diff --git a/docs/streams/upgrade-guide.html b/docs/streams/upgrade-guide.html
index b1af752754a..be7a95b69a8 100644
--- a/docs/streams/upgrade-guide.html
+++ b/docs/streams/upgrade-guide.html
@@ -86,6 +86,14 @@
More details about the new config StreamsConfig#TOPOLOGY_OPTIMIZATION
can be found in KIP-295.
+
+
+ The StreamsConfig
variable for configuration parameter "topology.optimization"
+ is renamed from TOPOLOGY_OPTIMIZATION
to TOPOLOGY_OPTIMIZATION_CONFIG
.
+ The old variable is deprecated. Note, that the parameter name itself is not affected.
+ (Cf. KIP-629.)
+
+
We added a new processing mode that improves application scalability using exactly-once guarantees
diff --git a/streams/src/main/java/org/apache/kafka/streams/StreamsConfig.java b/streams/src/main/java/org/apache/kafka/streams/StreamsConfig.java
index f9e75fe7fb9..af65fd05206 100644
--- a/streams/src/main/java/org/apache/kafka/streams/StreamsConfig.java
+++ b/streams/src/main/java/org/apache/kafka/streams/StreamsConfig.java
@@ -207,12 +207,12 @@ public class StreamsConfig extends AbstractConfig {
public static final String ADMIN_CLIENT_PREFIX = "admin.";
/**
- * Config value for parameter {@link #TOPOLOGY_OPTIMIZATION "topology.optimization"} for disabling topology optimization
+ * Config value for parameter {@link #TOPOLOGY_OPTIMIZATION_CONFIG "topology.optimization"} for disabling topology optimization
*/
public static final String NO_OPTIMIZATION = "none";
/**
- * Config value for parameter {@link #TOPOLOGY_OPTIMIZATION "topology.optimization"} for enabling topology optimization
+ * Config value for parameter {@link #TOPOLOGY_OPTIMIZATION_CONFIG "topology.optimization"} for enabling topology optimization
*/
public static final String OPTIMIZE = "all";
@@ -524,7 +524,7 @@ public class StreamsConfig extends AbstractConfig {
private static final String STATE_DIR_DOC = "Directory location for state store. This path must be unique for each streams instance sharing the same underlying filesystem.";
/** {@code topology.optimization} */
- public static final String TOPOLOGY_OPTIMIZATION = "topology.optimization";
+ public static final String TOPOLOGY_OPTIMIZATION_CONFIG = "topology.optimization";
private static final String TOPOLOGY_OPTIMIZATION_DOC = "A configuration telling Kafka Streams if it should optimize the topology, disabled by default";
/** {@code upgrade.from} */
@@ -552,6 +552,13 @@ public class StreamsConfig extends AbstractConfig {
private static final String PARTITION_GROUPER_CLASS_DOC = "Partition grouper class that implements the org.apache.kafka.streams.processor.PartitionGrouper
interface." +
" WARNING: This config is deprecated and will be removed in 3.0.0 release.";
+ /**
+ * {@code topology.optimization}
+ * @deprecated since 2.7; use {@link #TOPOLOGY_OPTIMIZATION_CONFIG} instead
+ */
+ @Deprecated
+ public static final String TOPOLOGY_OPTIMIZATION = TOPOLOGY_OPTIMIZATION_CONFIG;
+
private static final String[] NON_CONFIGURABLE_CONSUMER_DEFAULT_CONFIGS =
new String[] {ConsumerConfig.ENABLE_AUTO_COMMIT_CONFIG};
@@ -664,7 +671,7 @@ public class StreamsConfig extends AbstractConfig {
CommonClientConfigs.DEFAULT_SECURITY_PROTOCOL,
Importance.MEDIUM,
CommonClientConfigs.SECURITY_PROTOCOL_DOC)
- .define(TOPOLOGY_OPTIMIZATION,
+ .define(TOPOLOGY_OPTIMIZATION_CONFIG,
Type.STRING,
NO_OPTIMIZATION,
in(NO_OPTIMIZATION, OPTIMIZE),
diff --git a/streams/src/main/java/org/apache/kafka/streams/kstream/KStream.java b/streams/src/main/java/org/apache/kafka/streams/kstream/KStream.java
index bcc911c3af6..f5f02dc2e41 100644
--- a/streams/src/main/java/org/apache/kafka/streams/kstream/KStream.java
+++ b/streams/src/main/java/org/apache/kafka/streams/kstream/KStream.java
@@ -933,7 +933,7 @@ public interface KStream {
* For this case, all data of this stream will be redistributed through the repartitioning topic by writing all
* records to it, and rereading all records from it, such that the resulting {@link KTable} is partitioned
* correctly on its key.
- * Note that you cannot enable {@link StreamsConfig#TOPOLOGY_OPTIMIZATION} config for this case, because
+ * Note that you cannot enable {@link StreamsConfig#TOPOLOGY_OPTIMIZATION_CONFIG} config for this case, because
* repartition topics are considered transient and don't allow to recover the result {@link KTable} in cause of
* a failure; hence, a dedicated changelog topic is required to guarantee fault-tolerance.
