Using transactions in Quarkus
The quarkus-narayana-jta extension provides a Transaction Manager that coordinates and expose transactions to your applications as described in the link: Jakarta Transactions specification, formerly known as Java Transaction API (JTA).
When discussing Quarkus transactions, this guide refers to Jakarta Transactions transaction style and uses only the term transaction to address them.
Also, Quarkus does not support distributed transactions.
This means that models that propagate transaction context, such as Java Transaction Service (JTS), REST-AT, WS-Atomic Transaction, and others, are not supported by the narayana-jta extension.
Setting it up
You don’t need to worry about setting it up most of the time as extensions needing it will simply add it as a dependency. Hibernate ORM for example will include the transaction manager and set it up properly.
You might need to add it as a dependency explicitly if you are using transactions directly without Hibernate ORM for example. Add the following to your build file:
<dependency>
<groupId>io.quarkus</groupId>
<artifactId>quarkus-narayana-jta</artifactId>
</dependency>implementation("io.quarkus:quarkus-narayana-jta")Starting and stopping transactions: defining your boundaries
You can define your transaction boundaries either declaratively with @Transactional or programmatically with QuarkusTransaction. You can also use
the JTA UserTransaction API directly, however this is less user-friendly than QuarkusTransaction.
Declarative approach
The easiest way to define your transaction boundaries is to use the @Transactional annotation on your entry method (jakarta.transaction.Transactional).
@ApplicationScoped
public class SantaClausService {
@Inject ChildDAO childDAO;
@Inject SantaClausDAO santaDAO;
@Transactional (1)
public void getAGiftFromSanta(Child child, String giftDescription) {
// some transaction work
Gift gift = childDAO.addToGiftList(child, giftDescription);
if (gift == null) {
throw new OMGGiftNotRecognizedException(); (2)
}
else {
santaDAO.addToSantaTodoList(gift);
}
}
}| 1 | This annotation defines your transaction boundaries and will wrap this call within a transaction. |
| 2 | A RuntimeException crossing the transaction boundaries will roll back the transaction. |
@Transactional can be used to control transaction boundaries on any CDI bean at the method level or at the class level to ensure every method is transactional.
That includes REST endpoints.
You can control whether and how the transaction is started with parameters on @Transactional:
-
@Transactional(REQUIRED)(default): starts a transaction if none was started, stays with the existing one otherwise. -
@Transactional(REQUIRES_NEW): starts a transaction if none was started ; if an existing one was started, suspends it and starts a new one for the boundary of that method. -
@Transactional(MANDATORY): fails if no transaction was started ; works within the existing transaction otherwise. -
@Transactional(SUPPORTS): if a transaction was started, joins it ; otherwise works with no transaction. -
@Transactional(NOT_SUPPORTED): if a transaction was started, suspends it and works with no transaction for the boundary of the method ; otherwise works with no transaction. -
@Transactional(NEVER): if a transaction was started, raises an exception ; otherwise works with no transaction.
REQUIRED or NOT_SUPPORTED are probably the most useful ones.
This is how you decide whether a method is to be running within or outside a transaction.
Make sure to check the JavaDoc for the precise semantic.
The transaction context is propagated to all calls nested in the @Transactional method as you would expect (in this example childDAO.addToGiftList() and santaDAO.addToSantaTodoList()).
The transaction will commit unless a runtime exception crosses the method boundary.
You can override whether an exception forces the rollback or not by using @Transactional(dontRollbackOn=SomeException.class) (or rollbackOn).
You can also programmatically ask for a transaction to be marked for rollback.
Inject a TransactionManager for this.
@ApplicationScoped
public class SantaClausService {
@Inject TransactionManager tm; (1)
@Inject ChildDAO childDAO;
@Inject SantaClausDAO santaDAO;
@Transactional
public void getAGiftFromSanta(Child child, String giftDescription) {
// some transaction work
Gift gift = childDAO.addToGiftList(child, giftDescription);
if (gift == null) {
tm.setRollbackOnly(); (2)
}
else {
santaDAO.addToSantaTodoList(gift);
}
}
}| 1 | Inject the TransactionManager to be able to activate setRollbackOnly semantic. |
| 2 | Programmatically decide to set the transaction for rollback. |
Transaction configuration
Advanced configuration of the transaction is possible with the use of the @TransactionConfiguration annotation that is set in addition to the standard @Transactional annotation on your entry method or at the class level.
