2L-Querycache ImmutableNaturalKeyLookup not properly recognized as hasImmutableNaturalId() is called on wrong EntityMetamodel


If a entityclass with immutable naturalId has non-lazy properties (for instance a OneToOne association),
then more entityPersisters are involved in the query as it becomes a join-query eager loading the associated entity.

The method Loader#getResultFromQueryCache presumes that the first entityPersister in the array is the main entity class.

boolean isImmutableNaturalKeyLookup = queryParameters.isNaturalKeyLookup()
&& getEntityPersisters()[0].getEntityMetamodel().hasImmutableNaturalId();

Debugging attached testcase you can see indeed that getEntityPersisters()[0] returns SingleTableEntityPersister(hello.D)
which is the persister of the associated entity, not the main one!
Consequently the query is not considered as isImmutableNaturalKeyLookup, despite fullfilling all conditions.

Here my output (with enabled p6spy):
11:53:15,651 INFO Version:37 - Hibernate Annotations 3.5.0.Beta1
11:53:15,690 INFO Environment:563 - Hibernate 3.5.0-Beta-3
11:53:15,694 INFO Environment:596 - hibernate.properties not found
11:53:15,700 INFO Environment:774 - Bytecode provider name : javassist
11:53:15,706 INFO Environment:655 - using JDK 1.4 java.sql.Timestamp handling
11:53:15,863 INFO Version:38 - Hibernate EntityManager 3.5.0.Beta1
11:53:16,356 INFO AnnotationBinder:446 - Binding entity from annotated class: hello.A
11:53:16,424 INFO EntityBinder:471 - Bind entity hello.A on table A
11:53:16,526 INFO AnnotationBinder:446 - Binding entity from annotated class: hello.D
11:53:16,527 INFO EntityBinder:471 - Bind entity hello.D on table D
11:53:16,686 WARN Ejb3Configuration:943 - hibernate.connection.autocommit = false break the EJB3 specification
11:53:16,702 INFO HibernateSearchEventListenerRegister:75 - Unable to find org.hibernate.search.event.FullTextIndexEventListener on the classpath. Hibernate Search is not enabled.
11:53:16,718 INFO DriverManagerConnectionProvider:64 - Using Hibernate built-in connection pool (not for production use!)
11:53:16,719 INFO DriverManagerConnectionProvider:65 - Hibernate connection pool size: 20
11:53:16,720 INFO DriverManagerConnectionProvider:68 - autocommit mode: false
11:53:16,921 INFO DriverManagerConnectionProvider:103 - using driver: com.p6spy.engine.spy.P6SpyDriver at URL: jdbc:hsqldb:hsql://localhost
11:53:16,922 INFO DriverManagerConnectionProvider:109 - connection properties: {user=sa, password=****, autocommit=false, release_mode=after_transaction}
11:53:17,002 INFO SettingsFactory:117 - RDBMS: HSQL Database Engine, version: 1.8.0
11:53:17,004 INFO SettingsFactory:118 - JDBC driver: HSQL Database Engine Driver, version: 1.8.0
11:53:17,068 INFO Dialect:223 - Using dialect: org.hibernate.dialect.HSQLDialect
11:53:17,082 INFO JdbcSupportLoader:79 - Disabling contextual LOB creation as JDBC driver reported JDBC version [3] less than 4
11:53:17,085 INFO TransactionFactoryFactory:62 - Transaction strategy: org.hibernate.transaction.JDBCTransactionFactory
11:53:17,089 INFO TransactionManagerLookupFactory:80 - No TransactionManagerLookup configured (in JTA environment, use of read-write or transactional second-level cache is not recommended)
11:53:17,090 INFO SettingsFactory:169 - Automatic flush during beforeCompletion(): disabled
11:53:17,091 INFO SettingsFactory:173 - Automatic session close at end of transaction: disabled
11:53:17,097 INFO SettingsFactory:188 - Scrollable result sets: enabled
11:53:17,098 INFO SettingsFactory:196 - JDBC3 getGeneratedKeys(): disabled
11:53:17,099 INFO SettingsFactory:204 - Connection release mode: after_transaction
11:53:17,101 INFO SettingsFactory:231 - Default batch fetch size: 1
11:53:17,103 INFO SettingsFactory:235 - Generate SQL with comments: disabled
11:53:17,104 INFO SettingsFactory:239 - Order SQL updates by primary key: disabled
11:53:17,105 INFO SettingsFactory:243 - Order SQL inserts for batching: disabled
11:53:17,107 INFO SettingsFactory:410 - Query translator: org.hibernate.hql.ast.ASTQueryTranslatorFactory
11:53:17,114 INFO SettingsFactory:251 - Query language substitutions: {}
11:53:17,115 INFO SettingsFactory:256 - JPA-QL strict compliance: enabled
11:53:17,121 INFO SettingsFactory:261 - Second-level cache: enabled
11:53:17,130 INFO SettingsFactory:265 - Query cache: enabled
11:53:17,143 INFO SettingsFactory:395 - Cache region factory : org.hibernate.cache.impl.bridge.RegionFactoryCacheProviderBridge
11:53:17,144 INFO RegionFactoryCacheProviderBridge:61 - Cache provider: net.sf.ehcache.hibernate.SingletonEhCacheProvider
11:53:17,150 INFO SettingsFactory:275 - Optimize cache for minimal puts: disabled
11:53:17,151 INFO SettingsFactory:284 - Structured second-level cache entries: disabled
11:53:17,151 INFO SettingsFactory:374 - Query cache factory: org.hibernate.cache.StandardQueryCacheFactory
11:53:17,165 INFO SettingsFactory:313 - Statistics: disabled
11:53:17,166 INFO SettingsFactory:317 - Deleted entity synthetic identifier rollback: disabled
11:53:17,166 INFO SettingsFactory:332 - Default entity-mode: pojo
11:53:17,167 INFO SettingsFactory:336 - Named query checking : enabled
11:53:17,167 INFO SettingsFactory:340 - Check Nullability in Core (should be disabled when Bean Validation is on): disabled
11:53:17,240 INFO SessionFactoryImpl:197 - building session factory
11:53:17,260 WARN ConfigurationFactory:131 - No configuration found. Configuring ehcache from ehcache-failsafe.xml found in the classpath: jar:file:/home/pb00067/Desktop/ehcache-1.7.1/lib/ehcache-core-1.7.1.jar!/ehcache-failsafe.xml
11:53:17,361 WARN SingletonEhCacheProvider:92 - Could not find a specific ehcache configuration for cache named [hello.A]; using defaults.
11:53:17,566 WARN SingletonEhCacheProvider:92 - Could not find a specific ehcache configuration for cache named [hello.D]; using defaults.
11:53:17,661 INFO SchemaExport:226 - Running hbm2ddl schema export
11:53:17,663 INFO SchemaExport:251 - exporting generated schema to database
p6spy - 1264416797664|0|0|commit||
p6spy - 1264416797670|1|0|statement

