Package io.ebean

Interface Transaction

    • Field Summary

      Fields 
      Modifier and Type Field Description
      static int READ_COMMITTED
      Read Committed transaction isolation.
      static int READ_UNCOMMITTED
      Read Uncommitted transaction isolation.
      static int REPEATABLE_READ
      Repeatable read transaction isolation.
      static int SERIALIZABLE
      Serializable transaction isolation.

    • Method Summary

      All Methods Static Methods Instance Methods Abstract Methods Default Methods Deprecated Methods 
      Modifier and Type Method Description
      void addModification​(String tableName, boolean inserts, boolean updates, boolean deletes)
      Add table modification information to the TransactionEvent.
      void close()
      Synonym for end() to support AutoClosable.
      void commit()
      Commit the transaction.
      void commitAndContinue()
      Commits the transaction at this point with the expectation that another commit (or rollback or end) will occur later to complete the transaction.
      static Transaction current()
      Return the current transaction (of the default database) or null if there is no current transaction in scope.
      void end()
      If the transaction is active then perform rollback.
      void flush()
      The batch will be flushing automatically but you can use this to explicitly flush the batch if you like.
      void flushBatch()
      Deprecated.
      int getBatchSize()
      Return the current batch size.
      Connection getConnection()
      Return the underlying Connection object.
      Object getUserObject​(String name)
      Get an object added with putUserObject(String, Object).
      boolean isActive()
      Return true if the transaction is active.
      default boolean isBatchFlushOnQuery()
      Deprecated.
      boolean isBatchMode()
      Return the batch mode at the transaction level.
      boolean isBatchOnCascade()
      Return the batch mode at the request level.
      boolean isFlushOnQuery()
      Return true if the batch (of persisted beans or executed UpdateSql etc) should be flushed prior to executing a query.
      boolean isReadOnly()
      Return true if this transaction is read only.
      boolean isRollbackOnly()
      Return true if the transaction is marked as rollback only.
      boolean isSkipCache()
      Return true if the L2 cache should be skipped.
      void putUserObject​(String name, Object value)
      Add an arbitrary user object to the transaction.
      void register​(TransactionCallback callback)
      Register a TransactionCallback with this transaction.
      void rollback()
      Rollback the transaction.
      void rollback​(Throwable e)
      Rollback the transaction specifying a throwable that caused the rollback to occur.
      default void setBatchFlushOnMixed​(boolean batchFlushOnMixed)
      Deprecated.
      default void setBatchFlushOnQuery​(boolean batchFlushOnQuery)
      Deprecated.
      default void setBatchGetGeneratedKeys​(boolean getGeneratedKeys)
      Deprecated.
      void setBatchMode​(boolean useBatch)
      Turn on or off use of JDBC statement batching.
      void setBatchOnCascade​(boolean batchMode)
      Set the JDBC batch mode to use for a save() or delete() when cascading to children.
      void setBatchSize​(int batchSize)
      Specify the number of statements before a batch is flushed automatically.
      void setDocStoreBatchSize​(int batchSize)
      Set the batch size to use for sending messages to the document store.
      void setDocStoreMode​(io.ebean.annotation.DocStoreMode mode)
      Set the behavior for document store updates on this transaction.
      void setFlushOnMixed​(boolean batchFlushOnMixed)
      By default when mixing UpdateSql (or CallableSql) with Beans the batch is automatically flushed when you change (between persisting beans and executing UpdateSql or CallableSql).
      void setFlushOnQuery​(boolean batchFlushOnQuery)
      By default executing a query will automatically flush any batched statements (persisted beans, executed UpdateSql etc).
      void setGetGeneratedKeys​(boolean getGeneratedKeys)
      Specify if you want batched inserts to use getGeneratedKeys.
      void setLabel​(String label)
      Set a label on the transaction.
      void setNestedUseSavepoint()
      Set when we want nested transactions to use Savepoint's.
      void setPersistCascade​(boolean persistCascade)
      Explicitly turn off or on the cascading nature of save() and delete().
      void setReadOnly​(boolean readOnly)
      Set whether this transaction should be readOnly.
      void setRollbackOnly()
      Mark the transaction for rollback only.
      void setSkipCache​(boolean skipCache)
      Set if the L2 cache should be skipped for "find by id" and "find by natural key" queries.
      void setUpdateAllLoadedProperties​(boolean updateAllLoadedProperties)
      Set to true when you want all loaded properties to be included in the update (rather than just the changed properties).