The @TransactionConfiguration annotation allows to set a timeout property, in seconds, that applies to transactions created within the annotated method.
This annotation may only be placed on the top level method delineating the transaction. Annotated nested methods once a transaction has started will throw an exception.
If defined on a class, it is equivalent to defining it on all the methods of the class marked with @Transactional.
The configuration defined on a method takes precedence over the configuration defined on a class.
Reactive extensions
If your @Transactional-annotated method returns a reactive value, such as:
-
CompletionStage(from the JDK) -
Publisher(from Reactive-Streams) -
Any type that can be converted to one of the two previous types using Reactive Type Converters
then the behaviour is a bit different, because the transaction will not be terminated until the returned reactive value is terminated. In effect, the returned reactive value will be listened to and if it terminates exceptionally the transaction will be marked for rollback, and will be committed or rolled-back only at termination of the reactive value.
This allows your reactive methods to keep on working on the transaction asynchronously until their work is really done, and not just until the reactive method returns.
If you need to propagate your transaction context across your reactive pipeline, please see the Context Propagation guide.
Programmatic approach
You can use static methods on QuarkusTransaction to define transaction boundaries. This provides two different options,
a functional approach that allows you to run a lambda within the scope of a transaction, or by using explicit begin,
commit and rollback methods.
import io.quarkus.narayana.jta.QuarkusTransaction;
public class TransactionExample {
public void beginExample() {
QuarkusTransaction.begin();
//do work
QuarkusTransaction.commit();
QuarkusTransaction.begin(QuarkusTransaction.beginOptions()
.timeout(10));
//do work
QuarkusTransaction.rollback();
}
public void runnerExample() {
QuarkusTransaction.requiringNew().run(() -> {
//do work
});
QuarkusTransaction.joiningExisting().run(() -> {
//do work
});
int result = QuarkusTransaction.requiringNew()
.timeout(10)
.exceptionHandler((throwable) -> {
if (throwable instanceof SomeException) {
return TransactionExceptionResult.COMMIT;
}
return TransactionExceptionResult.ROLLBACK;
})
.call(() -> {
//do work
return 0;
});
}
}The above example shows a few different ways the API can be used.
The first method simply calls begin, does some work and commits it.
This created transaction is tied to the CDI request scope, so if it is still active when the request scope is destroyed then it will
be automatically rolled back. This removes the need to explicitly catch exceptions and call rollback, and acts as a safety net
against inadvertent transaction leaks, however it does mean that this can only be used when the request scope is active. The second
example in the method calls begin with a timeout option, and then rolls back the transaction.
The second method shows the use of lambda scoped transactions with QuarkusTransaction.runner(…);
the first example just runs a Runnable within a new transaction,
the second does the same but joining the existing transaction (if any),
and the third calls a Callable with some specific options.
In particular the exceptionHandler method can be used to control if the transaction is rolled back or not on exception.
The following semantics are supported:
QuarkusTransaction.disallowingExisting()/DISALLOW_EXISTING-
If a transaction is already associated with the current thread a
QuarkusTransactionExceptionwill be thrown, otherwise a new transaction is started, and follows all the normal lifecycle rules. QuarkusTransaction.joiningExisting()/JOIN_EXISTING-
If no transaction is active then a new transaction will be started, and committed when the method ends. If an exception is thrown the exception handler registered by
#exceptionHandler(Function)will be called to decide if the TX should be committed or rolled back. If an existing transaction is active then the method is run in the context of the existing transaction. If an exception is thrown the exception handler will be called, however a result ofExceptionResult#ROLLBACKwill result in the TX marked as rollback only, while a result ofExceptionResult#COMMITwill result in no action being taken. QuarkusTransaction.requiringNew()/REQUIRE_NEW-
If an existing transaction is already associated with the current thread then the transaction is suspended, then a new transaction is started which follows all the normal lifecycle rules, and when it’s complete the original transaction is resumed. Otherwise, a new transaction is started, and follows all the normal lifecycle rules.
