1
0

[HUDI-4273] Support inline schedule clustering for Flink stream (#5890)

* [HUDI-4273] Support inline schedule clustering for Flink stream

* delete deprecated clustering plan strategy and add clustering ITTest
This commit is contained in:
Zhaojing Yu
2022-06-24 11:28:06 +08:00
committed by GitHub
parent af9f09047d
commit 6456bd3a51
29 changed files with 1116 additions and 385 deletions

View File

@@ -70,9 +70,6 @@ public abstract class ClusteringPlanStrategy<T extends HoodieRecordPayload,I,K,O
String sparkSizeBasedClassName = HoodieClusteringConfig.SPARK_SIZED_BASED_CLUSTERING_PLAN_STRATEGY;
String sparkSelectedPartitionsClassName = "org.apache.hudi.client.clustering.plan.strategy.SparkSelectedPartitionsClusteringPlanStrategy";
String sparkRecentDaysClassName = "org.apache.hudi.client.clustering.plan.strategy.SparkRecentDaysClusteringPlanStrategy";
String flinkSizeBasedClassName = HoodieClusteringConfig.FLINK_SIZED_BASED_CLUSTERING_PLAN_STRATEGY;
String flinkSelectedPartitionsClassName = "org.apache.hudi.client.clustering.plan.strategy.FlinkSelectedPartitionsClusteringPlanStrategy";
String flinkRecentDaysClassName = "org.apache.hudi.client.clustering.plan.strategy.FlinkRecentDaysClusteringPlanStrategy";
String javaSelectedPartitionClassName = "org.apache.hudi.client.clustering.plan.strategy.JavaRecentDaysClusteringPlanStrategy";
String javaSizeBasedClassName = HoodieClusteringConfig.JAVA_SIZED_BASED_CLUSTERING_PLAN_STRATEGY;
@@ -85,14 +82,6 @@ public abstract class ClusteringPlanStrategy<T extends HoodieRecordPayload,I,K,O
config.setValue(HoodieClusteringConfig.PLAN_PARTITION_FILTER_MODE_NAME, ClusteringPlanPartitionFilterMode.SELECTED_PARTITIONS.name());
LOG.warn(String.format(logStr, className, sparkSizeBasedClassName, HoodieClusteringConfig.PLAN_PARTITION_FILTER_MODE_NAME.key(), ClusteringPlanPartitionFilterMode.SELECTED_PARTITIONS.name()));
return sparkSizeBasedClassName;
} else if (flinkRecentDaysClassName.equals(className)) {
config.setValue(HoodieClusteringConfig.PLAN_PARTITION_FILTER_MODE_NAME, ClusteringPlanPartitionFilterMode.RECENT_DAYS.name());
LOG.warn(String.format(logStr, className, sparkSizeBasedClassName, HoodieClusteringConfig.PLAN_PARTITION_FILTER_MODE_NAME.key(), ClusteringPlanPartitionFilterMode.RECENT_DAYS.name()));
return flinkSizeBasedClassName;
} else if (flinkSelectedPartitionsClassName.equals(className)) {
config.setValue(HoodieClusteringConfig.PLAN_PARTITION_FILTER_MODE_NAME, ClusteringPlanPartitionFilterMode.SELECTED_PARTITIONS.name());
LOG.warn(String.format(logStr, className, sparkSizeBasedClassName, HoodieClusteringConfig.PLAN_PARTITION_FILTER_MODE_NAME.key(), ClusteringPlanPartitionFilterMode.SELECTED_PARTITIONS.name()));
return flinkSizeBasedClassName;
} else if (javaSelectedPartitionClassName.equals(className)) {
config.setValue(HoodieClusteringConfig.PLAN_PARTITION_FILTER_MODE_NAME, ClusteringPlanPartitionFilterMode.RECENT_DAYS.name());
LOG.warn(String.format(logStr, className, javaSizeBasedClassName, HoodieClusteringConfig.PLAN_PARTITION_FILTER_MODE_NAME.key(), ClusteringPlanPartitionFilterMode.SELECTED_PARTITIONS.name()));
@@ -173,7 +162,7 @@ public abstract class ClusteringPlanStrategy<T extends HoodieRecordPayload,I,K,O
return metrics;
}
protected HoodieTable<T,I,K, O> getHoodieTable() {
protected HoodieTable<T, I, K, O> getHoodieTable() {
return this.hoodieTable;
}

View File

@@ -71,7 +71,7 @@ public class ScheduleCompactionActionExecutor<T extends HoodieRecordPayload, I,
if (!config.getWriteConcurrencyMode().supportsOptimisticConcurrencyControl()
&& !config.getFailedWritesCleanPolicy().isLazy()) {
// TODO(yihua): this validation is removed for Java client used by kafka-connect. Need to revisit this.
if (config.getEngineType() != EngineType.JAVA) {
if (config.getEngineType() == EngineType.SPARK) {
// if there are inflight writes, their instantTime must not be less than that of compaction instant time
table.getActiveTimeline().getCommitsTimeline().filterPendingExcludingCompaction().firstInstant()
.ifPresent(earliestInflight -> ValidationUtils.checkArgument(

