先说我我的解决方法:找到锁住的线程然后kill掉。
mysql> kill thr_id;
下面简单分析一下到底应该kill哪个线程。
show processlist查看所有运行的线程,发现所有线程的状态都是sleep,不清楚这么多sleep的线程,INNODB_TRX表里也没有对应RUNNING的记录,那些线程假死了?
mysql> show full processlist;
+------+------+---------------------+----------+---------+-------+-------+-----------------------+
| Id | User | Host | db | Command | Time | State | Info |
+------+------+---------------------+----------+---------+-------+-------+-----------------------+
| 1172 | root | 112.124.64.64:43083 | eye | Sleep | 12384 | | NULL |
| 1179 | root | 112.124.64.64:43086 | eye | Sleep | 596 | | NULL |
| 1181 | root | 112.124.64.64:43087 | eye | Sleep | 12467 | | NULL |
| 1182 | root | 112.124.64.64:43088 | eye | Sleep | 149 | | NULL |
| 1183 | root | 112.124.64.64:43089 | eye | Sleep | 612 | | NULL |
| 1184 | root | 112.124.64.64:43090 | eye | Sleep | 39 | | NULL |
| 1197 | root | 112.124.64.64:43092 | eye | Sleep | 148 | | NULL |
| 1198 | root | 112.124.64.64:43093 | eye | Sleep | 5415 | | NULL |
| 1200 | root | 112.124.64.64:43095 | eye | Sleep | 581 | | NULL |
| 1289 | root | 58.249.112.63:5657 | eye | Sleep | 1496 | | NULL |
| 1325 | root | localhost | NULL | Query | 0 | NULL | show full processlist |
+------+------+---------------------+----------+---------+-------+-------+-----------------------+
查看information_schema库中的INNODB_TRX表发现有几条记录,时间和用户反馈的时间是对上号的,id号在上结果中也有显示,表列太多,截图没截全:
show engine innodb status发现有以下几行:
---TRANSACTION 17E34B, ACTIVE 3296 sec
2 lock struct(s), heap size 376, 0 row lock(s), undo log entries 15
MySQL thread id 1196, OS thread handle 0x7f1f0759d700, query id 140864 112.124.64.64 root
Trx read view will not see trx with id >= 17E34C, sees < 17DFA9
---TRANSACTION 17E11F, ACTIVE 6222 sec
10 lock struct(s), heap size 3112, 9 row lock(s), undo log entries 11
MySQL thread id 1199, OS thread handle 0x7f1f07458700, query id 140844 112.124.64.64 root
Trx read view will not see trx with id >= 17E120, sees < 17DFA9
---TRANSACTION 17DFAA, ACTIVE 11653 sec
2 lock struct(s), heap size 376, 0 row lock(s), undo log entries 15
MySQL thread id 1178, OS thread handle 0x7f1f0af75700, query id 133760 112.124.64.64 root
Trx read view will not see trx with id >= 17DFAB, sees < 17DFA7
---TRANSACTION 17DFA9, ACTIVE 11655 sec
3 lock struct(s), heap size 376, 1 row lock(s), undo log entries 15
MySQL thread id 1177, OS thread handle 0x7f1f1b682700, query id 133743 112.124.64.64 root
Trx read view will not see trx with id >= 17DFAA, sees < 17DFA7
(TRANSACTION 17DFAA,2 lock struct(s), 0 row lock(s),0 row locks,那 2 lock struct(s)到底锁在了啥上面)【当事务超时之后show engine innodb status就不会显示此事务运行时的执行SQL,以上信息只能说明这个事务锁等待过,2016-08-20】,(不是特别理解"MySQL thread id 1196, OS thread handle 0x7f1f0759d700, query id 140864 112.124.64.64 root Trx read view will not see trx with id >= 17E34C, sees < 17DFA9",有什么提示信息,难道trx_id还必须连续?")【InnoDB使用MVCC来减少锁的使用和开销,允许非阻塞读,写操作时只锁定必要的记录,MVCC工作在READ-COMMITTED,REAPEATABLE-READ隔离级别,通过TRX_ID并发控制,2016-08-20】(undo log entries 15"的意思是15条没有insert进去,还是回滚操作没有回滚回来?)【undo log记录的是没有更改之前的数据,回滚用,这里应该是回滚了的条数,2016-08-20】。
先来尝试还原当时的情景,插入一条数据,因为未知原因事务一直未提交,获得的锁当然未释放,然后再去操作同一条数据两次,查询INNODB_TRX表,发现出现三条数据,前两条是后来更新操作的信息,trx_state字段为LOCK WAIT,第三个是没有提交事务的那条操作:
mysql> select * from INNODB_TRX;
+--------+-----------+---------------------+-----------------------+---------------------+------------+---------------------+---------------------------------------------------------------------------------------------------------------------------------------------+---------------------+-------------------+-------------------+------------------+-----------------------+-----------------+-------------------+-------------------------+---------------------+-------------------+------------------------+----------------------------+---------------------------+---------------------------+
| trx_id | trx_state | trx_started | trx_requested_lock_id | trx_wait_started | trx_weight | trx_mysql_thread_id | trx_query | trx_operation_state | trx_tables_in_use | trx_tables_locked | trx_lock_structs | trx_lock_memory_bytes | trx_rows_locked | trx_rows_modified | trx_concurrency_tickets | trx_isolation_level | trx_unique_checks | trx_foreign_key_checks | trx_last_foreign_key_error | trx_adaptive_hash_latched | trx_adaptive_hash_timeout |
+--------+-----------+---------------------+-----------------------+---------------------+------------+---------------------+---------------------------------------------------------------------------------------------------------------------------------------------+---------------------+-------------------+-------------------+------------------+-----------------------+-----------------+-------------------+-------------------------+---------------------+-------------------+------------------------+----------------------------+---------------------------+---------------------------+
| 1852B9 | LOCK WAIT | 2016-05-23 21:09:24 | 1852B9:0:3611:51 | 2016-05-23 21:09:25 | 2 | 7920 | UPDATE inspectionhistory SET evaluation = null, photoName = null WHERE inspectionID = '000154DD83C3313FA051' | starting index read | 1 | 1 | 2 | 376 | 1 | 0 | 0 | REPEATABLE READ | 1 | 1 | NULL | 0 | 9997 |
