The best way to fix the Hibernate MultipleBagFetchException

Introduction

If you’ve been using Hibernate for some time, there is a good chance you bumped into a MultipleBagFetchException issue:

 org.hibernate.loader.MultipleBagFetchException: cannot simultaneously fetch multiple bags 

In this article, we are going to see the reason Hibernate throws the MultipleBagFetchException as well as the best way to solve this issue.

Domain Model

Let’s consider that our application defines three entities: Post , PostComment , and Tag , which are associated as in the following diagram:

What we are mostly interested in this article is that the Post entity defines a bidirectional @OneToMany association with the PostComment child entity, as well as a unidirectional @ManyToMany association with the Tag entity.

@OneToMany(
    mappedBy = "post", 
    cascade = CascadeType.ALL, 
    orphanRemoval = true
)
private List<PostComment> comments = new ArrayList<>();

@ManyToMany(
    cascade = {
        CascadeType.PERSIST, 
        CascadeType.MERGE
    }
)
@JoinTable(
    name = "post_tag",
    joinColumns = @JoinColumn(name = "post_id"),
    inverseJoinColumns = @JoinColumn(name = "tag_id")
)
private List<Tag> tags = new ArrayList<>();

The reason why the @ManyToMany association cascades only the PERSIST and MERGE entity state transitions and not the REMOVE one is because the other side is not a child entity.

Since the Tag entity lifecycle is not tied to the Post entity, cascading REMOVE or enabling the orphanRemoval mechanism would be a mistake. For more details about this topic, check outthis article.

Hibernate throwing MultipleBagFetchException

Now, if we want to fetch the Post entities with the identifier values between 1 and 50, along with all their associated PostComment and Tag entities, we would write a query like the following one:

List<Post> posts = entityManager
.createQuery(
    "select p " +
    "from Post p " +
    "left join fetch p.comments " +
    "left join fetch p.tags " +
    "where p.id between :minId and :maxId", Post.class)
.setParameter("minId", 1L)
.setParameter("maxId", 50L)
.getResultList();

However, when running the entity query above, Hibernate throws a MultipleBagFetchException while compiling the JPQL query:

org.hibernate.loader.MultipleBagFetchException: cannot simultaneously fetch multiple bags [
  com.vladmihalcea.book.hpjp.hibernate.fetching.Post.comments, 
  com.vladmihalcea.book.hpjp.hibernate.fetching.Post.tags
]

So, no SQL query is executed by Hibernate. The reason why a MultipleBagFetchException is thrown by Hibernate is that duplicates can occur, and the unordered List , which is called a bag in Hibernate terminology, is not supposed to remove duplicates.

How NOT to “fix” the Hibernate MultipleBagFetchException

If you google the MultipleBagFetchException , you are going to see many wrong answers, like this one on StackOverflow , which, surprisingly, has over 280 upvotes.

So simple, yet so wrong!

Using Set instead of List

So, let’s change the association collection type from List to Set :

@OneToMany(
    mappedBy = "post", 
    cascade = CascadeType.ALL, 
    orphanRemoval = true
)
private Set<PostComment> comments = new HashSet<>();

@ManyToMany(
    cascade = {
        CascadeType.PERSIST, 
        CascadeType.MERGE
    }
)
@JoinTable(
    name = "post_tag",
    joinColumns = @JoinColumn(name = "post_id"),
    inverseJoinColumns = @JoinColumn(name = "tag_id")
)
private Set<Tag> tags = new HashSet<>();

And, now, when rerunning the previous entity query which fetched some Post entities along with their comments and tags associations, we can see that no MultipleBagFetchException is thrown.

