The Connection.setSavepoint method sets a savepoint within the current transaction. The Connection.rollback method is overloaded to take a savepoint argument.

The code example below inserts a row into a table, sets the savepoint svpt1, and then inserts a second row. When the transaction is later rolled back to svpt1, the second insertion is undone, but the first insertion remains intact. In other words, when the transaction is committed, only the row containing '1' will be added to TABLE1. conn.setAutoCommit(false); // Autocommit must be off to use savepoints. Statement stmt = conn.createStatement(); int rows = stmt.executeUpdate("INSERT INTO TABLE1 (COL1) VALUES(1)"); // set savepoint Savepoint svpt1 = conn.setSavepoint("S1"); rows = stmt.executeUpdate("INSERT INTO TABLE1 (COL1) VALUES (2)"); ... conn.rollback(svpt1); ... conn.commit();

The method Connection.releaseSavepoint takes a Savepoint object as a parameter and removes it from the current transaction. Once a savepoint has been released, attempting to reference it in a rollback operation will cause an SQLException to be thrown.

Any savepoints that have been created in a transaction are automatically released and become invalid when the transaction is committed or when the entire transaction is rolled back.

Rolling a transaction back to a savepoint automatically releases and makes invalid any other savepoints created after the savepoint in question.

The savepoint cannot be set within a batch of statements to enable partial recovery. If a savepoint is set any time before the method executeBatch is called, it is set before any of the statements that have been added to the batch are executed.

A savepoint can be reused after it has been released explicitly (by issuing a release of the savepoint) or implicitly (by issuing a connection commit/rollback to that savepoint or to a savepoint declared earlier than that savepoint).

It is possible to nest savepoints, but only in an embedded environment.