QuarkusTransaction.suspendingExisting()/SUSPEND_EXISTING-
If no transaction is active then these semantics are basically a no-op. If a transaction is active then it is suspended, and resumed after the task is run. The exception handler will never be consulted when these semantics are in use, specifying both an exception handler and these semantics are considered an error. These semantics allows for code to easily be run outside the scope of a transaction.
Legacy API approach
The less easy way is to inject a UserTransaction and use the various transaction demarcation methods.
@ApplicationScoped
public class SantaClausService {
@Inject ChildDAO childDAO;
@Inject SantaClausDAO santaDAO;
@Inject UserTransaction transaction;
public void getAGiftFromSanta(Child child, String giftDescription) {
// some transaction work
try {
transaction.begin();
Gift gift = childDAO.addToGiftList(child, giftDescription);
santaDAO.addToSantaTodoList(gift);
transaction.commit();
}
catch(SomeException e) {
// do something on Tx failure
transaction.rollback();
}
}
}|
You cannot use |
Configuring the transaction timeout
You can configure the default transaction timeout, the timeout that applies to all transactions managed by the transaction manager, via the property quarkus.transaction-manager.default-transaction-timeout, specified as a duration.
|
To write duration values, use the standard You can also use a simplified format, starting with a number:
In other cases, the simplified format is translated to the
|
The default value is 60 seconds.
Configuring transaction node name identifier
Narayana, as the underlying transaction manager, has a concept of a unique node identifier. This is important if you consider using XA transactions that involve multiple resources.
The node name identifier plays a crucial part in the identification of a transaction. The node name identifier is forged into the transaction id when the transaction is created. Based on the node name identifier, the transaction manager is capable of recognizing the XA transaction counterparts created in database or JMS broker. The identifier makes possible for the transaction manager to roll back the transaction counterparts during recovery.
The node name identifier needs to be unique per transaction manager deployment. And the node identifier needs to be stable over the transaction manager restarts.
The node name identifier may be configured via the property quarkus.transaction-manager.node-name.
|
The node name cannot be longer than 28 bytes.
To automatically shorten names longer than 28 bytes, set Shortening is implemented by hashing the node name, encoding the hash to Base64 and then truncating the result. As with all hashes, the resulting shortened node name could potentially conflict with another shortened node name, but it is very unlikely. |
Using @TransactionScoped to bind CDI beans to the transaction lifecycle
You can define beans that live for as long as a transaction, and through CDI lifecycle events perform actions when a transaction starts and ends.
Just assign the transaction scope to such beans using the @TransactionScoped annotation:
@TransactionScoped
public class MyTransactionScopedBean {
// The bean's state is bound to a specific transaction,
// and restored even after suspending then resuming the transaction.
int myData;
@PostConstruct
void onBeginTransaction() {
// This gets invoked after a transaction begins.
}
@PreDestroy
void onBeforeEndTransaction() {
// This gets invoked before a transaction ends (commit or rollback).
}
}Alternatively, if you don’t necessarily need to hold state during the transaction, and just want to react to transaction start/end events, you can simply declare event listeners in a differently scoped bean:
@ApplicationScoped
public class MyTransactionEventListeningBean {
void onBeginTransaction(@Observes @Initialized(TransactionScoped.class) Object event) {
// This gets invoked when a transaction begins.
}
void onBeforeEndTransaction(@Observes @BeforeDestroyed(TransactionScoped.class) Object event) {
// This gets invoked before a transaction ends (commit or rollback).
}
void onAfterEndTransaction(@Observes @Destroyed(TransactionScoped.class) Object event) {
// This gets invoked after a transaction ends (commit or rollback).
}
}
The event object represents the transaction ID, and defines toString()/equals()/hashCode() accordingly.
|
In listener methods, you can access more information about the transaction in progress by accessing the TransactionManager,
which is a CDI bean and can be @Injected.
|
Configure storing of Quarkus transaction logs in a database
In cloud environments where persistent storage is not available, such as when application containers are unable to use persistent volumes, you can configure the transaction management to store transaction logs in a database by using a Java Database Connectivity (JDBC) datasource.
