
+---------------+------------+
| Variable_name | Value      |
+---------------+------------+
| log_output    | FILE,TABLE |
+---------------+------------+
1 row in set (0.09 sec)

mysql> select sleep(10);
+-----------+
| sleep(10) |
+-----------+
|         0 |
+-----------+
1 row in set (10.09 sec)

mysql> select * from mysql.slow_log \G 我从打开table到查询有两条语句达到慢查询设定的时间
*************************** 1. row ***************************
    start_time: 2015-04-17 01:33:50
     user_host: root[root] @ localhost []
    query_time: 00:00:00
     lock_time: 00:00:00
     rows_sent: 0
rows_examined: 0
            db:
last_insert_id: 0
     insert_id: 0
     server_id: 13
      sql_text: set global log_output='FILE,TABLE'
     thread_id: 1
*************************** 2. row ***************************
    start_time: 2015-04-17 01:34:17
     user_host: root[root] @ localhost []
    query_time: 00:00:10
     lock_time: 00:00:00
     rows_sent: 1
rows_examined: 0
            db:
last_insert_id: 0
     insert_id: 0
     server_id: 13
      sql_text: select sleep(10)
     thread_id: 1
2 rows in set (0.01 sec)

mysql> 修改slow_log的存储引擎为myisam
mysql> alter table mysql.slow_log engine=myisam;
ERROR 1580 (HY000): You cannot 'ALTER' a log table if logging is enabled
mysql> set global slow_query_log=off;
Query OK, 0 rows affected (0.00 sec)

mysql> alter table mysql.slow_log engine=myisam;
Query OK, 3 rows affected (0.15 sec)
Records: 3  Duplicates: 0  Warnings: 0

mysql>