I've a situation where each new record should contain a unique and readable value.
This value has a business meaning for the user and will be handled as a natural id (next to primary key) in our database.
To give you an idea of the value's structure:
- record 1 has business value 'INVOICE_AAA_001'
- record 2 has business value 'INVOICE_AAA_002'
...
- record 999 has business value 'INVOICE_AAA_999'
- record 1000 has business value 'INVOICE_BAA_001'
- record 1001 has business value 'INVOICE_BAA_002'
...
This business value is created by a factory:
class BusinessFactory {
...
public String createUniqueValue() {
String lastValue = (String) getSession().createQuery("select businessValue " +
"from Invoice as invoice " +
"order by businessValue")
.setMaxResults(1)
.setReadOnly(true)
.uniqueResult();
// generate new value
return newValue;
}
}
The Service layer will call the factory and save a new Invoice:
#Transactional
public void saveNewInvoice() {
final String newValue = businessFactory.createUniqueValue();
Invoice invoice = new Invoice(newValue);
invoiceRepository.save(invoice);
}
The problem here is that a situation might exist where trx1 and trx2 read business value 'INVOICE_BAA_002'.
What happens next is that 2 trx's are working with the same value. The trx that first commits will succeed, the 2nd will fail due to a unique constraint exception.
Therefore I need to put a lock on Invoice table when reading out the latest Business value. I think this lock should be active until the new Invoice entity is saved to DB.
How should I do this in Hibernate?
Instead of using a conflict detection concurrency control mechanism, such a relying on unique constraints or optimistic locking, you can use pessimistic locking.
You need to have:
An InvoiceSequence table with an Entity mapped to it
This table has only one row, storing the latest invoice sequence value
You acquire an exclusive lock on the record:
InvoiceSequence invoiceSequence = em.find(InvoiceSequence.class, 1L, LockModeType.PESSIMISTIC_WRITE)
You increment the sequence using your business logic and modify the entity to store the latest value:
String currentSequence = invoiceSequence.getValue();
String nextSequence = sequenceGenerator.nextValue(currentSequence);
invoiceSequence.setValue(nextSequence);
The exclusive lock will prevent both concurrent read and writes too.
A sequence is a set of integers 1, 2, 3, ... that are generated in order on demand. Sequences are frequently used in databases because many applications require each row in a table to contain a unique value, and sequences provide an easy way to generate them.
One way you can do
class BusinessFactory {
public String createUniqueValue() {
String valueNotUsed = (String) getSession().createSQLQuery("select nextval('hibernate_sequence')");
// generate new value
return newValue;
}}
Related
The idea is basically to extend some Repositories with custom functionality. So I got this setup, which DOES work!
#MappedSuperclass
abstract class MyBaseEntity {
#Id
#GeneratedValue(strategy = GenerationType.IDENTITY)
var id: Int = 0
var eid: Int = 0
}
interface MyRepository<T : MyBaseEntity> {
#Transactional
fun saveInsert(entity: T): Optional<T>
}
open class MyRepositoryImpl<T : MyBaseEntity> : MyRepository<T> {
#Autowired
private lateinit var entityManager: EntityManager
#Transactional
override fun saveInsert(entity: T): Optional<T> {
// lock table
entityManager.createNativeQuery("LOCK TABLE myTable WRITE").executeUpdate()
// get current max EID
val result = entityManager.createNativeQuery("SELECT MAX(eid) FROM myTable LIMIT 1").singleResult as? Int ?: 0
// set entities EID with incremented result
entity.eid = result + 1
// test if table is locked. sending manually 2-3 POST requests to REST
Thread.sleep(5000)
// save
entityManager.persist(entity)
// unlock
entityManager.createNativeQuery("UNLOCK TABLES").executeUpdate()
return Optional.of(entity)
}
}
How would I do this more spring-Like?
At first, I thought the #Transactional would do the LOCK and UNLOCK stuff. I tried a couple of additional parameters and #Lock. I did go through docs and some tutorials but the abstract technical English is often not easy to understand. At the end, I did not get a working solution so I manually added the table-locking, which works fine. Still would prefer a more spring-like way to do it.
