mysql InnoDB锁等待的查看及分析

说明：前面已经了解了InnoDB关于在出现锁等待的时候，会根据参数innodb_lock_wait_timeout的配置，判断是否需要进行timeout的操作，本文档介绍在出现锁等待时候的查看及分析处理；

在InnoDB Plugin之前，一般通过show full processlist和show engine innodb status命令查看当前的数据库请求，然后再判断当前事务中锁的情况。随着mysql的发展，已经提供更加便捷的方法来监控数据库中的锁等待现象了。

在information_schema下面有三张表:INNODB_TRX、INNODB_LOCKS、INNODB_LOCK_WAITS，通过这三张表，可以更简单地监控当前的事务并分析可能存在的问题。

• INNODB_TRX表及结构

Column name	Description
TRX_ID	Unique transaction ID number, internal to InnoDB. (Starting in MySQL 5.6, these IDs are not created for transactions that are read-only and non-locking. See Optimizing InnoDB Read-Only Transactions for details.)
TRX_WEIGHT	The weight of a transaction, reflecting (but not necessarily the exact count of) the number of rows altered and the number of rows locked by the transaction. To resolve a deadlock, InnoDB selects the transaction with the smallest weight as the “victim” to rollback. Transactions that have changed non-transactional tables are considered heavier than others, regardless of the number of altered and locked rows.
TRX_STATE	Transaction execution state. One of RUNNING, LOCK WAIT, ROLLING BACK or COMMITTING.
TRX_STARTED	Transaction start time.
TRX_REQUESTED_LOCK_ID	ID of the lock the transaction is currently waiting for (if TRX_STATE is LOCK WAIT, otherwise NULL). Details about the lock can be found by joining with INNODB_LOCKS on LOCK_ID.
TRX_WAIT_STARTED	Time when the transaction started waiting on the lock (if TRX_STATE is LOCK WAIT, otherwise NULL).
TRX_MYSQL_THREAD_ID	MySQL thread ID. Can be used for joining with PROCESSLIST on ID. See Section 14.17.2.3.1, “Potential Inconsistency with PROCESSLIST Data”.
TRX_QUERY	The SQL query that is being executed by the transaction.
TRX_OPERATION_STATE	The transaction's current operation, or NULL.
TRX_TABLES_IN_USE	The number of InnoDB tables used while processing the current SQL statement of this transaction.
TRX_TABLES_LOCKED	Number of InnoDB tables that the current SQL statement has row locks on. (Because these are row locks, not table locks, the tables can usually still be read from and written to by multiple transactions, despite some rows being locked.)
TRX_LOCK_STRUCTS	The number of locks reserved by the transaction.
TRX_LOCK_MEMORY_BYTES	Total size taken up by the lock structures of this transaction in memory.
TRX_ROWS_LOCKED	Approximate number or rows locked by this transaction. The value might include delete-marked rows that are physically present but not visible to the transaction.
TRX_ROWS_MODIFIED	The number of modified and inserted rows in this transaction.
TRX_CONCURRENCY_TICKETS	A value indicating how much work the current transaction can do before being swapped out, as specified by the innodb_concurrency_tickets option.
TRX_ISOLATION_LEVEL	The isolation level of the current transaction.
TRX_UNIQUE_CHECKS	Whether unique checks are turned on or off for the current transaction. (They might be turned off during a bulk data load, for example.)
TRX_FOREIGN_KEY_CHECKS	Whether foreign key checks are turned on or off for the current transaction. (They might be turned off during a bulk data load, for example.)
TRX_LAST_FOREIGN_KEY_ERROR	Detailed error message for last FK error, or NULL.
TRX_ADAPTIVE_HASH_LATCHED	Whether or not the adaptive hash index is locked by the current transaction. (Only a single transaction at a time can modify the adaptive hash index.)
TRX_ADAPTIVE_HASH_TIMEOUT	Whether to relinquish the search latch immediately for the adaptive hash index, or reserve it across calls from MySQL. When there is no AHI contention, this value remains zero and statements reserve the latch until they finish. During times of contention, it counts down to zero, and statements release the latch immediately after each row lookup.
TRX_IS_READ_ONLY	A value of 1 indicates the transaction is read-only. (5.6.4 and up.)
TRX_AUTOCOMMIT_NON_LOCKING	A value of 1 indicates the transaction is a SELECT statement that does not use the FOR UPDATE or LOCK IN SHARED MODE clauses, and is executing with the autocommit setting turned on so that the transaction will only contain this one statement. (5.6.4 and up.) When this column and TRX_IS_READ_ONLY are both 1, InnoDB optimizes the transaction to reduce the overhead associated with transactions that change table data.