*
@@ -959,7 +959,7 @@ public interface KStream {
* For this case, all data of this stream will be redistributed through the repartitioning topic by writing all
* records to it, and rereading all records from it, such that the resulting {@link KTable} is partitioned
* correctly on its key.
- * Note that you cannot enable {@link StreamsConfig#TOPOLOGY_OPTIMIZATION} config for this case, because
+ * Note that you cannot enable {@link StreamsConfig#TOPOLOGY_OPTIMIZATION_CONFIG} config for this case, because
* repartition topics are considered transient and don't allow to recover the result {@link KTable} in cause of
* a failure; hence, a dedicated changelog topic is required to guarantee fault-tolerance.
*
@@ -986,7 +986,7 @@ public interface KStream {
* For this case, all data of this stream will be redistributed through the repartitioning topic by writing all
* records to it, and rereading all records from it, such that the resulting {@link KTable} is partitioned
* correctly on its key.
- * Note that you cannot enable {@link StreamsConfig#TOPOLOGY_OPTIMIZATION} config for this case, because
+ * Note that you cannot enable {@link StreamsConfig#TOPOLOGY_OPTIMIZATION_CONFIG} config for this case, because
* repartition topics are considered transient and don't allow to recover the result {@link KTable} in cause of
* a failure; hence, a dedicated changelog topic is required to guarantee fault-tolerance.
*
@@ -1014,7 +1014,7 @@ public interface KStream {
* For this case, all data of this stream will be redistributed through the repartitioning topic by writing all
* records to it, and rereading all records from it, such that the resulting {@link KTable} is partitioned
* correctly on its key.
- * Note that you cannot enable {@link StreamsConfig#TOPOLOGY_OPTIMIZATION} config for this case, because
+ * Note that you cannot enable {@link StreamsConfig#TOPOLOGY_OPTIMIZATION_CONFIG} config for this case, because
* repartition topics are considered transient and don't allow to recover the result {@link KTable} in cause of
* a failure; hence, a dedicated changelog topic is required to guarantee fault-tolerance.
*
diff --git a/streams/src/main/java/org/apache/kafka/streams/kstream/internals/InternalStreamsBuilder.java b/streams/src/main/java/org/apache/kafka/streams/kstream/internals/InternalStreamsBuilder.java
index e6aa2afc0b4..f859db8edd0 100644
--- a/streams/src/main/java/org/apache/kafka/streams/kstream/internals/InternalStreamsBuilder.java
+++ b/streams/src/main/java/org/apache/kafka/streams/kstream/internals/InternalStreamsBuilder.java
@@ -317,7 +317,7 @@ public class InternalStreamsBuilder implements InternalNameProvider {
private void maybePerformOptimizations(final Properties props) {
- if (props != null && StreamsConfig.OPTIMIZE.equals(props.getProperty(StreamsConfig.TOPOLOGY_OPTIMIZATION))) {
+ if (props != null && StreamsConfig.OPTIMIZE.equals(props.getProperty(StreamsConfig.TOPOLOGY_OPTIMIZATION_CONFIG))) {
LOG.debug("Optimizing the Kafka Streams graph for repartition nodes");
optimizeKTableSourceTopics();
maybeOptimizeRepartitionOperations();
diff --git a/streams/src/test/java/org/apache/kafka/streams/StreamsBuilderTest.java b/streams/src/test/java/org/apache/kafka/streams/StreamsBuilderTest.java
index f77b3e586e8..bdc60a78330 100644
--- a/streams/src/test/java/org/apache/kafka/streams/StreamsBuilderTest.java
+++ b/streams/src/test/java/org/apache/kafka/streams/StreamsBuilderTest.java
@@ -397,7 +397,7 @@ public class StreamsBuilderTest {
final String topic = "topic";
builder.table(topic, Materialized.>as("store"));
final Properties props = StreamsTestUtils.getStreamsConfig();
- props.put(StreamsConfig.TOPOLOGY_OPTIMIZATION, StreamsConfig.OPTIMIZE);
+ props.put(StreamsConfig.TOPOLOGY_OPTIMIZATION_CONFIG, StreamsConfig.OPTIMIZE);
final Topology topology = builder.build(props);
final InternalTopologyBuilder internalTopologyBuilder = TopologyWrapper.getInternalTopologyBuilder(topology);
diff --git a/streams/src/test/java/org/apache/kafka/streams/StreamsConfigTest.java b/streams/src/test/java/org/apache/kafka/streams/StreamsConfigTest.java
