Hibernate with Oracle sequence doesn't use it

谁说我不能喝 提交于 2019-11-27 19:06:47

This is because the SequenceGenerator is not really a sequence generator. It's a sequence hi-lo generator. This means that the first time it's invoked, it gets the next value from the sequence (6 for example), then multiplies this value by 50 and gives you the result (300). The next time it's invoked, it returns 301 (without going to the sequence), and so on until it reaches 349. Then it asks the sequence for the next value and obtains 7, which it multiplies by 50 again to give you 350. My algorithm description could be off by one, but you get the idea.

If you stop and start your application, it will thus have gaps. But it's more efficient than a pure sequence generator because it only makes a database call once in 50 generations.

See http://docs.jboss.org/hibernate/core/3.6/reference/en-US/html_single/#mapping-declaration-id-enhanced-optimizers and http://docs.jboss.org/hibernate/core/3.6/reference/en-US/html_single/#mapping-declaration-id-generator for details.

I take it that your question is that the values of the ID column in the database are not a natural sequence, but why you are seeing gaps:

A bit of background:

  • Every time you call select HIBERNATE_SEQUENCE.nextval from DUAL the value of the sequence is increased.
  • As your sequence name is generic rather than specific to the table, if you've got multiple entities which all use the HIBERNATE_SEQUENCE as id generator, then the values from the sequences are used in all entities.
  • If some other application uses HIBERNATE_SEQUENCE, then the value is skipped as well.
  • As you are using CACHE=20, Oracle will grab sequence numbers in blocks of 20 and then use an internal cache to return the numbers. This can lead to numbers being skipped if the cache is lost (e.g. if the DB is shut down).
  • If rows are deleted from your database, the sequence value does not change

For example, consider the following scenario:

You've got two entities Entity1 and Entity2 using HIBERNATE_SEQUENCE as the ID generator:

  1. Current HIBERNATE_SEQUENCE value is 100
  2. An Entity1 is inserted (uses HIBERNATE_SEQUENCE which returns 101)
  3. An Entity2 is inserted (uses HIBERNATE_SEQUENCE which returns 102)
  4. An Entity2 is inserted (uses HIBERNATE_SEQUENCE which returns 103)
  5. The Entity2 with ID 103 is deleted
  6. You manually execute select HIBERNATE_SEQUENCE.nextval from DUAL (returns 104)
  7. An Entity1 is inserted (uses HIBERNATE_SEQUENCE which returns 105)
  8. An Entity2 is inserted (uses HIBERNATE_SEQUENCE which returns 106)

So at the end of it you'll have:

  • Entity1 with IDs (101, 105)
  • Entity2 with IDs (102, 106)

which explains the gaps.

EDIT:

Even if the @SequenceGenerator were setup to use the SequenceGenerator rather than the SequenceHiLoGenerator (as pointed out by JB Nizet, which I think is a better explanation for the gaps), gaps in IDs generated by sequences are a common occurrence.

Yogi
CREATE SEQUENCE SEQ_SEQUENCENAME INCREMENT BY 1 START WITH 1 MINVALUE 1;
grant all on SEQ_SEQUENCENAME to public;

@Id
@Column(name = "ID", unique = true, nullable = false)
@SequenceGenerator(name = "SequenceIdGenerator", sequenceName = "SEQ_SEQUENCENAME")
@GeneratedValue(strategy = GenerationType.SEQUENCE, generator = "SequenceIdGenerator")
private int Id;
易学教程内所有资源均来自网络或用户发布的内容,如有违反法律规定的内容欢迎反馈
该文章没有解决你所遇到的问题?点击提问,说说你的问题,让更多的人一起探讨吧!