View File

@@ -1,65 +0,0 @@
/*
* Licensed to the Apache Software Foundation (ASF) under one
* or more contributor license agreements. See the NOTICE file
* distributed with this work for additional information
* regarding copyright ownership. The ASF licenses this file
* to you under the Apache License, Version 2.0 (the
* "License"); you may not use this file except in compliance
* with the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package org.apache.hudi.client.clustering.plan.strategy;
import org.apache.hudi.client.common.HoodieFlinkEngineContext;
import org.apache.hudi.common.model.HoodieRecordPayload;
import org.apache.hudi.config.HoodieWriteConfig;
import org.apache.hudi.table.HoodieFlinkCopyOnWriteTable;
import org.apache.hudi.table.HoodieFlinkMergeOnReadTable;
import org.apache.log4j.LogManager;
import org.apache.log4j.Logger;
import java.util.Comparator;
import java.util.List;
import java.util.stream.Collectors;
/**
* Clustering Strategy based on following.
* 1) Only looks at latest 'daybased.lookback.partitions' partitions.
* 2) Excludes files that are greater than 'small.file.limit' from clustering plan.
*/
public class FlinkRecentDaysClusteringPlanStrategy<T extends HoodieRecordPayload<T>>
extends FlinkSizeBasedClusteringPlanStrategy<T> {
private static final Logger LOG = LogManager.getLogger(FlinkRecentDaysClusteringPlanStrategy.class);
public FlinkRecentDaysClusteringPlanStrategy(HoodieFlinkCopyOnWriteTable<T> table,
HoodieFlinkEngineContext engineContext,
HoodieWriteConfig writeConfig) {
super(table, engineContext, writeConfig);
}
public FlinkRecentDaysClusteringPlanStrategy(HoodieFlinkMergeOnReadTable<T> table,
HoodieFlinkEngineContext engineContext,
HoodieWriteConfig writeConfig) {
super(table, engineContext, writeConfig);
}
@Override
protected List<String> filterPartitionPaths(List<String> partitionPaths) {
int targetPartitionsForClustering = getWriteConfig().getTargetPartitionsForClustering();
int skipPartitionsFromLatestForClustering = getWriteConfig().getSkipPartitionsFromLatestForClustering();
return partitionPaths.stream()
.sorted(Comparator.reverseOrder())
.skip(Math.max(skipPartitionsFromLatestForClustering, 0))
.limit(targetPartitionsForClustering > 0 ? targetPartitionsForClustering : partitionPaths.size())
.collect(Collectors.toList());
}
}

View File

@@ -1,67 +0,0 @@
/*
* Licensed to the Apache Software Foundation (ASF) under one
* or more contributor license agreements. See the NOTICE file
* distributed with this work for additional information
* regarding copyright ownership. The ASF licenses this file
* to you under the Apache License, Version 2.0 (the
* "License"); you may not use this file except in compliance
* with the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package org.apache.hudi.client.clustering.plan.strategy;
import org.apache.hudi.client.common.HoodieFlinkEngineContext;
import org.apache.hudi.common.model.HoodieRecordPayload;
import org.apache.hudi.config.HoodieWriteConfig;
import org.apache.hudi.table.HoodieFlinkCopyOnWriteTable;
import org.apache.hudi.table.HoodieFlinkMergeOnReadTable;
import org.apache.log4j.LogManager;
import org.apache.log4j.Logger;
import java.util.List;
import java.util.stream.Collectors;
import static org.apache.hudi.config.HoodieClusteringConfig.CLUSTERING_STRATEGY_PARAM_PREFIX;
/**
* Clustering Strategy to filter just specified partitions from [begin, end]. Note both begin and end are inclusive.
*/
public class FlinkSelectedPartitionsClusteringPlanStrategy<T extends HoodieRecordPayload<T>>
extends FlinkSizeBasedClusteringPlanStrategy<T> {
private static final Logger LOG = LogManager.getLogger(FlinkSelectedPartitionsClusteringPlanStrategy.class);
public static final String CONF_BEGIN_PARTITION = CLUSTERING_STRATEGY_PARAM_PREFIX + "cluster.begin.partition";
public static final String CONF_END_PARTITION = CLUSTERING_STRATEGY_PARAM_PREFIX + "cluster.end.partition";
public FlinkSelectedPartitionsClusteringPlanStrategy(HoodieFlinkCopyOnWriteTable<T> table,
HoodieFlinkEngineContext engineContext,
HoodieWriteConfig writeConfig) {
super(table, engineContext, writeConfig);
}
public FlinkSelectedPartitionsClusteringPlanStrategy(HoodieFlinkMergeOnReadTable<T> table,
HoodieFlinkEngineContext engineContext,
HoodieWriteConfig writeConfig) {
super(table, engineContext, writeConfig);
}
@Override
protected List<String> filterPartitionPaths(List<String> partitionPaths) {
String beginPartition = getWriteConfig().getProps().getProperty(CONF_BEGIN_PARTITION);
String endPartition = getWriteConfig().getProps().getProperty(CONF_END_PARTITION);
List<String> filteredPartitions = partitionPaths.stream()
.filter(path -> path.compareTo(beginPartition) >= 0 && path.compareTo(endPartition) <= 0)
.collect(Collectors.toList());
LOG.info("Filtered to the following partitions: " + filteredPartitions);
return filteredPartitions;
}
}