index 92d2881320d..1745cc7a453 100644
--- a/streams/src/test/java/org/apache/kafka/streams/StreamsConfigTest.java
+++ b/streams/src/test/java/org/apache/kafka/streams/StreamsConfigTest.java
@@ -47,7 +47,7 @@ import static org.apache.kafka.common.IsolationLevel.READ_COMMITTED;
import static org.apache.kafka.common.IsolationLevel.READ_UNCOMMITTED;
import static org.apache.kafka.streams.StreamsConfig.EXACTLY_ONCE;
import static org.apache.kafka.streams.StreamsConfig.EXACTLY_ONCE_BETA;
-import static org.apache.kafka.streams.StreamsConfig.TOPOLOGY_OPTIMIZATION;
+import static org.apache.kafka.streams.StreamsConfig.TOPOLOGY_OPTIMIZATION_CONFIG;
import static org.apache.kafka.streams.StreamsConfig.adminClientPrefix;
import static org.apache.kafka.streams.StreamsConfig.consumerPrefix;
import static org.apache.kafka.streams.StreamsConfig.producerPrefix;
@@ -895,22 +895,22 @@ public class StreamsConfigTest {
@Test
public void shouldSpecifyNoOptimizationWhenNotExplicitlyAddedToConfigs() {
final String expectedOptimizeConfig = "none";
- final String actualOptimizedConifig = streamsConfig.getString(TOPOLOGY_OPTIMIZATION);
+ final String actualOptimizedConifig = streamsConfig.getString(TOPOLOGY_OPTIMIZATION_CONFIG);
assertEquals("Optimization should be \"none\"", expectedOptimizeConfig, actualOptimizedConifig);
}
@Test
public void shouldSpecifyOptimizationWhenNotExplicitlyAddedToConfigs() {
final String expectedOptimizeConfig = "all";
- props.put(TOPOLOGY_OPTIMIZATION, "all");
+ props.put(TOPOLOGY_OPTIMIZATION_CONFIG, "all");
final StreamsConfig config = new StreamsConfig(props);
- final String actualOptimizedConifig = config.getString(TOPOLOGY_OPTIMIZATION);
+ final String actualOptimizedConifig = config.getString(TOPOLOGY_OPTIMIZATION_CONFIG);
assertEquals("Optimization should be \"all\"", expectedOptimizeConfig, actualOptimizedConifig);
}
@Test(expected = ConfigException.class)
public void shouldThrowConfigExceptionWhenOptimizationConfigNotValueInRange() {
- props.put(TOPOLOGY_OPTIMIZATION, "maybe");
+ props.put(TOPOLOGY_OPTIMIZATION_CONFIG, "maybe");
new StreamsConfig(props);
}
diff --git a/streams/src/test/java/org/apache/kafka/streams/integration/KStreamRepartitionIntegrationTest.java b/streams/src/test/java/org/apache/kafka/streams/integration/KStreamRepartitionIntegrationTest.java
index 03410e057b2..a12bb854cbf 100644
--- a/streams/src/test/java/org/apache/kafka/streams/integration/KStreamRepartitionIntegrationTest.java
+++ b/streams/src/test/java/org/apache/kafka/streams/integration/KStreamRepartitionIntegrationTest.java
@@ -131,7 +131,7 @@ public class KStreamRepartitionIntegrationTest {
streamsConfiguration.put(StreamsConfig.COMMIT_INTERVAL_MS_CONFIG, 100);
streamsConfiguration.put(StreamsConfig.DEFAULT_KEY_SERDE_CLASS_CONFIG, Serdes.Integer().getClass());
streamsConfiguration.put(StreamsConfig.DEFAULT_VALUE_SERDE_CLASS_CONFIG, Serdes.String().getClass());
- streamsConfiguration.put(StreamsConfig.TOPOLOGY_OPTIMIZATION, topologyOptimization);
+ streamsConfiguration.put(StreamsConfig.TOPOLOGY_OPTIMIZATION_CONFIG, topologyOptimization);
}
@After
diff --git a/streams/src/test/java/org/apache/kafka/streams/integration/KTableKTableForeignKeyJoinIntegrationTest.java b/streams/src/test/java/org/apache/kafka/streams/integration/KTableKTableForeignKeyJoinIntegrationTest.java
index 5ebd9ef8100..d774fa627fa 100644
--- a/streams/src/test/java/org/apache/kafka/streams/integration/KTableKTableForeignKeyJoinIntegrationTest.java
+++ b/streams/src/test/java/org/apache/kafka/streams/integration/KTableKTableForeignKeyJoinIntegrationTest.java
@@ -95,7 +95,7 @@ public class KTableKTableForeignKeyJoinIntegrationTest {
mkEntry(StreamsConfig.APPLICATION_ID_CONFIG, "app-" + safeTestName),
mkEntry(StreamsConfig.BOOTSTRAP_SERVERS_CONFIG, "asdf:0000"),
mkEntry(StreamsConfig.STATE_DIR_CONFIG, TestUtils.tempDirectory().getPath()),
- mkEntry(StreamsConfig.TOPOLOGY_OPTIMIZATION, optimization)
+ mkEntry(StreamsConfig.TOPOLOGY_OPTIMIZATION_CONFIG, optimization)
