All Downloads are FREE. Search and download functionalities are using the official Maven repository.

com.hazelcast.spi.impl.merge.AbstractContainerCollector Maven / Gradle / Ivy

There is a newer version: 4.5.4
Show newest version
/*
 * Copyright (c) 2008-2018, Hazelcast, Inc. All Rights Reserved.
 *
 * Licensed 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 com.hazelcast.spi.impl.merge;

import com.hazelcast.config.MergePolicyConfig;
import com.hazelcast.spi.ManagedService;
import com.hazelcast.spi.NodeEngine;
import com.hazelcast.spi.SplitBrainHandlerService;
import com.hazelcast.spi.impl.PartitionSpecificRunnable;
import com.hazelcast.spi.impl.operationexecutor.OperationExecutor;
import com.hazelcast.spi.impl.operationservice.impl.OperationServiceImpl;
import com.hazelcast.spi.merge.DiscardMergePolicy;
import com.hazelcast.spi.merge.SplitBrainMergePolicy;
import com.hazelcast.spi.merge.SplitBrainMergePolicyProvider;
import com.hazelcast.spi.partition.IPartitionService;

import java.util.Collection;
import java.util.Iterator;
import java.util.LinkedList;
import java.util.NoSuchElementException;
import java.util.concurrent.ConcurrentHashMap;
import java.util.concurrent.ConcurrentMap;
import java.util.concurrent.CountDownLatch;

/**
 * Collects non-threadsafe containers for {@link SplitBrainHandlerService}
 * capable data structures.
 * 

* The {@link com.hazelcast.internal.cluster.impl.ClusterMergeTask ClusterMergeTask} * collects all mergeable data from all Hazelcast services. The problem is that it * runs on an arbitrary thread, but the data is only modified by specific * {@link com.hazelcast.spi.impl.operationexecutor.impl.PartitionOperationThread partition threads}. * This can cause visibility issues due to a missing happens-before relation. *

* The collector fetches the data via {@link PartitionSpecificRunnable} (which * run on the partition threads) and stores them in a {@link ConcurrentHashMap}. * This guarantees the visibility for the {@code ClusterMergeTask}. *

* The collector can be implemented for data structures which reference their containers by *

    *
  • partition IDs using {@link AbstractContainerCollector}
  • *
  • container name using {@link AbstractNamedContainerCollector}
  • *
* Note: The collector will retain the collected containers for the merge runnable. * So {@link ManagedService#reset()} is not allowed to clear or destroy those containers. * To ensure this the collector removes the link between the data structure and a container * via {@link Iterator#remove()} in {@link CollectContainerRunnable#run()}. *

* The cleanup of the containers itself is done in {@link #destroy()} via {@link #destroy(Object)} * after the merge is done, or directly if the container is not collected. Containers from backup * partitions are directly cleaned via {@link #destroyBackup(Object)}. * * @param container of the data structure */ public abstract class AbstractContainerCollector { private final ConcurrentMap> containersByPartitionId = new ConcurrentHashMap>(); private final OperationExecutor operationExecutor; private final IPartitionService partitionService; private final SplitBrainMergePolicyProvider mergePolicyProvider; private CountDownLatch latch; protected AbstractContainerCollector(NodeEngine nodeEngine) { this.operationExecutor = ((OperationServiceImpl) nodeEngine.getOperationService()).getOperationExecutor(); this.partitionService = nodeEngine.getPartitionService(); this.mergePolicyProvider = nodeEngine.getSplitBrainMergePolicyProvider(); } /** * Collects the containers from the data structure in a thread-safe way. */ public final void run() { int partitionCount = partitionService.getPartitionCount(); latch = new CountDownLatch(partitionCount); for (int partitionId = 0; partitionId < partitionCount; partitionId++) { operationExecutor.execute(new CollectContainerRunnable(partitionId)); } try { latch.await(); } catch (InterruptedException e) { Thread.currentThread().interrupt(); } } /** * Returns the collected containers by partition ID. * * @return the collected containers */ public final ConcurrentMap> getCollectedContainers() { return containersByPartitionId; } /** * Destroys all collected containers. */ public final void destroy() { for (Collection containers : containersByPartitionId.values()) { for (C container : containers) { destroy(container); } } containersByPartitionId.clear(); onDestroy(); } /** * Will be called by {@link #destroy()}. *

* Can be overridden by implementations to cleanup local resources. */ protected void onDestroy() { // NOP } /** * Returns all containers of the data structure for the given partition ID. * * @return {@link Iterator} over the containers of the given partition */ protected abstract Iterator containerIterator(int partitionId); /** * Returns the {@link MergePolicyConfig} of the container. * * @return the {@link MergePolicyConfig} of the container */ protected abstract MergePolicyConfig getMergePolicyConfig(C container); /** * Destroys the owned data in the container. *

* Is called if a container is not collected or after the merge has been done. */ protected abstract void destroy(C container); /** * Destroys the backup data in the container. *

* Is called if the container is not collected, since it's in a backup partition. */ protected abstract void destroyBackup(C container); /** * Returns the number of collected merging values in this collector. *

* The count is used to calculate the timeout value to wait for merge operations to complete. *

* Note: Depending on the data structure, this can be the number of collected containers * or the number of merging values within the collected containers. * * @return the number of collected merge values */ protected abstract int getMergingValueCount(); /** * Determines if the container should be merged. *

* Can be overridden if there are additional restrictions beside the merge policy, * if a container is mergeable or not. * * @return {@code true} if the container is mergeable, {@code false} otherwise */ protected boolean isMergeable(C container) { return true; } /** * Empty iterator for {@link #containerIterator(int)} calls, if the requested partition is empty. */ protected final class EmptyIterator implements Iterator { public EmptyIterator() { } @Override public boolean hasNext() { return false; } @Override public C next() { throw new NoSuchElementException(); } @Override public void remove() { throw new UnsupportedOperationException(); } } /** * Collects containers from a specific partition ID. */ private final class CollectContainerRunnable implements PartitionSpecificRunnable { private final Collection containers = new LinkedList(); private final int partitionId; CollectContainerRunnable(int partitionId) { this.partitionId = partitionId; } @Override public int getPartitionId() { return partitionId; } @Override public void run() { try { Iterator iterator = containerIterator(partitionId); while (iterator.hasNext()) { C container = iterator.next(); collect(container); iterator.remove(); } } finally { if (!containers.isEmpty()) { containersByPartitionId.put(partitionId, containers); } latch.countDown(); } } private void collect(C container) { // just collect owned partitions if (partitionService.isPartitionOwner(partitionId)) { MergePolicyConfig mergePolicyconfig = getMergePolicyConfig(container); SplitBrainMergePolicy mergePolicy = mergePolicyProvider.getMergePolicy(mergePolicyconfig.getPolicy()); if (isMergeable(container) && !(mergePolicy instanceof DiscardMergePolicy)) { containers.add(container); } else { destroy(container); } } else { destroyBackup(container); } } } }





© 2015 - 2024 Weber Informatics LLC | Privacy Policy