比较常用的列：

trx_id:InnoDB存储引擎内部唯一的事物ID
trx_status:当前事务的状态
trx_status:事务的开始时间
trx_requested_lock_id：等待事务的锁ID
trx_wait_started：事务等待的开始时间
trx_weight：事务的权重，反应一个事务修改和锁定的行数，当发现死锁需要回滚时，权重越小的值被回滚
trx_mysql_thread_id：MySQL中的进程ID，与show processlist中的ID值相对应
trx_query：事务运行的SQL语句

• INNODB_LOCKS

Column name	Description
LOCK_ID	Unique lock ID number, internal to InnoDB. Treat it as an opaque string. Although LOCK_ID currently contains TRX_ID, the format of the data in LOCK_ID is not guaranteed to remain the same in future releases. Do not write programs that parse the LOCK_ID value.
LOCK_TRX_ID	ID of the transaction holding this lock. Details about the transaction can be found by joining with INNODB_TRX on TRX_ID.
LOCK_MODE	Mode of the lock. One of S, X, IS, IX, S_GAP, X_GAP, IS_GAP, IX_GAP, or AUTO_INC for shared, exclusive, intention shared, intention exclusive row locks, shared and exclusive gap locks, intention shared and intention exclusive gap locks, and auto-increment table level lock, respectively. Refer to the sections Section 14.5.3, “InnoDB Lock Modes” and Section 14.5.2, “The InnoDB Transaction Model and Locking” for information on InnoDB locking.
LOCK_TYPE	Type of the lock. One of RECORD or TABLE for record (row) level or table level locks, respectively.
LOCK_TABLE	Name of the table that has been locked or contains locked records.
LOCK_INDEX	Name of the index if LOCK_TYPE='RECORD', otherwise NULL.
LOCK_SPACE	Tablespace ID of the locked record if LOCK_TYPE='RECORD', otherwise NULL.
LOCK_PAGE	Page number of the locked record if LOCK_TYPE='RECORD', otherwise NULL.
LOCK_REC	Heap number of the locked record within the page if LOCK_TYPE='RECORD', otherwise NULL.
LOCK_DATA	Primary key value(s) of the locked record if LOCK_TYPE='RECORD', otherwise NULL. This column contains the value(s) of the primary key column(s) in the locked row, formatted as a valid SQL string (ready to be copied to SQL commands). If there is no primary key then the InnoDB internal unique row ID number is used. If a gap lock is taken for key values or ranges above the largest value in the index, LOCK_DATA reports “supremum pseudo-record”. When the page containing the locked record is not in the buffer pool (in the case that it was paged out to disk while the lock was held), InnoDB does not fetch the page from disk, to avoid unnecessary disk operations. Instead, LOCK_DATA is set to NULL.

• INNODB_LOCK_WAITS

Column name	Description
REQUESTING_TRX_ID	ID of the requesting transaction.
REQUESTED_LOCK_ID	ID of the lock for which a transaction is waiting. Details about the lock can be found by joining with INNODB_LOCKS on LOCK_ID.
BLOCKING_TRX_ID	ID of the blocking transaction.
BLOCKING_LOCK_ID	ID of a lock held by a transaction blocking another transaction from proceeding. Details about the lock can be found by joining with INNODB_LOCKS on LOCK_ID.

以上这些表，其实只要知道其中比较常用的字段，就差不多能够满足日常的工作需求了，下面通过测试进行演示；

一、准备工作

1、在test下面随便创建一张表john，并取消自动commit操作，脚本如下：

mysql> use information_schema
Database changed

mysql> select count(*) from tables;
+----------+
| count(*) |
+----------+
|       81 |
+----------+
1 row in set (0.06 sec)

mysql> create table test.john as select * from tables;
Query OK, 82 rows affected (0.29 sec)
Records: 82  Duplicates: 0  Warnings: 0

mysql> insert into john select * from john;
Query OK, 671744 rows affected (2 min 19.03 sec)
Records: 671744  Duplicates: 0  Warnings: 0

（经过几次插入后john表的数据671744行）

mysql> set @@autocommit=0;
Query OK, 0 rows affected (0.00 sec)

（取消数据库的自动commit）

二、进行表john加锁操作，脚本如下：

mysql> select count(*) from john for update;
+----------+
| count(*) |
+----------+
|  2686976 |
+----------+
1 row in set (8.19 sec)

