com.hazelcast.spi.merge.SplitBrainMergePolicy Maven / Gradle / Ivy
/*
* Copyright (c) 2008-2021, 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.merge;
import com.hazelcast.core.HazelcastInstance;
import com.hazelcast.core.HazelcastInstanceAware;
import com.hazelcast.instance.impl.Node;
import com.hazelcast.internal.services.NodeAware;
import com.hazelcast.nio.serialization.DataSerializable;
/**
* Policy for merging data structure values
* after a split-brain has been healed.
*
* The values of merging and existing {@link MergingValue}s are
* always in the in-memory format of the backing data structure.
* This can be a serialized format, so the content cannot be
* processed without deserialization. For most merge policies
* this will be fine, since the key or value are not used.
*
* The deserialization is not done eagerly for two main reasons:
*
* - The deserialization is quite expensive and
* should be avoided, if the result is not needed.
* - There is no need to locate classes of stored entries
* on the server side, when the entries are not deserialized.
* So you can put entries from a client by using {@link
* com.hazelcast.config.InMemoryFormat#BINARY} with a different
* classpath on client and server. In this case a deserialization
* could throw a {@link java.lang.ClassNotFoundException}.
*
* If you need the deserialized data you can call
* {@link MergingValue#getValue()}
* or {@link MergingEntry#getKey()},
* which will deserialize the data lazily.
*
* A merge policy can implement {@link HazelcastInstanceAware} to get the
* {@link HazelcastInstance} injected. This can be used to retrieve the
* user context via {@link HazelcastInstance#getUserContext()}, which is
* an easy way to get user dependencies that are otherwise hard to obtain.
*
* A merge policy can also implement {@link NodeAware}
* to get an instance of {@link Node} injected.
*
* @param the (deserialized) type of the merging value
* @param the type of the required merging value, e.g. a simple
* {@code MergingValue} or a composition like {@code
* MergingEntry & MergingHits & MergingLastAccessTime}
* @param the type of the merged value as returned by {@link #merge(MergingValue, MergingValue)}
* @since 3.10
*/
public interface SplitBrainMergePolicy, R>
extends DataSerializable {
/**
* Selects the value of either the merging or the
* existing {@link MergingValue} which should be merged.
*
* Note that the existing {@link MergingValue} instance may be {@code null}
* if no matching data could be found to the merging {@link MergingValue}.
*
* @param mergingValue {@link MergingValue} instance that has the
* merging data of the smaller sub-cluster
* @param existingValue {@link MergingValue} instance that has the existing
* data or {@code null} if no matching data exists
* @return the selected value for merging
*/
R merge(T mergingValue, T existingValue);
}