I don't understand INSERT sets an exclusive lock on the inserted row. part of this document.

https://dev.mysql.com/doc/refman/8.0/en/innodb-locks-set.html

我不明白INSERT在插入的行上设置了排他锁。本文档的一部分。Https://dev.mysql.com/doc/refman/8.0/en/innodb-locks-set.html

In below part,

在下面的部分，

Session 1:
START TRANSACTION;
INSERT INTO t1 VALUES(1);

Session 2:
START TRANSACTION;
INSERT INTO t1 VALUES(1);

Session 3:
START TRANSACTION;
INSERT INTO t1 VALUES(1);

I thought they would go in the order below

我以为他们会按下面的顺序排

1. First transaction got exclusive lock for number 1 row.(didn't committed yet)


2. Second transaction also try to get exclusive lock.


3. Second transaction is going to wait, because First transaction already has exclusive lock.

But they didn't like above, the document said

但他们不喜欢上面的内容，文件说

The operations by sessions 2 and 3 both result in a duplicate-key error and they both request a shared lock for the row.

I don't understand why they got duplicate exception.

For a duplicate error to occur, the row must already be committed.

But they didn't. First transaction didn't commit yet.

我不明白为什么他们会有重复的例外。若要发生重复错误，该行必须已提交。但他们没有。第一笔交易还没有提交。

I suspect this is because before inserting an unique value (unique index), InnoDB has to do a search (secondary index scan), this will create next-key locks (next-key locks = record locks + gap locks).

我怀疑这是因为在插入唯一值(唯一索引)之前，InnoDB必须执行搜索(二级索引扫描)，这将创建下一键锁(下一键锁=记录锁+间隙锁)。

Because there is a gap lock, doing insertion will require transaction to accquire the Insert Intention Locks first. The purpose of this lock is only to allow different transactions to insert rows into a gap at the same time if the rows are not duplicate. If the rows are duplicate then duplicate-key error will be raised.

因为有空隙锁，所以做插入需要交易先获取插入意向锁。此锁的用途只是允许不同的事务在行不重复的情况下同时将行插入到间隙中。如果行重复，则会引发重复键错误。

In this example, transaction 1 accquired next-key locks from the minimum value to 1, then exclusive lock on 1 and inserted the row before committing. Transaction 2 and 3 accquired Insert Intention Locks as there is a next-key locks, and they realize their insertions are duplicate of each other.

在本例中，事务1获取从最小值到1的Next-key锁，然后在1上获得排他锁，并在提交之前插入行。事务2和事务3获得插入意向锁，因为存在下一键锁，并且它们意识到它们的插入是彼此重复的。