However, this is SQL query that Hibernate executed for the aforementioned JPQL query:

SELECT 
    p.id AS id1_0_0_,
    pc.id AS id1_1_1_,
    t.id AS id1_3_2_,
    p.title AS title2_0_0_,
    pc.post_id AS post_id3_1_1_,
    pc.review AS review2_1_1_,
    t.name AS name2_3_2_,
    pt.post_id AS post_id1_2_1__,
    pt.tag_id AS tag_id2_2_1__
FROM 
    post p
LEFT OUTER JOIN 
    post_comment pc ON p.id = pc.post_id
LEFT OUTER JOIN 
    post_tag pt ON p.id = pt.post_id
LEFT OUTER JOIN 
    tag t ON pt.tag_id = t.id
WHERE 
    p.id BETWEEN 1 AND 50

So, what’s wrong with this SQL query?

The post and post_comment are associated via the post_id Foreign Key column, so the join produces a result set containing all post table rows with the Primary Key values between 1 and 50 along with their associated post_comment table rows.

The post and tag tables are also associated via the post_id and tag_id post_tag Foreign Key columns, so these two joins produce a result set containing all post table rows with the Primary Key values between 1 and 50 along with their associated tag table rows.

Now, to merge the two result sets, the database can only use a Cartesian Product, so the final result set contains 50 post rows multiplied by the associated post_comment and tag table rows.

So, if we have 50 post rows associated with 20 post_comment and 10 tag rows, the final result set will contain 10_000 records (e.g., 50 x 20 x 10), as illustrated by the following test case:

List<Post> posts = entityManager.createQuery(
    "select p " +
    "from Post p " +
    "left join fetch p.comments " +
    "left join fetch p.tags " +
    "where p.id between :minId and :maxId", Post.class)
.setParameter("minId", 1L)
.setParameter("maxId", 50L)
.getResultList();

assertEquals(
    POST_COUNT * POST_COMMENT_COUNT * TAG_COUNT, 
    posts.size()
);

That’s so terrible from a performance perspective!

How to fix the Hibernate MultipleBagFetchException

To avoid a Cartesian Product, you can fetch at most one association at a time. So, instead of executing a single JPQL query that fetches two associations, we can execute two JPQL queries instead:

List<Post> posts = doInJPA(entityManager -> {
    List<Post> _posts = entityManager
    .createQuery(
        "select distinct p " +
        "from Post p " +
        "left join fetch p.comments " +
        "where p.id between :minId and :maxId ", Post.class)
    .setParameter("minId", 1L)
    .setParameter("maxId", 50L)
    .setHint(QueryHints.PASS_DISTINCT_THROUGH, false)
    .getResultList();

    entityManager
    .createQuery(
        "select distinct p " +
        "from Post p " +
        "left join fetch p.tags t " +
        "where p in :posts ", Post.class)
    .setParameter("posts", _posts)
    .setHint(QueryHints.PASS_DISTINCT_THROUGH, false)
    .getResultList();

    return _posts;
});

assertEquals(POST_COUNT, posts.size());

for(Post post : posts) {
    assertEquals(POST_COMMENT_COUNT, post.getComments().size());
    assertEquals(TAG_COUNT, post.getTags().size());
}

The first JPQL query defines the main filtering criteria and fetches the Post entities along with the associated PostComment records.

The PASS_DISTINCT_THROUGH query hint allows you to avoid passing the DISTINCT keyword to the SQL statement, and only use it to remove Java entity duplicates caused by the parent-child joined result set. For more details about the PASS_DISTINCT_THROUGH query hint, check outthis article.

Now, we have to fetch the Post entities along with their associated Tag entities, and, thanks to thePersistence Context, Hibernate will set the tags collection of the previously fetched Post entities.

Cool, right?

If you enjoyed this article, I bet you are going to love my Book and Video Courses as well.

Conclusion

There are so many blog posts, videos, books, and forum answers, providing the wrong solution to the MultipleBagFetchException Hibernate issues. All these resources tell you that using a Set instead of a List is the right way to avoid this exception.

However, the MultipleBagFetchException tells you that a Cartesian Product might be generated, and, most of the time, that’s undesirable when fetching entities as it can lead to terrible data access performance issues.

The best way to fetch multiple entity collections with JPA and Hibernate is to load at most one collection at a time while relying on the Hibernate Persistence Context guarantee that only a single entity object can be loading at a time in a given JPA EntityManager or Hibernate Session .