innodb 鎖機制

innodb存儲引擎不須要鎖升級，由於一個鎖和多個鎖的開銷是相同的。

鎖是數據庫系統區別於文件系統的一個關鍵特性。鎖機制用於管理對共享資源的併發訪問。Innodb引擎中使用鎖的地方有多個：在行級別上對錶數據上鎖；操做緩衝池中的LRU列表，刪除，添加，移動LRU列表中的元素。

鎖信息的查看命令：

• show engine innodb status;
• information_schema架構下的表：innodb_trx，innodb_locks，innodb_lock_waits

鎖的類型

innodb實現了兩種標準的行級鎖：

• 共享鎖（S LOCK),容許事務讀一行數據
• 排他鎖（X lock），容許事務刪除或更新一行數據

鎖兼容：若是一個事務T1已經得到了r行的共享鎖，那麼另外的事務T2能夠當即得到行r的共享鎖，由於讀取並無改變行r的數據。

鎖不兼容：若是T3想得到r行的x鎖，則必須等T1,T2釋放行r上的S鎖

 

 

S鎖和X鎖的兼容性

	X	S
X	不兼容	不兼容
S	不兼容	兼容

innodb支持兩個鐘意向鎖（在innodb中即爲表鎖）：

• 意向共享鎖（IS lock），事務想要得到一張表中某幾行的共享鎖
• 意向排他鎖（IX lock），事務想要得到一張表中某幾行的排他鎖

查看innodb隔離級別：

mysql> SELECT @@global.tx_isolation;
+-----------------------+
| @@global.tx_isolation |
+-----------------------+
| REPEATABLE-READ       |
+-----------------------+
1 row in set (0.00 sec)

　　

實驗過程：

建表test：

mysql> CREATE TABLE `test` (
    ->   `id` bigint(20) NOT NULL,
    ->   `name` varchar(20) NOT NULL DEFAULT "name",
    ->   PRIMARY KEY (id),
    ->   KEY `index_name` (`name`)
    -> ) ENGINE=InnoDB DEFAULT CHARSET=utf8;
Query OK, 0 rows affected (0.99 sec)

填充數據：

mysql> insert into test (id) values(1),(2),(3),(9);
Query OK, 4 rows affected (0.08 sec)
Records: 4  Duplicates: 0  Warnings: 0

mysql> select * from test;
+----+------+
| id | name |
+----+------+
|  1 | name |
|  2 | name |
|  3 | name |
|  9 | name |
+----+------+
4 rows in set (0.00 sec)

　