1) There might be a problem with your current design as well. The persist does not instantly INSERT a row in the database. That happens on transaction commit when the method returns.
So you unlock the table before the actual insert:
// save
entityManager.persist(entity) // -> There is no INSERT at this point.
// unlock
entityManager.createNativeQuery("UNLOCK TABLES").executeUpdate()
2) Going back to how to do it only with JPA without natives (it still requires a bit of a workaround as it is not supported by default):
// lock table by loading one existing entity and setting the LockModeType
Entity lockedEntity = entityManager.find(Entity.class, 1, LockModeType.PESSIMISTIC_WRITE);
// get current max EID, TRY NOT TO USE NATIVE QUERY HERE
// set entities EID with incremented result
// save
entityManager.persist(entity)
entityManager.flush() // -> Force an actual INSERT
// unlock by passing the previous entity
entityManager.lock(lockedEntity, LockModeType.NONE)
I want to get unique ID from my domain object (table) ID each time a method is called. So that ID's do not repeat. I have a function that returns unique ID.
public static Long generateID (Short company)
throws Exception
{
IDDAO iDDAO = SpringApplicationContext.getBean (IDDAO.class);
ID iD = iDDAO.findByNaturalKey (new IDNatKey (company);
if (iD != null)
{
// Check if ID has reached limit, then reset the ID to the first ID
if (iD.getLatestIDno ().longValue () == iD.getLastIDno ().longValue ())
{
iD.setLatestIDno (iD.getFrstIDno ());
}
// Get next ID
iD.setLatestIDno (iD.getLatestIDno () + 1);
// update database with latest id
iDDAO.update (iD);
return iD.getLatestIDno ();
}
}
The issue is that if access the application from two machines and press button from UI to generate ID exactly at the same time, there are sometimes duplicate IDs returned from this method
e.g.
Long ID = TestClass.generateID (123);
This gives me duplicate sometimes.
I made the method like this
public static synchronized Long generateID (Short company)
throws Exception
so that only one thread can go in this function at a time, but the duplicate issue is still there.
I do not want to use database sequences as I do not want gaps in the ID sequences if the transaction rolls back, in that case sequence will be incremented still, which I do not want.Gaps at the middle are OK but not at end. E.g we have 1, 2 and 3 as IDs and 2 rolls back, that is OK. But if 3 rolls back, we should get 3 again when another user comes, in case of sequence, it will give 4
Please help me tell what I am doing incorrect ? static synchronized will still cause other threads to go inside this function at same time ? I have many other static (but not synchronized) functions in the class. Will this cause issue with them too if I make it static synchronized ?
Thanks
Aiden
You can use java.util.UUID. it will generate a universal uniqueId.
Keep 2 unique IDs:
a db-provided, internal transaction ID, created by an autoincrement every time a new transaction is built. Gaps may appear if transactions are rolled back.
a pretty, gap-less "ticket ID", assigned only once the transaction commits successfully.
Assign both from the DB - it is best to keep all shared state there, as the DB will guarantee ACID, while Java concurrency is far trickier to get right.
In that case I think you try the below :
synchronized(this){if (iD != null)
{
// Check if ID has reached limit, then reset the ID to the first ID
if (iD.getLatestIDno ().longValue () == iD.getLastIDno ().longValue ())
{
iD.setLatestIDno (iD.getFrstIDno ());
}
// Get next ID
iD.setLatestIDno (iD.getLatestIDno () + 1);
// update database with latest id
iDDAO.update (iD);
return iD.getLatestIDno ();
}}
Entity with id autogenerated from oracle trigger sequence.
#Entity
#Table(name = "REPORT", schema = "WEBPORTAL")
public class Report {
private Integer id;
....
#Id
#GeneratedValue(strategy = GenerationType.SEQUENCE, generator="report_sequence")
#SequenceGenerator(name="report_sequence", sequenceName = "report_id_seq")
#Column(name="REPORT_ID", unique = true, nullable = false)
public Integer getId() {
return id;
}
....