    • Method Detail

      • setLabel

        void setLabel​(String label)
        Set a label on the transaction.

        This label is used to group transaction execution times for performance metrics reporting.

      • isReadOnly

        boolean isReadOnly()
        Return true if this transaction is read only.

      • setReadOnly

        void setReadOnly​(boolean readOnly)
        Set whether this transaction should be readOnly.

      • commitAndContinue

        void commitAndContinue()
        Commits the transaction at this point with the expectation that another commit (or rollback or end) will occur later to complete the transaction.

        This is similar to commit() but leaves the transaction "Active".

        Functions

        • Flush the JDBC batch buffer
        • Call commit on the underlying JDBC connection
        • Trigger any registered TransactionCallbacks
        • Perform post-commit processing updating L2 cache, ElasticSearch etc

      • commit

        void commit()
        Commit the transaction.

        This performs commit and completes the transaction closing underlying resources and marking the transaction as "In active".

        Functions

        • Flush the JDBC batch buffer
        • Call commit on the underlying JDBC connection
        • Trigger any registered TransactionCallbacks
        • Perform post-commit processing updating L2 cache, ElasticSearch etc
        • Close any underlying resources, closing the underlying JDBC connection
        • Mark the transaction as "Inactive"

      • rollback

        void rollback()
       throws javax.persistence.PersistenceException
        Rollback the transaction.

        This performs rollback, closes underlying resources and marks the transaction as "In active".

        Functions

        • Call rollback on the underlying JDBC connection
        • Trigger any registered TransactionCallbacks
        • Close any underlying resources, closing the underlying JDBC connection
        • Mark the transaction as "Inactive"
        Throws:
        javax.persistence.PersistenceException

      • rollback

        void rollback​(Throwable e)
       throws javax.persistence.PersistenceException
        Rollback the transaction specifying a throwable that caused the rollback to occur.

        If you are using transaction logging this will log the throwable in the transaction logs.

        Throws:
        javax.persistence.PersistenceException

      • setNestedUseSavepoint

        void setNestedUseSavepoint()
        Set when we want nested transactions to use Savepoint's.

        This means that for a nested transaction:

        • begin transaction maps to creating a savepoint
        • commit transaction maps to releasing a savepoint
        • rollback transaction maps to rollback a savepoint

      • setRollbackOnly

        void setRollbackOnly()
        Mark the transaction for rollback only.

      • isRollbackOnly

        boolean isRollbackOnly()
        Return true if the transaction is marked as rollback only.

      • end

        void end()
        If the transaction is active then perform rollback. Otherwise do nothing.

      • isActive

        boolean isActive()
        Return true if the transaction is active.

      • setDocStoreMode

        void setDocStoreMode​(io.ebean.annotation.DocStoreMode mode)
        Set the behavior for document store updates on this transaction.

        For example, set the mode to DocStoreEvent.IGNORE for this transaction and then any changes via this transaction are not sent to the doc store. This would be used when doing large bulk inserts into the database and we want to control how that is sent to the document store.

      • setDocStoreBatchSize

        void setDocStoreBatchSize​(int batchSize)
        Set the batch size to use for sending messages to the document store.

        You might set this if you know the changes in this transaction result in especially large or especially small payloads and want to adjust the batch size to match.

        Setting this overrides the default of DocStoreConfig.getBulkBatchSize()

      • setPersistCascade

        void setPersistCascade​(boolean persistCascade)
        Explicitly turn off or on the cascading nature of save() and delete(). This gives the developer exact control over what beans are saved and deleted rather than Ebean cascading detecting 'dirty/modified' beans etc.

        This is useful if you can getting back entity beans from a layer of code (potentially remote) and you prefer to have exact control.

        This may also be useful if you are using jdbc batching with jdbc drivers that do not support getGeneratedKeys.

      • setUpdateAllLoadedProperties

        void setUpdateAllLoadedProperties​(boolean updateAllLoadedProperties)
        Set to true when you want all loaded properties to be included in the update (rather than just the changed properties).