alter table A drop constraint FK41F7BDC12E
p6spy - 1264416797672|0|0|statement

drop table A if exists
p6spy - 1264416797673|0|0|statement

drop table D if exists
p6spy - 1264416797674|1|0|statement

drop table hibernate_sequences if exists
p6spy - 1264416797675|1|0|statement

create table A (oid bigint not null, name varchar(255), version integer not null, assSingleD_oid bigint, primary key (oid), unique (name))
p6spy - 1264416797676|0|0|statement

create table D (oid bigint not null, version integer not null, primary key (oid))
p6spy - 1264416797678|1|0|statement

alter table A add constraint FK41F7BDC12E foreign key (assSingleD_oid) references D
p6spy - 1264416797679|1|0|statement

create table hibernate_sequences ( sequence_name varchar(255), sequence_next_hi_value integer )
11:53:17,679 INFO SchemaExport:268 - schema export complete
11:53:17,684 INFO UpdateTimestampsCache:57 - starting update timestamps cache at region: org.hibernate.cache.UpdateTimestampsCache
11:53:17,686 WARN SingletonEhCacheProvider:92 - Could not find a specific ehcache configuration for cache named [org.hibernate.cache.UpdateTimestampsCache]; using defaults.
11:53:17,689 INFO StandardQueryCache:75 - starting query cache at region: org.hibernate.cache.StandardQueryCache
11:53:17,690 WARN SingletonEhCacheProvider:92 - Could not find a specific ehcache configuration for cache named [org.hibernate.cache.StandardQueryCache]; using defaults.
11:53:17,841 DEBUG ConnectionManager:444 - opening JDBC connection
p6spy - 1264416797874|6|1|statement

sequence_name = 'A'
p6spy - 1264416797876|0|1|statement

(sequence_name, sequence_next_hi_value)
('A', 0)
p6spy - 1264416797878|0|1|statement

sequence_next_hi_value = 1
sequence_next_hi_value = 0
and sequence_name = 'A'
p6spy - 1264416797879|0|1|commit||
p6spy - 1264416797915|0|0|statement

(assSingleD_oid, name, version, oid)
('', 'name1', 0, 1)
11:53:17,919 DEBUG UpdateTimestampsCache:66 - Pre-invalidating space [A]
p6spy - 1264416797931|0|0|commit||
11:53:17,934 DEBUG UpdateTimestampsCache:81 - Invalidating space [A], timestamp: 5179051204337665
11:53:17,974 DEBUG StandardQueryCache:127 - checking cached query results in region: org.hibernate.cache.StandardQueryCache
11:53:17,975 DEBUG StandardQueryCache:132 - query results were not found in cache
p6spy - 1264416797978|0|0|statement

this_.oid as oid0_1_,
this_.assSingleD_oid as assSingleD4_0_1_,
this_.name as name0_1_,
this_.version as version0_1_,
d2_.oid as oid1_0_,
d2_.version as version1_0_
A this_
left outer join
D d2_
on this_.assSingleD_oid=d2_.oid
p6spy - 1264416797981|-1||resultset|oid=1, oid=null
11:53:17,985 DEBUG StandardQueryCache:94 - caching query results in region: org.hibernate.cache.StandardQueryCache; timestamp=5179051204546560
11:53:17,992 DEBUG StandardQueryCache:127 - checking cached query results in region: org.hibernate.cache.StandardQueryCache
11:53:17,993 DEBUG StandardQueryCache:185 - Checking query spaces for up-to-dateness: [A]
11:53:17,994 DEBUG UpdateTimestampsCache:102 - [A] last update timestamp: 5179051204337665, result set timestamp: 5179051204546560
11:53:17,994 DEBUG StandardQueryCache:142 - returning cached query results
11:53:17,997 INFO SessionFactoryImpl:908 - closing