}
Service
#Service("reportService")
public class ReportServiceImpl implements ReportService {
....
#Transactional(readOnly=false)
public void saveOrUpdate(Report report) {
reportDAO.saveOrUpdate(report);
}
}
DAO
#Repository
public class ReportDAOImpl implements ReportDAO {
....
#Override
public Report save(Report report) {
try {
Session session = sessionFactory.getCurrentSession();
session.save(report);
} catch (Exception e) {
logger.error("error", e);
}
return report;
}
}
And When I call service's saveOrUpdate and then try to reach id of entity I get different value than persisted in database. Values on database with autogeneration all is ok. Any suggestions?
reportService.saveOrUpdate(report);
System.out.println(report.getId());
prints: 4150
but saved id in database is: 84
NOTE: My purpose to get Id comes from that I wanted to save childs with cascade. But foreign key on child was different in database(the values of id that I get with getId()).
And Id generated in database is incremented by 2. EX: 80, 82, 84.
UPDATE:
Oracle trigger for sequence generation
CREATE OR REPLACE TRIGGER REPORT_ID_TRIG
BEFORE INSERT ON WEBPORTAL.REPORT
FOR EACH ROW
BEGIN
SELECT report_id_seq.NEXTVAL
INTO :new.report_id
FROM dual;
END;
ANSWER: Trigger should check if id is null
CREATE OR REPLACE TRIGGER REPORT_ID_TRIG
BEFORE INSERT ON WEBPORTAL.REPORT
FOR EACH ROW
WHEN (new.report_id is null)
BEGIN
SELECT report_id_seq.NEXTVAL
INTO :new.report_id
FROM dual;
END;
DESCRIPTION:
#GeneratedValue is not just a sequence generator. It's bit of HiLo algorithm.When it first requests id from database it multiplies it with 50(it can differ) and next 50 new entities will be given ids consequently, and than next request to database. This is to decrease request to database.
The numbers that I get from java was right numbers that should be saved on report.
Without id null value check Hibernate firstly requested for id from database and sequence.nextval called. When hibernate was persisting it(completing transaction) the database called sequence.next second time and set that value to database. So on ReportDetails there was true id value of report and on the Report id it was id set from database.
The problem is that two separate mechanisms are in place to generate the key:
one at Hibernate level which is to call a sequence and use the value to populate an Id column and send it to the database as the insert key
and another mechanism at the database that Hibernate does not know about: the column is incremented via a trigger.
Hibernate thinks that the insert was made with the value of the sequence, but in the database something else occurred. The simplest solution would probably be to remove the trigger mechanism, and let Hibernate populate the key based on the sequence only.
Another Solution:
check your trigger definition that should be in this format
(WHEN (new.report_id is null) ) is important.
CREATE OR REPLACE TRIGGER TRIGGER_NAME
BEFORE INSERT ON TABLE_NAME
FOR EACH ROW
WHEN (new.id is null)
BEGIN
SELECT SEQUENCE_NAME.NEXTVAL
INTO :new.id
FROM dual;
END
I need to save data into 2 tables (an entity and an association table).
I simply save my entity with the save() method from my entity repository.
Then, for performances, I need to insert rows into an association table in native sql. The rows have a reference on the entity I saved before.
The issue comes here : I get an integrity constraint exception concerning a Foreign Key. The entity saved first isn't known in this second query.