));
}
diff --git a/streams/src/test/java/org/apache/kafka/streams/integration/LagFetchIntegrationTest.java b/streams/src/test/java/org/apache/kafka/streams/integration/LagFetchIntegrationTest.java
index 2e6d6c06f66..d288c3f9ee4 100644
--- a/streams/src/test/java/org/apache/kafka/streams/integration/LagFetchIntegrationTest.java
+++ b/streams/src/test/java/org/apache/kafka/streams/integration/LagFetchIntegrationTest.java
@@ -156,7 +156,7 @@ public class LagFetchIntegrationTest {
props.put(StreamsConfig.InternalConfig.INTERNAL_TASK_ASSIGNOR_CLASS, FallbackPriorTaskAssignor.class.getName());
props.put(StreamsConfig.APPLICATION_SERVER_CONFIG, "localhost:" + i);
props.put(StreamsConfig.CLIENT_ID_CONFIG, "instance-" + i);
- props.put(StreamsConfig.TOPOLOGY_OPTIMIZATION, optimization);
+ props.put(StreamsConfig.TOPOLOGY_OPTIMIZATION_CONFIG, optimization);
props.put(StreamsConfig.NUM_STANDBY_REPLICAS_CONFIG, 1);
props.put(StreamsConfig.STATE_DIR_CONFIG, TestUtils.tempDirectory(stateStoreName + i).getAbsolutePath());
diff --git a/streams/src/test/java/org/apache/kafka/streams/integration/OptimizedKTableIntegrationTest.java b/streams/src/test/java/org/apache/kafka/streams/integration/OptimizedKTableIntegrationTest.java
index fbf40e2835e..2507436f65a 100644
--- a/streams/src/test/java/org/apache/kafka/streams/integration/OptimizedKTableIntegrationTest.java
+++ b/streams/src/test/java/org/apache/kafka/streams/integration/OptimizedKTableIntegrationTest.java
@@ -180,7 +180,7 @@ public class OptimizedKTableIntegrationTest {
private Properties streamsConfiguration() {
final String safeTestName = safeUniqueTestName(getClass(), testName);
final Properties config = new Properties();
- config.put(StreamsConfig.TOPOLOGY_OPTIMIZATION, StreamsConfig.OPTIMIZE);
+ config.put(StreamsConfig.TOPOLOGY_OPTIMIZATION_CONFIG, StreamsConfig.OPTIMIZE);
config.put(StreamsConfig.APPLICATION_ID_CONFIG, "app-" + safeTestName);
config.put(StreamsConfig.APPLICATION_SERVER_CONFIG, "localhost:" + (++port));
config.put(StreamsConfig.BOOTSTRAP_SERVERS_CONFIG, cluster.bootstrapServers());
diff --git a/streams/src/test/java/org/apache/kafka/streams/integration/RestoreIntegrationTest.java b/streams/src/test/java/org/apache/kafka/streams/integration/RestoreIntegrationTest.java
index faac172b639..0dad8c31233 100644
--- a/streams/src/test/java/org/apache/kafka/streams/integration/RestoreIntegrationTest.java
+++ b/streams/src/test/java/org/apache/kafka/streams/integration/RestoreIntegrationTest.java
@@ -135,7 +135,7 @@ public class RestoreIntegrationTest {
final StreamsBuilder builder = new StreamsBuilder();
final Properties props = props();
- props.put(StreamsConfig.TOPOLOGY_OPTIMIZATION, StreamsConfig.OPTIMIZE);
+ props.put(StreamsConfig.TOPOLOGY_OPTIMIZATION_CONFIG, StreamsConfig.OPTIMIZE);
// restoring from 1000 to 4000 (committed), and then process from 4000 to 5000 on each of the two partitions
final int offsetLimitDelta = 1000;
diff --git a/streams/src/test/java/org/apache/kafka/streams/integration/StandbyTaskCreationIntegrationTest.java b/streams/src/test/java/org/apache/kafka/streams/integration/StandbyTaskCreationIntegrationTest.java
index 3bf2d8ffaa4..79f14b5d2ec 100644
--- a/streams/src/test/java/org/apache/kafka/streams/integration/StandbyTaskCreationIntegrationTest.java
+++ b/streams/src/test/java/org/apache/kafka/streams/integration/StandbyTaskCreationIntegrationTest.java
@@ -127,9 +127,9 @@ public class StandbyTaskCreationIntegrationTest {
@Test
public void shouldCreateStandByTasksForMaterializedAndOptimizedSourceTables() throws Exception {
final Properties streamsConfiguration1 = streamsConfiguration();
- streamsConfiguration1.put(StreamsConfig.TOPOLOGY_OPTIMIZATION, StreamsConfig.OPTIMIZE);
+ streamsConfiguration1.put(StreamsConfig.TOPOLOGY_OPTIMIZATION_CONFIG, StreamsConfig.OPTIMIZE);
final Properties streamsConfiguration2 = streamsConfiguration();
- streamsConfiguration2.put(StreamsConfig.TOPOLOGY_OPTIMIZATION, StreamsConfig.OPTIMIZE);
