org.apache.flink.connectors.test.common.testsuites

Class SourceTestSuiteBase<T>

java.lang.Object

org.apache.flink.connectors.test.common.testsuites.SourceTestSuiteBase<T>

@ExtendWith(value={ConnectorTestingExtension.class,TestLoggerExtension.class,TestCaseInvocationContextProvider.class})
 @TestInstance(value=PER_CLASS)
 @Experimental
public abstract class SourceTestSuiteBase<T>
extends Object

Base class for all test suites.

All cases should have well-descriptive JavaDoc, including:

• What's the purpose of this case
• Simple description of how this case works
• Condition to fulfill in order to pass this case
• Requirement of running this case

Constructor Summary

Constructors

Constructor and Description
SourceTestSuiteBase()

Method Summary

Modifier and Type	Method and Description
protected List<T>	generateAndWriteTestData(int splitIndex, ExternalContext<T> externalContext) Generate a set of test records and write it to the given split writer.
void	testIdleReader(TestEnvironment testEnv, ExternalContext<T> externalContext) Test connector source with an idle reader.
void	testMultipleSplits(TestEnvironment testEnv, ExternalContext<T> externalContext) Test connector source with multiple splits in the external system
void	testSourceSingleSplit(TestEnvironment testEnv, ExternalContext<T> externalContext) Test connector source with only one split in the external system.
void	testTaskManagerFailure(TestEnvironment testEnv, ExternalContext<T> externalContext, ClusterControllable controller) Test connector source with task manager failover.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

SourceTestSuiteBase

public SourceTestSuiteBase()

Method Detail

testSourceSingleSplit

@TestTemplate
 @DisplayName(value="Test source with single split")
public void testSourceSingleSplit(TestEnvironment testEnv,
                                                                                                     ExternalContext<T> externalContext)
                                                                                              throws Exception

Test connector source with only one split in the external system.

This test will create one split in the external system, write test data into it, and consume back via a Flink job with 1 parallelism.

The number and order of records consumed by Flink need to be identical to the test data written to the external system in order to pass this test.

A bounded source is required for this test.

Throws:

Exception

testMultipleSplits

@TestTemplate
 @DisplayName(value="Test source with multiple splits")
public void testMultipleSplits(TestEnvironment testEnv,
                                                                                                     ExternalContext<T> externalContext)
                                                                                              throws Exception

Test connector source with multiple splits in the external system

This test will create 4 splits in the external system, write test data to all splits, and consume back via a Flink job with 4 parallelism.

The number and order of records in each split consumed by Flink need to be identical to the test data written into the external system to pass this test. There's no requirement for record order across splits.

A bounded source is required for this test.

Throws:

Exception

testIdleReader

@TestTemplate
 @DisplayName(value="Test source with at least one idle parallelism")
public void testIdleReader(TestEnvironment testEnv,
                                                                                                               ExternalContext<T> externalContext)
                                                                                                        throws Exception

Test connector source with an idle reader.

This test will create 4 split in the external system, write test data to all splits, and consume back via a Flink job with 5 parallelism, so at least one parallelism / source reader will be idle (assigned with no splits). If the split enumerator of the source doesn't signal NoMoreSplitsEvent to the idle source reader, the Flink job will never spin to FINISHED state.

The number and order of records in each split consumed by Flink need to be identical to the test data written into the external system to pass this test. There's no requirement for record order across splits.

A bounded source is required for this test.

Throws:

Exception

testTaskManagerFailure

@TestTemplate
 @DisplayName(value="Test TaskManager failure")
public void testTaskManagerFailure(TestEnvironment testEnv,
                                                                                                 ExternalContext<T> externalContext,
                                                                                                 ClusterControllable controller)
                                                                                          throws Exception

Test connector source with task manager failover.

This test will create 1 split in the external system, write test record set A into the split, restart task manager to trigger job failover, write test record set B into the split, and terminate the Flink job finally.

The number and order of records consumed by Flink should be identical to A before the failover and B after the failover in order to pass the test.

An unbounded source is required for this test, since TaskManager failover will be triggered in the middle of the test.

Throws:

Exception

generateAndWriteTestData

protected List<T> generateAndWriteTestData(int splitIndex,
                                           ExternalContext<T> externalContext)

Generate a set of test records and write it to the given split writer.

Parameters:

externalContext - External context

Returns:

List of generated test records