I'm getting a warning in the Server log "firstResult/maxResults specified with collection fetch; applying in memory!". However everything working fine. But I don't want this warning.
public employee find(int id) {
return (employee) getEntityManager().createQuery(QUERY).setParameter("id", id).getSingleResult();
}
QUERY = "from employee as emp left join fetch emp.salary left join fetch emp.department where emp.id = :id"
Reason for this warning is that when fetch join is used, order in result sets is defined only by ID of selected entity (and not by join fetched).
If this sorting in memory is causing problems, do not use firsResult/maxResults with JOIN FETCH.
To avoid this WARNING you have to change the call getSingleResult to
getResultList().get(0)
Although you are getting valid results, the SQL query fetches all data and it's not as efficient as it should.
As I explained in this article, you have two options.
The easiest way to fix this issue is to execute two queries:
. The first query will fetch the root entity identifiers matching the provided filtering criteria. . The second query will use the previously extracted root entity identifiers to fetch the parent and the child entities.
This approach is very easy to implement and looks as follows:
List<Long> postIds = entityManager
.createQuery(
"select p.id " +
"from Post p " +
"where p.title like :titlePattern " +
"order by p.createdOn", Long.class)
.setParameter(
"titlePattern",
"High-Performance Java Persistence %"
)
.setMaxResults(5)
.getResultList();
List<Post> posts = entityManager
.createQuery(
"select distinct p " +
"from Post p " +
"left join fetch p.comments " +
"where p.id in (:postIds) " +
"order by p.createdOn", Post.class)
.setParameter("postIds", postIds)
.setHint(
"hibernate.query.passDistinctThrough",
false
)
.getResultList();
The second approach is to use SDENSE_RANK over the result set of parent and child entities that match our filtering criteria and restrict the output for the first N post entries only.
The SQL query can look as follows:
@NamedNativeQuery(
name = "PostWithCommentByRank",
query =
"SELECT * " +
"FROM ( " +
" SELECT *, dense_rank() OVER (ORDER BY \"p.created_on\", \"p.id\") rank " +
" FROM ( " +
" SELECT p.id AS \"p.id\", " +
" p.created_on AS \"p.created_on\", " +
" p.title AS \"p.title\", " +
" pc.id as \"pc.id\", " +
" pc.created_on AS \"pc.created_on\", " +
" pc.review AS \"pc.review\", " +
" pc.post_id AS \"pc.post_id\" " +
" FROM post p " +
" LEFT JOIN post_comment pc ON p.id = pc.post_id " +
" WHERE p.title LIKE :titlePattern " +
" ORDER BY p.created_on " +
" ) p_pc " +
") p_pc_r " +
"WHERE p_pc_r.rank <= :rank ",
resultSetMapping = "PostWithCommentByRankMapping"
)
@SqlResultSetMapping(
name = "PostWithCommentByRankMapping",
entities = {
@EntityResult(
entityClass = Post.class,
fields = {
@FieldResult(name = "id", column = "p.id"),
@FieldResult(name = "createdOn", column = "p.created_on"),
@FieldResult(name = "title", column = "p.title"),
}
),
@EntityResult(
entityClass = PostComment.class,
fields = {
@FieldResult(name = "id", column = "pc.id"),
@FieldResult(name = "createdOn", column = "pc.created_on"),
@FieldResult(name = "review", column = "pc.review"),
@FieldResult(name = "post", column = "pc.post_id"),
}
)
}
)
The @NamedNativeQuery fetches all Post entities matching the provided title along with their associated PostComment child entities. The DENSE_RANK Window Function is used to assign the rank for each Post and PostComment joined record so that we can later filter just the amount of Post records we are interested in fetching.
The SqlResultSetMapping provides the mapping between the SQL-level column aliases and the JPA entity properties that need to be populated.
Now, we can execute the PostWithCommentByRank @NamedNativeQuery like this:
List<Post> posts = entityManager
.createNamedQuery("PostWithCommentByRank")
.setParameter(
"titlePattern",
"High-Performance Java Persistence %"
)
.setParameter(
"rank",
5
)
.unwrap(NativeQuery.class)
.setResultTransformer(
new DistinctPostResultTransformer(entityManager)
)
.getResultList();
Now, by default, a native SQL query like the PostWithCommentByRank one would fetch the Post and the PostComment in the same JDBC row, so we will end up with an Object[] containing both entities.
However, we want to transform the tabular Object[] array into a tree of parent-child entities, and for this reason, we need to use the Hibernate ResultTransformer For more details about the ResultTransformer, check out this article.
The DistinctPostResultTransformer looks as follows:
public class DistinctPostResultTransformer
extends BasicTransformerAdapter {
private final EntityManager entityManager;
public DistinctPostResultTransformer(
EntityManager entityManager) {
this.entityManager = entityManager;
}
@Override
public List transformList(
List list) {
Map<Serializable, Identifiable> identifiableMap =
new LinkedHashMap<>(list.size());
for (Object entityArray : list) {
if (Object[].class.isAssignableFrom(entityArray.getClass())) {
Post post = null;
PostComment comment = null;
Object[] tuples = (Object[]) entityArray;
for (Object tuple : tuples) {
if(tuple instanceof Identifiable) {
entityManager.detach(tuple);
if (tuple instanceof Post) {
post = (Post) tuple;
}
else if (tuple instanceof PostComment) {
comment = (PostComment) tuple;
}
else {
throw new UnsupportedOperationException(
"Tuple " + tuple.getClass() + " is not supported!"
);
}
}
}
if (post != null) {
if (!identifiableMap.containsKey(post.getId())) {
identifiableMap.put(post.getId(), post);
post.setComments(new ArrayList<>());
}
if (comment != null) {
post.addComment(comment);
}
}
}
}
return new ArrayList<>(identifiableMap.values());
}
}
The DistinctPostResultTransformer must detach the entities being fetched because we are overwriting the child collection and we don’t want that to be propagated as an entity state transition:
post.setComments(new ArrayList<>());
Fore more details, check out this article.
This warning tells you Hibernate is performing in memory java pagination. This can cause high JVM memory consumption. Since a developer can miss this warning, I contributed to Hibernate by adding a flag allowing to throw an exception instead of logging the warning (https://hibernate.atlassian.net/browse/HHH-9965).
The flag is hibernate.query.fail_on_pagination_over_collection_fetch.
I recommend everyone to enable it.
The flag is defined in org.hibernate.cfg.AvailableSettings :
/**
* Raises an exception when in-memory pagination over collection fetch is about to be performed.
* Disabled by default. Set to true to enable.
*
* @since 5.2.13
*/
String FAIL_ON_PAGINATION_OVER_COLLECTION_FETCH = "hibernate.query.fail_on_pagination_over_collection_fetch";
the problem is you will get cartesian product doing JOIN. The offset will cut your recordset without looking if you are still on same root identity class
I guess the emp has many departments which is a One to Many relationship. Hibernate will fetch many rows for this query with fetched department records. So the order of result set can not be decided until it has really fetch the results to the memory. So the pagination will be done in memory.
If you do not want to fetch the departments with emp, but still want to do some query based on the department, you can achieve the result with out warning (without doing ordering in the memory). For that simply you have to remove the "fetch" clause. So something like as follows:
QUERY = "from employee as emp left join emp.salary sal left join emp.department dep where emp.id = :id and dep.name = 'testing' and sal.salary > 5000 "