+ streamsConfiguration2.put(StreamsConfig.TOPOLOGY_OPTIMIZATION_CONFIG, StreamsConfig.OPTIMIZE);
final StreamsBuilder builder = new StreamsBuilder();
builder.table(INPUT_TOPIC, Consumed.with(Serdes.Integer(), Serdes.Integer()), Materialized.as("source-table"));
diff --git a/streams/src/test/java/org/apache/kafka/streams/integration/StoreQueryIntegrationTest.java b/streams/src/test/java/org/apache/kafka/streams/integration/StoreQueryIntegrationTest.java
index fb7310178ff..ef34def3221 100644
--- a/streams/src/test/java/org/apache/kafka/streams/integration/StoreQueryIntegrationTest.java
+++ b/streams/src/test/java/org/apache/kafka/streams/integration/StoreQueryIntegrationTest.java
@@ -389,7 +389,7 @@ public class StoreQueryIntegrationTest {
private Properties streamsConfiguration() {
final String safeTestName = safeUniqueTestName(getClass(), testName);
final Properties config = new Properties();
- config.put(StreamsConfig.TOPOLOGY_OPTIMIZATION, StreamsConfig.OPTIMIZE);
+ config.put(StreamsConfig.TOPOLOGY_OPTIMIZATION_CONFIG, StreamsConfig.OPTIMIZE);
config.put(StreamsConfig.APPLICATION_ID_CONFIG, "app-" + safeTestName);
config.put(StreamsConfig.APPLICATION_SERVER_CONFIG, "localhost:" + (++port));
config.put(StreamsConfig.BOOTSTRAP_SERVERS_CONFIG, cluster.bootstrapServers());
diff --git a/streams/src/test/java/org/apache/kafka/streams/kstream/RepartitionTopicNamingTest.java b/streams/src/test/java/org/apache/kafka/streams/kstream/RepartitionTopicNamingTest.java
index 5638dbdc5cd..c286b142459 100644
--- a/streams/src/test/java/org/apache/kafka/streams/kstream/RepartitionTopicNamingTest.java
+++ b/streams/src/test/java/org/apache/kafka/streams/kstream/RepartitionTopicNamingTest.java
@@ -192,7 +192,7 @@ public class RepartitionTopicNamingTest {
kGroupedStream.windowedBy(TimeWindows.of(Duration.ofMillis(10L))).count();
kGroupedStream.windowedBy(TimeWindows.of(Duration.ofMillis(30L))).count();
final Properties properties = new Properties();
- properties.put(StreamsConfig.TOPOLOGY_OPTIMIZATION, StreamsConfig.OPTIMIZE);
+ properties.put(StreamsConfig.TOPOLOGY_OPTIMIZATION_CONFIG, StreamsConfig.OPTIMIZE);
final Topology topology = builder.build(properties);
assertThat(getCountOfRepartitionTopicsFound(topology.describe().toString(), repartitionTopicPattern), is(1));
}
@@ -224,7 +224,7 @@ public class RepartitionTopicNamingTest {
public void shouldPassWithSameRepartitionTopicNameUsingSameKGroupedStreamOptimized() {
final StreamsBuilder builder = new StreamsBuilder();
final Properties properties = new Properties();
- properties.put(StreamsConfig.TOPOLOGY_OPTIMIZATION, StreamsConfig.OPTIMIZE);
+ properties.put(StreamsConfig.TOPOLOGY_OPTIMIZATION_CONFIG, StreamsConfig.OPTIMIZE);
final KGroupedStream kGroupedStream = builder.stream("topic")
.selectKey((k, v) -> k)
.groupByKey(Grouped.as("grouping"));
@@ -500,7 +500,7 @@ public class RepartitionTopicNamingTest {
final Properties properties = new Properties();
- properties.put(StreamsConfig.TOPOLOGY_OPTIMIZATION, optimizationConfig);
+ properties.put(StreamsConfig.TOPOLOGY_OPTIMIZATION_CONFIG, optimizationConfig);
return builder.build(properties);
}
diff --git a/streams/src/test/java/org/apache/kafka/streams/kstream/internals/CogroupedKStreamImplTest.java b/streams/src/test/java/org/apache/kafka/streams/kstream/internals/CogroupedKStreamImplTest.java
index d6195bea405..d780fa7d5a1 100644
--- a/streams/src/test/java/org/apache/kafka/streams/kstream/internals/CogroupedKStreamImplTest.java
+++ b/streams/src/test/java/org/apache/kafka/streams/kstream/internals/CogroupedKStreamImplTest.java
@@ -379,7 +379,7 @@ public class CogroupedKStreamImplTest {
@Test
public void shouldInsertRepartitionsTopicForUpstreamKeyModificationWithGroupedReusedInSameCogroupsWithOptimization() {
final Properties properties = new Properties();
- properties.setProperty(StreamsConfig.TOPOLOGY_OPTIMIZATION, StreamsConfig.OPTIMIZE);
+ properties.setProperty(StreamsConfig.TOPOLOGY_OPTIMIZATION_CONFIG, StreamsConfig.OPTIMIZE);
final StreamsBuilder builder = new StreamsBuilder();
final KStream stream1 = builder.stream("one", stringConsumed);