| 1852A8 | LOCK WAIT | 2016-05-23 21:09:13 | 1852A8:0:3611:51 | 2016-05-23 21:09:13 | 2 | 7861 | UPDATE inspectionhistory SET evaluation = null, photoName = null WHERE inspectionID = '000154DD83C3313FA051' | starting index read | 1 | 1 | 2 | 376 | 1 | 0 | 0 | REPEATABLE READ | 1 | 1 | NULL | 0 | 9997 |
| 18529F | RUNNING | 2016-05-23 21:09:06 | NULL | NULL | 31 | 7863 | NULL | NULL | 0 | 0 | 14 | 3112 | 17 | 17 | 0 | REPEATABLE READ | 1 | 1 | NULL | 0 | 9982 |
+--------+-----------+---------------------+-----------------------+---------------------+------------+---------------------+---------------------------------------------------------------------------------------------------------------------------------------------+---------------------+-------------------+-------------------+------------------+-----------------------+-----------------+-------------------+-------------------------+---------------------+-------------------+------------------------+----------------------------+---------------------------+---------------------------+
上表里前两个事务等待超时返回错误信息,然后INNODB_TRX表里记录自动删除,剩下未提交事务的那条记录,如下:
--- The error occurred in mit/datamodel/ibatis/inspectionhistory.xml.
--- The error occurred while applying a parameter map.
--- Check the updateInspectionHistory-InlineParameterMap.
--- Check the statement (update failed).
--- Cause: java.sql.SQLException: Lock wait timeout exceeded; try restarting transaction
+--------+-----------+---------------------+-----------------------+------------------+------------+---------------------+-----------+---------------------+-------------------+-------------------+------------------+-----------------------+-----------------+-------------------+-------------------------+---------------------+-------------------+------------------------+----------------------------+---------------------------+---------------------------+
| trx_id | trx_state | trx_started | trx_requested_lock_id | trx_wait_started | trx_weight | trx_mysql_thread_id | trx_query | trx_operation_state | trx_tables_in_use | trx_tables_locked | trx_lock_structs | trx_lock_memory_bytes | trx_rows_locked | trx_rows_modified | trx_concurrency_tickets | trx_isolation_level | trx_unique_checks | trx_foreign_key_checks | trx_last_foreign_key_error | trx_adaptive_hash_latched | trx_adaptive_hash_timeout |
+--------+-----------+---------------------+-----------------------+------------------+------------+---------------------+-----------+---------------------+-------------------+-------------------+------------------+-----------------------+-----------------+-------------------+-------------------------+---------------------+-------------------+------------------------+----------------------------+---------------------------+---------------------------+
| 18529F | RUNNING | 2016-05-23 21:09:06 | NULL | NULL | 31 | 7863 | NULL | NULL | 0 | 0 | 14 | 3112 | 17 | 17 | 0 | REPEATABLE READ | 1 | 1 | NULL | 0 | 9982 |
+--------+-----------+---------------------+-----------------------+------------------+------------+---------------------+-----------+---------------------+-------------------+-------------------+------------------+-----------------------+-----------------+-------------------+-------------------------+---------------------+-------------------+------------------------+----------------------------+---------------------------+---------------------------+
当提交事务时,INNODB_TRX表里无记录。
mysql> select * from INNODB_TRX;
Empty set (0.00 sec)
如果真是这样,那么问题来了,(为什么上面截图里面5个事务一直RUNNING状态,什么导致了它们不去提交事务。)【什么导致了5个事务一直RUNNING现在还不清原因,也还原不了了,但是前段时间遇到一种这样的情况,在一次事务没有提交时,我更改了这张表的索引字段,导致卡死,之后强制关闭命令行,再重新打开去show processlist,发现更改索引的那个事务一直RUNNING,2016-08-20】(以至于后来查询其它数据也等待超时,难道是它们之间及到了X,S,GAP锁?或者因为 Trx read view will not see?)【确实是因为没有提交的事务锁定了某些行数据,如果字段没有建索引,会锁住整张表数据,2016-08-20】
ps:无聊时随手看了下INNIDB_TRX表信息,它的Engine竟然是Memory,其实也该想到了,只是经验太少,从没有实践机会去用Memory引擎。
+------------+--------+---------+------------+------+----------------+-------------+-----------------+--------------+-----------+----------------+---------------------+-------------+------------+-----------------+----------+----------------+---------+
| Name | Engine | Version | Row_format | Rows | Avg_row_length | Data_length | Max_data_length | Index_length | Data_free | Auto_increment | Create_time | Update_time | Check_time | Collation | Checksum | Create_options | Comment |
+------------+--------+---------+------------+------+----------------+-------------+-----------------+--------------+-----------+----------------+---------------------+-------------+------------+-----------------+----------+----------------+---------+
| INNODB_TRX | MEMORY | 10 | Fixed | NULL | 4534 | 0 | 16766732 | 0 | 0 | NULL | 2016-05-23 22:06:37 | NULL | NULL | utf8_general_ci | NULL | max_rows=3700 | |
+------------+--------+---------+------------+------+----------------+-------------+-----------------+--------------+-----------+----------------+---------------------+-------------+------------+-----------------+----------+----------------+---------+