Here is my code :
The repo :
public interface DistributionRepository extends JpaRepository<Distribution, Long>, QueryDslPredicateExecutor<Distribution> {
#Modifying
#Query(value = "INSERT INTO DISTRIBUTION_PERIMETER(DISTRIBUTION_ID, SERVICE_ID) SELECT :distId, p.id FROM PERIMETER p "
+ "WHERE p.id in (:serviceIds) AND p.discriminator = 'SRV' ", nativeQuery = true)
void insertDistributionPerimeter(#Param(value = "distId") Long distributionId, #Param(value = "serviceIds") Set<Long> servicesIds);
}
The service :
#Service
public class DistributionServiceImpl implements IDistributionService {
#Inject
private DistributionRepository distributionRepository;
#Override
#Transactional
public DistributionResource distribute(final DistributionResource distribution) {
// 1. Entity creation and saving
Distribution created = new Distribution();
final Date distributionDate = new Date();
created.setStatus(EnumDistributionStatus.distributing);
created.setDistributionDate(distributionDate);
created.setDistributor(agentRepository.findOne(distribution.getDistributor().getMatricule()));
created.setDocument(documentRepository.findOne(distribution.getDocument().getTechId()));
created.setEntity(entityRepository.findOne(distribution.getEntity().getTechId()));
created = distributionRepository.save(created);
// 2. Association table
final Set<Long> serviceIds = new HashSet<Long>();
for (final ServiceResource sr : distribution.getServices()) {
serviceIds.add(sr.getTechId());
}
// EXCEPTION HERE
distributionRepository.insertDistributionPerimeter(created.getId(), serviceIds);
}
}
The 2 queries seem to be in different transactions whereas I set the #Transactionnal annotation. I also tried to execute my second query with an entityManager.createNativeQuery() and got the same result...
Invoke entityManager.flush() before you execute your native queries or use saveAndFlush instead.
I your specific case I would recommend to use
created = distributionRepository.saveAndFlush(created);
Important: your "native" queries must use the same transaction! (or you need a now transaction isolation level)
you also wrote:
I don't really understand why the flush action is not done by default
Flushing is handled by Hibernate (it can been configured, default is "auto"). This mean that hibernate will flush the data at any point in time. But always before you commit the transaction or execute an other SQL statement VIA HIBERNATE. - So normally this is no problem, but in your case, you bypass hibernate with your native query, so hibernate will not know about this statement and therefore it will not flush its data.
See also this answer of mine: https://stackoverflow.com/a/17889017/280244 about this topic
I want the first to be generated:
#Id
#Column(name = "PRODUCT_ID", unique = true, nullable = false, precision = 12,
scale = 0)
#GeneratedValue(strategy = GenerationType.SEQUENCE, generator = "PROD_GEN")
#BusinessKey
public Long getAId() {
return this.aId;
}
I want the bId to be initially exactly as the aId. One approach is to insert the entity, then get the aId generated by the DB (2nd query) and then update the entity, setting the bId to be equal to aId (3rd query). Is there a way to get the bId to get the same generated value as aId?
Note that afterwards, I want to be able to update bId from my gui.
If the solution is JPA, even better.
Choose your poison:
Option #1
you could annotate bId as org.hibernate.annotations.Generated and use a database trigger on insert (I'm assuming the nextval has already been assigned to AID so we'll assign the curval to BID):
CREATE OR REPLACE TRIGGER "MY_TRIGGER"
before insert on "MYENTITY"
for each row
begin
select "MYENTITY_SEQ".curval into :NEW.BID from dual;
end;
I'm not a big fan of triggers and things that happen behind the scene but this seems to be the easiest option (not the best one for portability though).
Option #2
Create a new entity, persist it, flush the entity manager to get the id assigned, set the aId on bId, merge the entity.
em.getTransaction().begin();
MyEntity e = new MyEntity();
...
em.persist(e);
em.flush();
e.setBId(e.getAId());
em.merge(e);
...
em.getTransaction().commit();
Ugly, but it works.
Option #3
Use callback annotations to set the bId in-memory (until it gets written to the database):
#PostPersist
#PostLoad
public void initialiazeBId() {
if (this.bId == null) {
this.bId = aId;
}
}
This should work if you don't need the id to be written on insert (but in that case, see Option #4).
Option #4
You could actually add some logic in the getter of bId instead of using callbacks:
public Long getBId() {
if (this.bId == null) {
return this.aId;
}
return this.bId;
}
Again, this will work if you don't need the id to be persisted in the database on insert.
If you use JPA, after inserting the new A the id should be set to the generated value, i tought (maybe it depends on which jpa provider you use), so no 2nd query needed. then set bld to ald value in your DAO?