@@ -528,7 +528,7 @@ public class CogroupedKStreamImplTest {
final StreamsBuilder builder = new StreamsBuilder();
final Properties properties = new Properties();
- properties.setProperty(StreamsConfig.TOPOLOGY_OPTIMIZATION, StreamsConfig.OPTIMIZE);
+ properties.setProperty(StreamsConfig.TOPOLOGY_OPTIMIZATION_CONFIG, StreamsConfig.OPTIMIZE);
final KStream stream1 = builder.stream("one", stringConsumed);
final KStream stream2 = builder.stream("two", stringConsumed);
@@ -653,7 +653,7 @@ public class CogroupedKStreamImplTest {
final StreamsBuilder builder = new StreamsBuilder();
final Properties properties = new Properties();
- properties.setProperty(StreamsConfig.TOPOLOGY_OPTIMIZATION, StreamsConfig.OPTIMIZE);
+ properties.setProperty(StreamsConfig.TOPOLOGY_OPTIMIZATION_CONFIG, StreamsConfig.OPTIMIZE);
final KStream stream1 = builder.stream("one", stringConsumed);
final KStream stream2 = builder.stream("two", stringConsumed);
@@ -707,7 +707,7 @@ public class CogroupedKStreamImplTest {
final StreamsBuilder builder = new StreamsBuilder();
final Properties properties = new Properties();
- properties.setProperty(StreamsConfig.TOPOLOGY_OPTIMIZATION, StreamsConfig.OPTIMIZE);
+ properties.setProperty(StreamsConfig.TOPOLOGY_OPTIMIZATION_CONFIG, StreamsConfig.OPTIMIZE);
final KStream stream1 = builder.stream("one", stringConsumed);
final KStream stream2 = builder.stream("two", stringConsumed);
diff --git a/streams/src/test/java/org/apache/kafka/streams/kstream/internals/KStreamKStreamJoinTest.java b/streams/src/test/java/org/apache/kafka/streams/kstream/internals/KStreamKStreamJoinTest.java
index 56190eed3aa..4dd7f70d6bd 100644
--- a/streams/src/test/java/org/apache/kafka/streams/kstream/internals/KStreamKStreamJoinTest.java
+++ b/streams/src/test/java/org/apache/kafka/streams/kstream/internals/KStreamKStreamJoinTest.java
@@ -84,7 +84,7 @@ public class KStreamKStreamJoinTest {
public void shouldReuseRepartitionTopicWithGeneratedName() {
final StreamsBuilder builder = new StreamsBuilder();
final Properties props = new Properties();
- props.put(StreamsConfig.TOPOLOGY_OPTIMIZATION, StreamsConfig.NO_OPTIMIZATION);
+ props.put(StreamsConfig.TOPOLOGY_OPTIMIZATION_CONFIG, StreamsConfig.NO_OPTIMIZATION);
final KStream stream1 = builder.stream("topic", Consumed.with(Serdes.String(), Serdes.String()));
final KStream stream2 = builder.stream("topic2", Consumed.with(Serdes.String(), Serdes.String()));
final KStream stream3 = builder.stream("topic3", Consumed.with(Serdes.String(), Serdes.String()));
@@ -98,7 +98,7 @@ public class KStreamKStreamJoinTest {
public void shouldCreateRepartitionTopicsWithUserProvidedName() {
final StreamsBuilder builder = new StreamsBuilder();
final Properties props = new Properties();
- props.put(StreamsConfig.TOPOLOGY_OPTIMIZATION, StreamsConfig.NO_OPTIMIZATION);
+ props.put(StreamsConfig.TOPOLOGY_OPTIMIZATION_CONFIG, StreamsConfig.NO_OPTIMIZATION);
final KStream stream1 = builder.stream("topic", Consumed.with(Serdes.String(), Serdes.String()));
final KStream stream2 = builder.stream("topic2", Consumed.with(Serdes.String(), Serdes.String()));
final KStream stream3 = builder.stream("topic3", Consumed.with(Serdes.String(), Serdes.String()));
diff --git a/streams/src/test/java/org/apache/kafka/streams/kstream/internals/KStreamKTableJoinTest.java b/streams/src/test/java/org/apache/kafka/streams/kstream/internals/KStreamKTableJoinTest.java
index a5368cde9c3..c5067797595 100644
--- a/streams/src/test/java/org/apache/kafka/streams/kstream/internals/KStreamKTableJoinTest.java
+++ b/streams/src/test/java/org/apache/kafka/streams/kstream/internals/KStreamKTableJoinTest.java
@@ -118,7 +118,7 @@ public class KStreamKTableJoinTest {
public void shouldReuseRepartitionTopicWithGeneratedName() {
final StreamsBuilder builder = new StreamsBuilder();
final Properties props = new Properties();
- props.put(StreamsConfig.TOPOLOGY_OPTIMIZATION, StreamsConfig.NO_OPTIMIZATION);
+ props.put(StreamsConfig.TOPOLOGY_OPTIMIZATION_CONFIG, StreamsConfig.NO_OPTIMIZATION);
