Spring/spring-framework
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Rollback reactive transaction on cancel

Browse Source 
This commit introduces a change in reactive transaction semantics for
cancel signals. Canceling a subscription now rolls back a reactive transaction
to achieve a deterministic transaction outcome.

Previously, cancel signals committed a transaction which could
cause partially committed transactions depending on when the cancel happened.
This commit is contained in:
Mark Paluch
2020-06-16 11:00:58 +02:00
committed by Juergen Hoeller
parent 8853f4b883
commit 217b6e37a6
8 changed files with 240 additions and 35 deletions
Download Patch File Download Diff File Expand all files Collapse all files

46
src/docs/asciidoc/data-access.adoc
View File

@@ -184,7 +184,7 @@ transaction management. The following listing shows the definition of the
@Throws(TransactionException::class)
fun rollback(status: TransactionStatus)
}
----
----
This is primarily a service provider interface (SPI), although you can use it
<<transaction-programmatic-ptm, programmatically>> from your application code. Because
@@ -241,7 +241,7 @@ listing shows the transaction strategy defined by
@Throws(TransactionException::class)
fun rollback(status: ReactiveTransaction): Mono<Void>
}
----
----
The reactive transaction manager is primarily a service provider interface (SPI),
although you can use it <<transaction-programmatic-rtm, programmatically>> from your
@@ -1698,7 +1698,7 @@ in the application context:
@Transactional("account")
public void doSomething() { ... }
@Transactional("reactive-account")
public Mono<Void> doSomethingReactive() { ... }
}
@@ -2399,11 +2399,11 @@ the `TransactionOperator` resembles the next example:
}
public Mono<Object> someServiceMethod() {
// the code in this method executes in a transactional context
Mono<Object> update = updateOperation1();
return update.then(resultOfUpdateOperation2).as(transactionalOperator::transactional);
}
}
@@ -2463,9 +2463,7 @@ In Reactive Streams, a `Subscriber` can cancel its `Subscription` and terminate
`Publisher`. Operators in Project Reactor, as well as in other libraries, such as `next()`,
`take(long)`, `timeout(Duration)`, and others can issue cancellations. There is no way to
know the reason for the cancellation, whether it is due to an error or a simply lack of
interest to consume further, and in version 5.2 the `TransactionalOperator` defaults to
committing the transaction on cancel. In version 5.3 this behavior will change and
transactions will be roll back on cancel to create a reliable and deterministic outcome.
interest to consume further. Since version 5.3 cancel signals lead to a roll back.
As a result it is important to consider the operators used downstream from a transaction
`Publisher`. In particular in the case of a `Flux` or other multi-value `Publisher`,
the full output must be consumed to allow the transaction to complete.
@@ -2490,7 +2488,7 @@ following example shows customization of the transactional settings for a specif
public SimpleService(ReactiveTransactionManager transactionManager) {
DefaultTransactionDefinition definition = new DefaultTransactionDefinition();
// the transaction settings can be set here explicitly if so desired
definition.setIsolationLevel(TransactionDefinition.ISOLATION_READ_UNCOMMITTED);
definition.setTimeout(30); // 30 seconds
@@ -2588,9 +2586,9 @@ following example shows how to do so:
def.setPropagationBehavior(TransactionDefinition.PROPAGATION_REQUIRED);
Mono<ReactiveTransaction> reactiveTx = txManager.getReactiveTransaction(def);
reactiveTx.flatMap(status -> {
Mono<Object> tx = ...; // execute your business logic here
return tx.then(txManager.commit(status))
@@ -2942,7 +2940,7 @@ this `DataSource`. The following example autowires a `DataSource`:
----
@Repository
class JdbcMovieFinder(dataSource: DataSource) : MovieFinder {
private val jdbcTemplate = JdbcTemplate(dataSource)
// ...
@@ -3202,8 +3200,8 @@ The following query finds and populates a single domain object:
----
val actor = jdbcTemplate.queryForObject(
"select first_name, last_name from t_actor where id = ?",
arrayOf(1212L)) { rs, _ ->
Actor(rs.getString("first_name"), rs.getString("last_name"))
arrayOf(1212L)) { rs, _ ->
Actor(rs.getString("first_name"), rs.getString("last_name"))
}
----
@@ -3455,7 +3453,7 @@ method with `@Autowired`. The following example shows how to do so:
class JdbcCorporateEventDao(dataSource: DataSource) : CorporateEventDao { // <2>
private val jdbcTemplate = JdbcTemplate(dataSource) // <3>
// JDBC-backed implementations of the methods on the CorporateEventDao follow...
}
----
@@ -3794,10 +3792,10 @@ translator:
private val jdbcTemplate = JdbcTemplate(dataSource).apply {
// create a custom translator and set the DataSource for the default translation lookup
exceptionTranslator = CustomSQLErrorCodesTranslator().apply {
this.dataSource = dataSource
this.dataSource = dataSource
}
}
fun updateShippingCharge(orderId: Long, pct: Long) {
// use the prepared JdbcTemplate for this update
this.jdbcTemplate!!.update("update orders" +
@@ -4021,8 +4019,8 @@ on Oracle but may not work on other platforms:
val name = "Rob"
val keyHolder = GeneratedKeyHolder()
jdbcTemplate.update({
it.prepareStatement (INSERT_SQL, arrayOf("id")).apply { setString(1, name) }
jdbcTemplate.update({
it.prepareStatement (INSERT_SQL, arrayOf("id")).apply { setString(1, name) }
}, keyHolder)
// keyHolder.getKey() now contains the generated key
@@ -4960,7 +4958,7 @@ the constructor of your `SimpleJdbcCall`. The following example shows this confi
private var procReadActor = SimpleJdbcCall(JdbcTemplate(dataSource).apply {
isResultsMapCaseInsensitive = true
}).withProcedureName("read_actor")
// ... additional methods
}
----
@@ -5718,7 +5716,7 @@ the supplied `ResultSet`, as follows:
import org.springframework.jdbc.core.RowMapper
class GenreMapper : RowMapper<Genre> {
override fun mapRow(rs: ResultSet, rowNum: Int): Genre {
return Genre(rs.getString("name"))
}
@@ -6836,7 +6834,7 @@ implementation resembles the following example, based on the plain Hibernate API
.Kotlin
----
class ProductDaoImpl(private val sessionFactory: SessionFactory) : ProductDao {
fun loadProductsByCategory(category: String): Collection<*> {
return sessionFactory.currentSession
.createQuery("from test.Product product where product.category=?")
@@ -7044,7 +7042,7 @@ and an example for a business method implementation:
----
class ProductServiceImpl(transactionManager: PlatformTransactionManager,
private val productDao: ProductDao) : ProductService {
private val transactionTemplate = TransactionTemplate(transactionManager)
fun increasePriceOfAllProductsInCategory(category: String) {