        You might set this when using JDBC batch in order to get multiple updates with slightly different sets of changed properties into the same statement and hence better JDBC batch performance.

      • setSkipCache

        void setSkipCache​(boolean skipCache)
        Set if the L2 cache should be skipped for "find by id" and "find by natural key" queries.

        By default DatabaseConfig.isSkipCacheAfterWrite() is true and that means that for "find by id" and "find by natural key" queries which normally hit L2 bean cache automatically - will not do so after a persist/write on the transaction.

        This method provides explicit control over whether "find by id" and "find by natural key" will skip the L2 bean cache or not (regardless of whether the transaction is considered "read only").

        Refer to DatabaseConfig.setSkipCacheAfterWrite(boolean) for configuring the default behavior for using the L2 bean cache in transactions spanning multiple query/persist requests. 

        

   // assume Customer has L2 bean caching enabled ...

   try (Transaction transaction = DB.beginTransaction()) {

     // this uses L2 bean cache as the transaction
     // ... is considered "query only" at this point
     Customer.find.byId(42);

     // transaction no longer "query only" once
     // ... a bean has been saved etc
     someBean.save();

     // will NOT use L2 bean cache as the transaction
     // ... is no longer considered "query only"
     Customer.find.byId(55);



     // explicit control - please use L2 bean cache

     transaction.setSkipCache(false);
     Customer.find.byId(77); // hit the l2 bean cache


     // explicit control - please don't use L2 bean cache

     transaction.setSkipCache(true);
     Customer.find.byId(99); // skips l2 bean cache


     transaction.commit();
   }
        See Also:
        DatabaseConfig.isSkipCacheAfterWrite()

      • isSkipCache

        boolean isSkipCache()
        Return true if the L2 cache should be skipped. More accurately if true then find by id and find by natural key queries should NOT automatically use the L2 bean cache.

      • setBatchMode

        void setBatchMode​(boolean useBatch)
        Turn on or off use of JDBC statement batching.

        Calls to save(), delete(), insert() and execute() all support batch processing. This includes normal beans, CallableSql and UpdateSql. 

        

 try (Transaction transaction = database.beginTransaction()) {

   // turn on JDBC batch
   transaction.setBatchMode(true);

   // tune the batch size
   transaction.setBatchSize(50);

   ...

   transaction.commit();
 }

        getGeneratedKeys

        Often with large batch inserts we want to turn off getGeneratedKeys. We do this via setGetGeneratedKeys(boolean). Also note that some JDBC drivers do not support getGeneratedKeys in JDBC batch mode. 

        

 try (Transaction transaction = database.beginTransaction()) {

   transaction.setBatchMode(true);
   transaction.setBatchSize(100);
   // insert but don't bother getting back the generated keys
   transaction.setBatchGetGeneratedKeys(false);


   // perform lots of inserts ...
   ...

   transaction.commit();
 }

        Flush

        The batch is automatically flushed when it hits the batch size and also when we execute queries or when we mix UpdateSql and CallableSql with save and delete of beans.

        We use flush() to explicitly flush the batch and we can use setFlushOnQuery(boolean) and setFlushOnMixed(boolean) to control the automatic flushing behaviour.

        Example: batch processing of CallableSql executing every 10 rows 

        

 String data = "This is a simple test of the batch processing"
             + " mode and the transaction execute batch method";

 String[] da = data.split(" ");

 String sql = "{call sp_t3(?,?)}";

 CallableSql cs = new CallableSql(sql);
 cs.registerOut(2, Types.INTEGER);

 // (optional) inform Ebean this stored procedure
 // inserts into a table called sp_test
 cs.addModification("sp_test", true, false, false);

 try (Transaction txn = DB.beginTransaction()) {
   txn.setBatchMode(true);
   txn.setBatchSize(10);

   for (int i = 0; i < da.length;) {
     cs.setParameter(1, da[i]);
     DB.execute(cs);
   }

   // Note: commit implicitly flushes
   txn.commit();
 }

      • isBatchMode

        boolean isBatchMode()
        Return the batch mode at the transaction level.

      • setBatchOnCascade

        void setBatchOnCascade​(boolean batchMode)
        Set the JDBC batch mode to use for a save() or delete() when cascading to children.

