Adding support for letting Elasticsearch generate Id for document [DATAES-42]

[spring-projects-issues]

Mohsin Husen opened DATAES-42 and commented

Adding support for letting Elasticsearch generate Id for document

---

Referenced from: commits 08a6a9b, fb95bb0

6 votes, 8 watchers

[spring-projects-issues]

Daniel Stoch commented

Does mapped entity object should also has id generated? Because it looks like entity id is still null when deserialization, in ES document _source.id=null. ElasticsearchTemplate.index() calls setPersistentEntityId on this entity, but it is not stored in ES.

For example when saving a Book object (from tests):

public class Book {
  @Id
  private String id;
  private String name;
...
}

Book book = new Book();
book.setName("xyz");
bookRepository.save(book);

Document in ES looks like:

{
    _index: book
    _type: book
    _id: XuriiXpDRei2Z9rRQ84b5w
    _version: 1
    _score: 1
    _source: {
        id: null
        name: xyz
...

[spring-projects-issues]

Mohsin Husen commented

Daniel,

Thanks for reporting this.
One of the problem storing the elasticsearch generated Id will be second round trip to update the document once elasticsearch index and return the actual documentId.

I will have a look at spring data mongodb as it does something similar to update mongodb generated Id, meanwhile if you have some hint please share

Mohsin

[spring-projects-issues]

Daniel Stoch commented

If we want to avoid the second round trip to update Id, maybe we should generate this Id by ourselves before saving it to ES? Is there any advantage with using ES documentId as an entity Id, is it necessary?

[spring-projects-issues]

Mohsin Husen commented

That's valid point, we can generate unique Id and give it back to elasticsearch.
Reopening this card, will have a look

Thanks
Mohsin

[spring-projects-issues]

Jan Durian commented

I vote for increasing a priority of this one.

Thanks.
Jan

[spring-projects-issues]

Shubham Aggarwal commented

Looks like the id is still not generated. Is there any auto-generation strategy added?

[spring-projects-issues]

Gregory Czerwinski commented

It would be really great solution. We have many indexes where we do have many datasources per index

[spring-projects-issues]

Florin Popa commented

Another vote for adding this functionality as quickly as possible and increasing the Priority for this task.

Thank you,
Florin

[spring-projects-issues]

sothawo commented

With the release of version 4.0 this now can be easily handled in the users application by defining a custom BeforeConvertCallback that sets an id if necessary. For the given Book class this would be (using Elasticsearch's UUIDs class):

@Component
public class BookBeforeConvertCallback implements BeforeConvertCallback<Book> {

    @Override
    public Book onBeforeConvert(Book book, IndexCoordinates indexCoordinates) {

        if (book.getId() == null) {
            book.setId(UUIDs.randomBase64UUID());
        }

        return book;
    }
}