final KStream streamA = builder.stream("topic", Consumed.with(Serdes.String(), Serdes.String()));
final KTable tableB = builder.table("topic2", Consumed.with(Serdes.String(), Serdes.String()));
final KTable tableC = builder.table("topic3", Consumed.with(Serdes.String(), Serdes.String()));
@@ -133,7 +133,7 @@ public class KStreamKTableJoinTest {
public void shouldCreateRepartitionTopicsWithUserProvidedName() {
final StreamsBuilder builder = new StreamsBuilder();
final Properties props = new Properties();
- props.put(StreamsConfig.TOPOLOGY_OPTIMIZATION, StreamsConfig.NO_OPTIMIZATION);
+ props.put(StreamsConfig.TOPOLOGY_OPTIMIZATION_CONFIG, StreamsConfig.NO_OPTIMIZATION);
final KStream streamA = builder.stream("topic", Consumed.with(Serdes.String(), Serdes.String()));
final KTable tableB = builder.table("topic2", Consumed.with(Serdes.String(), Serdes.String()));
final KTable tableC = builder.table("topic3", Consumed.with(Serdes.String(), Serdes.String()));
diff --git a/streams/src/test/java/org/apache/kafka/streams/kstream/internals/graph/StreamsGraphTest.java b/streams/src/test/java/org/apache/kafka/streams/kstream/internals/graph/StreamsGraphTest.java
index d8e8303f633..96016027434 100644
--- a/streams/src/test/java/org/apache/kafka/streams/kstream/internals/graph/StreamsGraphTest.java
+++ b/streams/src/test/java/org/apache/kafka/streams/kstream/internals/graph/StreamsGraphTest.java
@@ -89,7 +89,7 @@ public class StreamsGraphTest {
final Properties properties = new Properties();
properties.setProperty(StreamsConfig.APPLICATION_ID_CONFIG, "test-application");
properties.setProperty(StreamsConfig.BOOTSTRAP_SERVERS_CONFIG, "localhost:9092");
- properties.setProperty(StreamsConfig.TOPOLOGY_OPTIMIZATION, StreamsConfig.OPTIMIZE);
+ properties.setProperty(StreamsConfig.TOPOLOGY_OPTIMIZATION_CONFIG, StreamsConfig.OPTIMIZE);
final StreamsBuilder builder = new StreamsBuilder();
final KStream inputStream = builder.stream("inputTopic");
@@ -118,7 +118,7 @@ public class StreamsGraphTest {
final Properties properties = new Properties();
properties.setProperty(StreamsConfig.APPLICATION_ID_CONFIG, "test-application");
properties.setProperty(StreamsConfig.BOOTSTRAP_SERVERS_CONFIG, "localhost:9092");
- properties.setProperty(StreamsConfig.TOPOLOGY_OPTIMIZATION, StreamsConfig.OPTIMIZE);
+ properties.setProperty(StreamsConfig.TOPOLOGY_OPTIMIZATION_CONFIG, StreamsConfig.OPTIMIZE);
final StreamsBuilder builder = new StreamsBuilder();
initializer = () -> "";
@@ -222,7 +222,7 @@ public class StreamsGraphTest {
.to("output_topic");
final Properties properties = new Properties();
- properties.setProperty(StreamsConfig.TOPOLOGY_OPTIMIZATION, StreamsConfig.OPTIMIZE);
+ properties.setProperty(StreamsConfig.TOPOLOGY_OPTIMIZATION_CONFIG, StreamsConfig.OPTIMIZE);
final Topology topology = streamsBuilder.build(properties);
assertEquals(expectedMergeOptimizedTopology, topology.describe().toString());
@@ -242,7 +242,7 @@ public class StreamsGraphTest {
final StreamsBuilder builder = new StreamsBuilder();
final Properties properties = new Properties();
- properties.put(StreamsConfig.TOPOLOGY_OPTIMIZATION, optimizeConfig);
+ properties.put(StreamsConfig.TOPOLOGY_OPTIMIZATION_CONFIG, optimizeConfig);
final KStream inputStream = builder.stream("input");
final KStream mappedKeyStream = inputStream.selectKey((k, v) -> k + v);
@@ -259,7 +259,7 @@ public class StreamsGraphTest {
final StreamsBuilder builder = new StreamsBuilder();
final Properties properties = new Properties();
- properties.put(StreamsConfig.TOPOLOGY_OPTIMIZATION, optimizeConfig);
+ properties.put(StreamsConfig.TOPOLOGY_OPTIMIZATION_CONFIG, optimizeConfig);
final KStream inputStream = builder.stream("input");
final KStream mappedKeyStream = inputStream.selectKey((k, v) -> k + v).through("through-topic");
@@ -274,7 +274,7 @@ public class StreamsGraphTest {
private Topology getTopologyWithRepartitionOperation(final String optimizeConfig) {
final StreamsBuilder builder = new StreamsBuilder();
final Properties properties = new Properties();