實驗數據：

說明：事務一，事務二都有數據，順序是先執行事務一，再執行事務二。

事務一	事務二	innodb_trx	innodb_locks	innodb_lock_waits
mysql> begin; select * from test where id=1 for update; Query OK, 0 rows affected (0.00 sec) +----+------+ | id | name | +----+------+ | 1 | name | +----+------+ 1 row in set (0.00 sec)		mysql> select * from information_schema.INNODB_TRX\G *************************** 1. row *************************** trx_id: 1074396261 trx_state: RUNNING trx_started: 2017-04-26 15:18:25 trx_requested_lock_id: NULL trx_wait_started: NULL trx_weight: 2 trx_mysql_thread_id: 8950735 trx_query: NULL trx_operation_state: NULL trx_tables_in_use: 0 trx_tables_locked: 0 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 trx_is_read_only: 0 trx_autocommit_non_locking: 0 1 row in set (0.00 sec)	mysql> select * from information_schema.INNODB_LOCKS\G Empty set (0.00 sec)	mysql> SELECT * FROM INFORMATION_SCHEMA.INNODB_LOCK_WAITS\G Empty set (0.00 sec)
	mysql> begin;select * from test where id=1 lock in share mode; Query OK, 0 rows affected (0.00 sec)	mysql> select * from information_schema.INNODB_TRX\G *************************** 1. row *************************** trx_id: 1074397998 trx_state: LOCK WAIT trx_started: 2017-04-26 15:21:02 trx_requested_lock_id: 1074397998:5594:3:2 trx_wait_started: 2017-04-26 15:21:02 trx_weight: 2 trx_mysql_thread_id: 8959888 trx_query: select * from test where id=1 lock in share mode 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 trx_is_read_only: 0 trx_autocommit_non_locking: 0 *************************** 2. row *************************** trx_id: 1074396261 trx_state: RUNNING trx_started: 2017-04-26 15:18:25 trx_requested_lock_id: NULL trx_wait_started: NULL trx_weight: 2 trx_mysql_thread_id: 8950735 trx_query: NULL trx_operation_state: NULL trx_tables_in_use: 0 trx_tables_locked: 0 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 trx_is_read_only: 0 trx_autocommit_non_locking: 0 2 rows in set (0.00 sec)	mysql> select * from information_schema.INNODB_LOCKS\G *************************** 1. row *************************** lock_id: 1074397998:5594:3:2 lock_trx_id: 1074397998 lock_mode: S lock_type: RECORD lock_table: `ztest`.`test` lock_index: PRIMARY lock_space: 5594 lock_page: 3 lock_rec: 2 lock_data: 1 *************************** 2. row *************************** lock_id: 1074396261:5594:3:2 lock_trx_id: 1074396261 lock_mode: X lock_type: RECORD lock_table: `ztest`.`test` lock_index: PRIMARY lock_space: 5594 lock_page: 3 lock_rec: 2 lock_data: 1 2 rows in set (0.00 sec)	mysql> SELECT * FROM INFORMATION_SCHEMA.INNODB_LOCK_WAITS\G *************************** 1. row *************************** requesting_trx_id: 1074397998 requested_lock_id: 1074397998:5594:3:2 blocking_trx_id: 1074396261 blocking_lock_id: 1074396261:5594:3:2 1 row in set (0.00 sec)
mysql> update test set name="name1" where id=1; Query OK, 1 row affected (0.00 sec) Rows matched: 1 Changed: 1 Warnings: 0 mysql> select * from test; +----+-------+ | id | name | +----+-------+ | 2 | name | | 3 | name | | 9 | name | | 1 | name1 | +----+-------+ 4 rows in set (0.00 sec)	mysql> select * from test; +----+------+ | id | name | +----+------+ | 1 | name | | 2 | name | | 3 | name | | 9 | name | +----+------+ 4 rows in set (0.00 sec) 　　該查詢下，id爲1的行，並無修改name，由於事務一的修改並無提交。
mysql> commit -> ; Query OK, 0 rows affected (0.08 sec) mysql> select * from test; +----+-------+ | id | name | +----+-------+ | 2 | name | | 3 | name | | 9 | name | | 1 | name1 | +----+-------+ 4 rows in set (0.00 sec)	mysql> select * from test; +----+------+ | id | name | +----+------+ | 1 | name | | 2 | name | | 3 | name | | 9 | name | +----+------+ 4 rows in set (0.00 sec) 　　事務一，已經提交，該查詢下，id爲1的行，name沒有發生變化，代表在RR隔離級別下的，可重複讀：同一個事務的，同一個select讀出的數據徹底一致。
	mysql> commit; Query OK, 0 rows affected (0.00 sec) mysql> select * from test; +----+-------+ | id | name | +----+-------+ | 2 | name | | 3 | name | | 9 | name | | 1 | name1 | +----+-------+ 4 rows in set (0.00 sec) 　　兩個事務都結束了。
實驗二
事務一	事務二	innodb_trx	innodb_locks	innodb_lock_waits
mysql> begin;select * from test where id>0 and id<5 for update; Query OK, 0 rows affected (0.00 sec) +----+-------+ | id | name | +----+-------+ | 2 | name | | 3 | name | | 1 | name1 | +----+-------+ 3 rows in set (0.00 sec)	mysql> begin;update test set name="nihao" where id=1; Query OK, 0 rows affected (0.00 sec) select * from test; 　　事務二的兩個操做都被阻塞		mysql> select * from information_schema.INNODB_LOCKS\G *************************** 1. row *************************** lock_id: 1074413840:5594:3:6 lock_trx_id: 1074413840 lock_mode: X lock_type: RECORD lock_table: `ztest`.`test` lock_index: PRIMARY lock_space: 5594 lock_page: 3 lock_rec: 6 lock_data: 1 *************************** 2. row *************************** lock_id: 1074413574:5594:3:6 lock_trx_id: 1074413574 lock_mode: X lock_type: RECORD lock_table: `ztest`.`test` lock_index: PRIMARY lock_space: 5594 lock_page: 3 lock_rec: 6 lock_data: 1 2 rows in set (0.00 sec)	mysql> SELECT * FROM INFORMATION_SCHEMA.INNODB_LOCK_WAITS\G *************************** 1. row *************************** requesting_trx_id: 1074413840 requested_lock_id: 1074413840:5594:3:6 blocking_trx_id: 1074413574 blocking_lock_id: 1074413574:5594:3:6 1 row in set (0.00 sec)
mysql> update test set name="fjsld" where id=1; Query OK, 1 row affected (0.00 sec) Rows matched: 1 Changed: 1 Warnings: 0 mysql> select * from test; +----+-------+ | id | name | +----+-------+ | 1 | fjsld | | 2 | name | | 3 | name | | 9 | name | +----+-------+ 4 rows in set (0.00 sec)
mysql> commit; Query OK, 0 rows affected (0.02 sec)	mysql> update test set name="nihao" where id=1; Query OK, 1 row affected (1 min 45.44 sec) Rows matched: 1 Changed: 1 Warnings: 0 　　 事務一提交，事務二更新成功  事務一優先獲取了X鎖，此後，事務二想要獲取X鎖，須要等候