在另外一个窗口中监控innodb锁的状态；

mysql> SELECT  * FROM INNODB_TRX\G;
*************************** 1. row ***************************
                    trx_id: B14                                                /请记住该trx_id/
                 trx_state: RUNNING                                       /当前状态/
               trx_started: 2014-11-29 14:07:51
     trx_requested_lock_id: NULL
          trx_wait_started: NULL
                trx_weight: 15905
       trx_mysql_thread_id: 10                                          /在process 里面的id值/
                 trx_query: select count(*) from john for update;    /当前执行的语句/
       trx_operation_state: fetching rows
         trx_tables_in_use: 1
         trx_tables_locked: 1
          trx_lock_structs: 15905
     trx_lock_memory_bytes: 1554872
           trx_rows_locked: 1360743
         trx_rows_modified: 0
   trx_concurrency_tickets: 0
       trx_isolation_level: REPEATABLE READ
         trx_unique_checks: 1
    trx_foreign_key_checks: 1
trx_last_foreign_key_error: NULL
 trx_adaptive_hash_latched: 0
 trx_adaptive_hash_timeout: 10000
1 row in set (0.02 sec)

trx_id: B14 只是持有锁，但并没有产生锁等待；

三、模拟锁等待

3.1 在另外一个窗口中，执行语句：

mysql> select count(*) from john where table_name='CHARACTER_SETS' for update;

3.2 查看当前锁等待的情况

INNODB_TRX的锁情况：

mysql> SELECT  * FROM INNODB_TRX\G;
*************************** 1. row ***************************
                    trx_id: B15                      
                 trx_state: LOCK WAIT                       //状态为锁等待//
               trx_started: 2014-11-29 14:12:28
     trx_requested_lock_id: B15:0:32777:2
          trx_wait_started: 2014-11-29 14:12:28
                trx_weight: 2
       trx_mysql_thread_id: 10                           //在process里面可以看到相应的状态// 
                 trx_query: select count(*) from john where table_name='CHARACTER_SETS' for update                //锁等待的语句//
       trx_operation_state: starting index read
         trx_tables_in_use: 1
         trx_tables_locked: 1
          trx_lock_structs: 2
     trx_lock_memory_bytes: 376
           trx_rows_locked: 1
         trx_rows_modified: 0
   trx_concurrency_tickets: 0
       trx_isolation_level: REPEATABLE READ
         trx_unique_checks: 1
    trx_foreign_key_checks: 1
trx_last_foreign_key_error: NULL
 trx_adaptive_hash_latched: 0
 trx_adaptive_hash_timeout: 10000
*************************** 2. row ***************************
                    trx_id: B14                                   
                 trx_state: RUNNING
               trx_started: 2014-11-29 14:07:51
     trx_requested_lock_id: NULL
          trx_wait_started: NULL
                trx_weight: 31777
       trx_mysql_thread_id: 8
                 trx_query: NULL
       trx_operation_state: NULL
         trx_tables_in_use: 0
         trx_tables_locked: 0
          trx_lock_structs: 31777
     trx_lock_memory_bytes: 3094968
           trx_rows_locked: 2718752
         trx_rows_modified: 0
   trx_concurrency_tickets: 0
       trx_isolation_level: REPEATABLE READ
         trx_unique_checks: 1
    trx_foreign_key_checks: 1
trx_last_foreign_key_error: NULL
 trx_adaptive_hash_latched: 0
 trx_adaptive_hash_timeout: 10000
2 rows in set (0.02 sec)

请注意：因为我们只有模拟两个session，所以这边只有两个会话。（因此一个处于锁等待，另外一个必然就是持有锁的对象。实际的生产环境中可能这边会出现很多列，所以需要用下面的语句才能判断：锁等待和持有锁对象的匹配关系）

3.3 锁等待和持有锁的相互关系

mysql> SELECT * FROM INNODB_LOCK_WAITS\G;
*************************** 1. row ***************************
requesting_trx_id: B15
requested_lock_id: B15:0:32777:2
  blocking_trx_id: B14
 blocking_lock_id: B14:0:32777:2
1 row in set (0.03 sec)

ERROR: 
No query specified

通过视图INNODB_LOCK_WAITS可以清晰的看到B14持有锁，而B15处于锁等待；

3.4 锁等待的原因

mysql> SELECT * FROM INNODB_LOCKS\G;
*************************** 1. row ***************************
    lock_id: B15:0:32777:2
lock_trx_id: B15
  lock_mode: X
  lock_type: RECORD
 lock_table: `test`.`john`
 lock_index: `GEN_CLUST_INDEX`
 lock_space: 0
  lock_page: 32777
   lock_rec: 2
  lock_data: 0x000000640000
*************************** 2. row ***************************
    lock_id: B14:0:32777:2
lock_trx_id: B14
  lock_mode: X
  lock_type: RECORD
 lock_table: `test`.`john`
 lock_index: `GEN_CLUST_INDEX`
 lock_space: 0
  lock_page: 32777
   lock_rec: 2
  lock_data: 0x000000640000
2 rows in set (0.01 sec)

可以看到持有锁的模式、对象

3.5 在进程里面查看状态

Id值为8的进程，Info显示为NULL值，可以推断当前的session由于未进行commit导致锁未释放的；

总结：通过以上几个视图，就可以很快速的判断出锁等待的对象及原因了，从这上面也可以看出mysql管理更加便捷和容易了；