- properties.put(StreamsConfig.TOPOLOGY_OPTIMIZATION, optimizeConfig);
+ properties.put(StreamsConfig.TOPOLOGY_OPTIMIZATION_CONFIG, optimizeConfig);
final KStream inputStream = builder.stream("input").selectKey((k, v) -> k + v);
diff --git a/streams/src/test/java/org/apache/kafka/streams/processor/internals/RepartitionOptimizingTest.java b/streams/src/test/java/org/apache/kafka/streams/processor/internals/RepartitionOptimizingTest.java
index 3dcb2a3f4e1..681119829f4 100644
--- a/streams/src/test/java/org/apache/kafka/streams/processor/internals/RepartitionOptimizingTest.java
+++ b/streams/src/test/java/org/apache/kafka/streams/processor/internals/RepartitionOptimizingTest.java
@@ -200,7 +200,7 @@ public class RepartitionOptimizingTest {
.withOtherValueSerde(Serdes.Long()))
.to(JOINED_TOPIC, Produced.as("join-to"));
- streamsConfiguration.setProperty(StreamsConfig.TOPOLOGY_OPTIMIZATION, optimizationConfig);
+ streamsConfiguration.setProperty(StreamsConfig.TOPOLOGY_OPTIMIZATION_CONFIG, optimizationConfig);
final Topology topology = builder.build(streamsConfiguration);
topologyTestDriver = new TopologyTestDriver(topology, streamsConfiguration);
diff --git a/streams/src/test/java/org/apache/kafka/streams/processor/internals/RepartitionWithMergeOptimizingTest.java b/streams/src/test/java/org/apache/kafka/streams/processor/internals/RepartitionWithMergeOptimizingTest.java
index bd2656cb4b0..b50ef8ecee0 100644
--- a/streams/src/test/java/org/apache/kafka/streams/processor/internals/RepartitionWithMergeOptimizingTest.java
+++ b/streams/src/test/java/org/apache/kafka/streams/processor/internals/RepartitionWithMergeOptimizingTest.java
@@ -120,7 +120,7 @@ public class RepartitionWithMergeOptimizingTest {
private void runTest(final String optimizationConfig, final int expectedNumberRepartitionTopics) {
- streamsConfiguration.setProperty(StreamsConfig.TOPOLOGY_OPTIMIZATION, optimizationConfig);
+ streamsConfiguration.setProperty(StreamsConfig.TOPOLOGY_OPTIMIZATION_CONFIG, optimizationConfig);
final StreamsBuilder builder = new StreamsBuilder();
diff --git a/streams/src/test/java/org/apache/kafka/streams/processor/internals/StreamsPartitionAssignorTest.java b/streams/src/test/java/org/apache/kafka/streams/processor/internals/StreamsPartitionAssignorTest.java
index b3d4169cd9f..a2c2b4602e9 100644
--- a/streams/src/test/java/org/apache/kafka/streams/processor/internals/StreamsPartitionAssignorTest.java
+++ b/streams/src/test/java/org/apache/kafka/streams/processor/internals/StreamsPartitionAssignorTest.java
@@ -1941,7 +1941,7 @@ public class StreamsPartitionAssignorTest {
createDefaultMockTaskManager();
EasyMock.expect(taskManager.mainConsumer()).andStubReturn(consumerClient);
- configurePartitionAssignorWith(singletonMap(StreamsConfig.TOPOLOGY_OPTIMIZATION, StreamsConfig.OPTIMIZE));
+ configurePartitionAssignorWith(singletonMap(StreamsConfig.TOPOLOGY_OPTIMIZATION_CONFIG, StreamsConfig.OPTIMIZE));
overwriteInternalTopicManagerWithMock(false);
EasyMock.expect(consumerClient.committed(changelogs))
diff --git a/streams/streams-scala/src/test/scala/org/apache/kafka/streams/scala/TopologyTest.scala b/streams/streams-scala/src/test/scala/org/apache/kafka/streams/scala/TopologyTest.scala
index 3107db6f1f2..3cf0059b409 100644
--- a/streams/streams-scala/src/test/scala/org/apache/kafka/streams/scala/TopologyTest.scala
+++ b/streams/streams-scala/src/test/scala/org/apache/kafka/streams/scala/TopologyTest.scala
@@ -335,10 +335,10 @@ class TopologyTest {
def shouldBuildIdenticalTopologyInJavaNScalaProperties(): Unit = {
val props = new Properties()
- props.put(StreamsConfig.TOPOLOGY_OPTIMIZATION, StreamsConfig.OPTIMIZE)
+ props.put(StreamsConfig.TOPOLOGY_OPTIMIZATION_CONFIG, StreamsConfig.OPTIMIZE)
val propsNoOptimization = new Properties()
- propsNoOptimization.put(StreamsConfig.TOPOLOGY_OPTIMIZATION, StreamsConfig.NO_OPTIMIZATION)
+ propsNoOptimization.put(StreamsConfig.TOPOLOGY_OPTIMIZATION_CONFIG, StreamsConfig.NO_OPTIMIZATION)
val AGGREGATION_TOPIC = "aggregationTopic"
val REDUCE_TOPIC = "reduceTopic"