 

常見概念：

1. select ...for update
   1. 鎖定查詢出來的行，其餘事務要修改該行，須要等待,該行會上X鎖
2. select ...lock in share mode;
   1. 鎖定查詢出來的行，其餘事務須要修改改行，須要等待，該行會上S鎖
3. 四種事務隔離級別：
   1. RR:可重複讀
   2. RC：提交讀
   3. RU：非提交讀
   4. SEARILIZED:事務串行執行
4. innodb支持的兩種級別舉例分析
   1. RR:repeatable read：可重複讀
   2. RC:read commit：提交讀
5. 髒讀，幻讀，不可重複讀
   1. 髒讀
      1. 事務一，能讀到事務二尚未提交的更改信息
      2. 隔離級別最低的RU中會出現此種狀況
   2. 幻讀
      1. 不可重複讀的特殊狀況：同一條sql語句，是範圍查找類型的，兩次獲得的結果不同
      2. 事務二在某個範圍內插入一條數據，並提交；事務一，再次範圍查找獲得不同的結果
      3. innodb的RR下，此種類型不會出現
   3. 不可重複讀
      1. 在同一個事務中，一樣一條sql語句，執行兩次，獲得的結果不同
      2. 適用場景：update
      3. innodb的RR下，此種類型不會出現
6. 設置innodb_lock_wait_timeout
   1. 相關操做

      1. mysql> show global variables like "%timeout%";
         +-----------------------------+----------+
         | Variable_name               | Value    |
         +-----------------------------+----------+
         | connect_timeout             | 10       |
         | delayed_insert_timeout      | 300      |
         | innodb_flush_log_at_timeout | 1        |
         | innodb_lock_wait_timeout    | 120      |
         | innodb_rollback_on_timeout  | OFF      |
         | interactive_timeout         | 28800    |
         | lock_wait_timeout           | 31536000 |
         | net_read_timeout            | 30       |
         | net_write_timeout           | 60       |
         | slave_net_timeout           | 3600     |
         | wait_timeout                | 28800    |
         +-----------------------------+----------+
         11 rows in set (0.00 sec)
         
         mysql> set innodb_lock_wait_timeout=3600;
         Query OK, 0 rows affected (0.00 sec)