3.5.0-Beta3, HSQLDB, EHCache as 2L cache implementation


December 2, 2010, 5:23 PM

if you proved that criteria.add(Restrictions.naturalId().set("otherAttribute", "someValue"));
does not throw any exception although "otherAttribute" is'nt declared naturalId at all,
then this is in fact a weakness in the robustness of hibernate (if a user uses hibernate in the wrong way!).
But IMHO this has nothing to with the actual problem I described here.
So I think your objection is correct, but should maybe dealt with a separate issue.

I tried doing that without having marked the attribute as NaturalId and it didn't complain. I thought it was a problem with my version of Hibernate (core and annotations) as described here but it wasn't. The problem seems to be the current opened defect (I was doing a Criteria query specifying FetchMode.JOIN for some relationships. That's how I was taken to this Jira issue.)

I don't have a way to prove it but to tell you what I did ... Shall I create a new Jira for this? As you said, this might be a different issue but it's still an issue.

>>@Guenther -> the proposed patch wouldn't work if the criteria has been created with a different alias, right?
What do you mean exactly with "different alias"?

If I create a query as follows:

and the patch does something like:

private boolean hasImmutableNaturalId() {
int i=0;
for (String alias : getAliases()){
if (alias.equals("this_")) {
return getEntityPersisters()[i].getEntityMetamodel().hasImmutableNaturalId();
return false;\\}

then you would be comparing against — this_ when the real alias is — myNewAlias

Am I right? I might have misunderstood your suggestion ...


Guenther Demetz
December 3, 2010, 10:03 AM

Hi Nicolas,

unfortunately currently I have not the time to test and usually I use not aliases, anyway
I strongly believe your analysis regarding the alias is correct and
consequently my proposed patch has to be enhanced in way that it takes also aliases into consideration.

For the problem with the 'otherAttribute' as misplaced NaturalId indeed
I still think it should be resolved in a separate issue, as it is an issue with minor priority.

Thanks for your attention and effort.

December 3, 2010, 1:33 PM

Hi again,

I found another problem related to this same issue.

When creating a criteria query as follows:

I would be creating a criteria query specifying

  • only one natural id restriction (— Restrictions.naturalId().set("myNaturalIdAttribute", "myValue"))

  • Join fetch mode for the collection of clients (by default is lazily loaded)

  • Join fetch mode for the collection of items of the clients (that's why I create a sub criteria)(by default is lazily loaded)

Given this situation, I expected that — queryParameters.isNaturalKeyLookup() from the below code would return true, but it doesn't.

I digged into the code, until I found this other part:


And given that my criteria has one — subcriteriaList, the code returns false which is not what it should be returning.

I understand that this code also would need to be changed as part of the current Jira issue as it is related to it.

Please let me know if I can contribute somehow to fix this issue.


P.S.1: Your proposed patch would still work (for situations not like the one I've just mentioned though ...) as Hibernate holds the user defined aliases in an attribute called userAliases in the Loader class and the automatically generated ones in aliases

P.S.2: To make your patch more robust you could find the following analysis useful:

The EntityMetaModel has these attributes (among others):

  • StandardProperty[] properties

  • int[] naturalIdPropertyNumbers
    From naturalIdPropertyNumbers you can get the index in properties and from the referenced StandardProperty you could get the name of that natural id attribute which you could use to compare it against the one you received. This is also quite dodgy as that information is hard to get by this time. I found it anyway in Loader#translator#rootCriteria#criterionEntries#... and so on ...

Gail Badner
October 4, 2011, 12:12 AM

I've committed the fix to master and 3.6.

I forgot to prefix the master commit message with "HHH-4838". See commit b307a801a748eed42810e10d7e413f185953fe43 for details: https://github.com/hibernate/hibernate-core/commit/b307a801a748eed42810e10d7e413f185953fe43

This fixes looking up an entity by an immutable natural ID from the query cache when that entity has non-lazy associations.

The lookup by natural ID optimization is not used for a Criteria if it has projections, subcriteria, or any restrictions (other than Restrictions.naturalId()).

Brett Meyer
March 7, 2014, 10:08 PM

Bulk closing tickets resolved in released versions



Gail Badner


Guenther Demetz

Fix versions





Suitable for new contributors


Requires Release Note


Pull Request




Affects versions