However, in cloud-native apps, using a database to store transaction logs has additional requirements.
The narayana-jta extension, which manages these transactions, requires stable storage, a unique reusable node identifier, and a steady IP address to work correctly.
While the JDBC object store provides a stable storage, users must still plan how to meet the other two requirements.
Quarkus, after you evaluate whether using a database to store transaction logs is right for you, allows the following JDBC-specific configuration of the object store included in quarkus.transaction-manager.object-store.<property> properties, where <property> can be:
-
type(string): Configure this property tojdbcto enable usage of a Quarkus JDBC datasource for storing transaction logs. The default value isfile-system. -
datasource(string): Specify the name of the datasource for the transaction log storage. If no value is provided for thedatasourceproperty, Quarkus uses the default datasource. -
create-table(boolean): When set totrue, the transaction log table gets automatically created if it does not already exist. The default value isfalse. -
drop-table(boolean): When set totrue, the tables are dropped on startup if they already exist. The default value isfalse. -
table-prefix(string): Specify the prefix for a related table name. The default value isquarkus_.
For more configuration information, see the Narayana JTA - Transaction manager section of the Quarkus All configuration options reference.
-
Create the transaction log table during the initial setup by setting the
create-tableproperty totrue. -
JDBC datasources and ActiveMQ Artemis allow the enlistment and automatically register the
XAResourceRecovery.-
JDBC datasources is part of
quarkus-agroal, and it needs to usequarkus.datasource.jdbc.transactions=XA. -
ActiveMQ Artemis is part of
quarkus-pooled-jms, and it needs to usequarkus.pooled-jms.transaction=XA.
-
-
To ensure data integrity in case of application crashes or failures, enable the transaction crash recovery with the
quarkus.transaction-manager.enable-recovery=trueconfiguration.
|
To work around the current known issue of Agroal having a different view on running transaction checks, set the datasource transaction type for the datasource responsible for writing the transaction logs to quarkus.datasource.TX_LOG.jdbc.transactions=disabled This example uses TX_LOG as the datasource name. |
Why always having a transaction manager?
- Does it work everywhere I want to?
-
Yep, it works in your Quarkus application, in your IDE, in your tests, because all of these are Quarkus applications. JTA has some bad press for some people. I don’t know why. Let’s just say that this is not your grandpa’s JTA implementation. What we have is perfectly embeddable and lean.
- Does it do 2 Phase Commit and slow down my app?
-
No, this is an old folk tale. Let’s assume it essentially comes for free and let you scale to more complex cases involving several datasources as needed.
- I don’t need transaction when I do read only operations, it’s faster.
-
Wrong.
First off, just disable the transaction by marking your transaction boundary with@Transactional(NOT_SUPPORTED)(orNEVERorSUPPORTSdepending on the semantic you want).
Second, it’s again fairy tale that not using transaction is faster. The answer is, it depends on your DB and how many SQL SELECTs you are making. No transaction means the DB does have a single operation transaction context anyway.
Third, when you do several SELECTs, it’s better to wrap them in a single transaction because they will all be consistent with one another. Say your DB represents your car dashboard, you can see the number of kilometers remaining and the fuel gauge level. By reading it in one transaction, they will be consistent. If you read one and the other from two different transactions, then they can be inconsistent. It can be more dramatic if you read data related to rights and access management for example. - Why do you prefer JTA vs Hibernate’s transaction management API
-
Managing the transactions manually via
entityManager.getTransaction().begin()and friends lead to a butt ugly code with tons of try catch finally that people get wrong. Transactions are also about JMS and other database access, so one API makes more sense. - It’s a mess because I don’t know if my Jakarta Persistence persistence unit is using
JTAorResource-levelTransaction -
It’s not a mess in Quarkus :) Resource-level was introduced to support Jakarta Persistence in a non-managed environment. But Quarkus is both lean and a managed environment, so we can safely always assume we are in JTA mode. The end result is that the difficulties of running Hibernate ORM + CDI + a transaction manager in Java SE mode are solved by Quarkus.