        This only takes effect when batch mode on the transaction has not already meant that JDBC batch mode is being used.

        This is useful when the single save() or delete() cascades. For example, inserting a 'master' cascades and inserts a collection of 'detail' beans. The detail beans can be inserted using JDBC batch.

        This is effectively already turned on for all platforms apart from older Sql Server.

        Parameters:
        batchMode - the batch mode to use per save(), insert(), update() or delete()
        See Also:
        DatabaseConfig.setPersistBatchOnCascade(PersistBatch)

      • isBatchOnCascade

        boolean isBatchOnCascade()
        Return the batch mode at the request level.

      • setBatchSize

        void setBatchSize​(int batchSize)
        Specify the number of statements before a batch is flushed automatically.

      • getBatchSize

        int getBatchSize()
        Return the current batch size.

      • setGetGeneratedKeys

        void setGetGeneratedKeys​(boolean getGeneratedKeys)
        Specify if you want batched inserts to use getGeneratedKeys.

        By default batched inserts will try to use getGeneratedKeys if it is supported by the underlying jdbc driver and database.

        You may want to turn getGeneratedKeys off when you are inserting a large number of objects and you don't care about getting back the ids.

      • setFlushOnMixed

        void setFlushOnMixed​(boolean batchFlushOnMixed)
        By default when mixing UpdateSql (or CallableSql) with Beans the batch is automatically flushed when you change (between persisting beans and executing UpdateSql or CallableSql).

        If you want to execute both WITHOUT having the batch automatically flush you need to call this with batchFlushOnMixed = false.

        Note that UpdateSql and CallableSql are ALWAYS executed first (before the beans are executed). This is because the UpdateSql and CallableSql have already been bound to their PreparedStatements. The beans on the other hand have a 2 step process (delayed binding).

      • setFlushOnQuery

        void setFlushOnQuery​(boolean batchFlushOnQuery)
        By default executing a query will automatically flush any batched statements (persisted beans, executed UpdateSql etc).

        Calling this method with batchFlushOnQuery = false means that you can execute a query and the batch will not be automatically flushed.

      • isFlushOnQuery

        boolean isFlushOnQuery()
        Return true if the batch (of persisted beans or executed UpdateSql etc) should be flushed prior to executing a query.

        The default is for this to be true.

      • flush

        void flush()
    throws javax.persistence.PersistenceException
        The batch will be flushing automatically but you can use this to explicitly flush the batch if you like.

        Flushing occurs automatically when:

        • the batch size is reached
        • A query is executed on the same transaction
        • UpdateSql or CallableSql are mixed with bean save and delete
        • Transaction commit occurs
        • A getter method is called on a batched bean
        Throws:
        javax.persistence.PersistenceException

      • flushBatch

        @Deprecated
void flushBatch()
         throws javax.persistence.PersistenceException
        Deprecated.
        Deprecated - migrate to flush().

        flush() is preferred as it matches the JPA flush() method.

        Throws:
        javax.persistence.PersistenceException

      • getConnection

        Connection getConnection()
        Return the underlying Connection object.

        Useful where a Developer wishes to use the JDBC API directly. Note that the commit() rollback() and end() methods on the Transaction should still be used. Calling these methods on the Connection would be a big no no unless you know what you are doing.

        Examples of when a developer may wish to use the connection directly are: Savepoints, advanced CLOB BLOB use and advanced stored procedure calls.

      • addModification

        void addModification​(String tableName,
                     boolean inserts,
                     boolean updates,
                     boolean deletes)
        Add table modification information to the TransactionEvent.

        Use this in conjunction with getConnection() and raw JDBC.

        This effectively informs Ebean of the data that has been changed by the transaction and this information is normally automatically handled by Ebean when you save entity beans or use UpdateSql etc.

        If you use raw JDBC then you can use this method to inform Ebean for the tables that have been modified. Ebean uses this information to keep its caches in synch and maintain text indexes.

      • putUserObject

        void putUserObject​(String name,
                   Object value)
        Add an arbitrary user object to the transaction. The objects added have no impact on any internals of ebean and are solely meant as a convenient method push user information (although somewhat replaced by TransactionCallback).