xichen01 commented on code in PR #5354: URL: https://github.com/apache/ozone/pull/5354#discussion_r1407627084
########## hadoop-hdds/server-scm/src/main/java/org/apache/hadoop/hdds/scm/pipeline/choose/algorithms/CapacityPipelineChoosePolicy.java: ########## @@ -0,0 +1,123 @@ +/** + * 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 + * <p> + * http://www.apache.org/licenses/LICENSE-2.0 + * <p> + * 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.hadoop.hdds.scm.pipeline.choose.algorithms; + +import org.apache.hadoop.hdds.scm.PipelineChoosePolicy; +import org.apache.hadoop.hdds.scm.PipelineRequestInformation; +import org.apache.hadoop.hdds.scm.container.placement.metrics.SCMNodeMetric; +import org.apache.hadoop.hdds.scm.node.NodeManager; +import org.apache.hadoop.hdds.scm.pipeline.Pipeline; +import org.slf4j.Logger; +import org.slf4j.LoggerFactory; + +import java.util.ArrayDeque; +import java.util.ArrayList; +import java.util.Comparator; +import java.util.Deque; +import java.util.List; +import java.util.Objects; + +/** + * The capacity pipeline choose policy that chooses pipeline + * with relatively more remaining datanode space. + */ +public class CapacityPipelineChoosePolicy implements PipelineChoosePolicy { + + private static final Logger LOG = + LoggerFactory.getLogger(PipelineChoosePolicy.class); + + private final NodeManager nodeManager; + + private final PipelineChoosePolicy healthPolicy; + + public CapacityPipelineChoosePolicy(NodeManager nodeManager) { + this.nodeManager = nodeManager; + healthPolicy = new HealthyPipelineChoosePolicy(nodeManager); + } + + @Override + public Pipeline choosePipeline(List<Pipeline> pipelineList, + PipelineRequestInformation pri) { + Pipeline pipeline1 = healthPolicy.choosePipeline(pipelineList, pri); Review Comment: >@xichen01 Thanks for review ! About the logic of selection, there are links to this original papers in [HDFS-11564](https://issues.apache.org/jira/browse/HDFS-11564). The algorithms of choosing 2 random nodes and then placing the container on the lower utilization node is discussed in great depth in this survey paper. https://pdfs.semanticscholar.org/3597/66cb47572028eb70c797115e987ff203e83f.pdf In addition, SCMContainerPlacementCapacity#chooseNode also uses this algorithm. So, I wonder if it is not necessary to find the pipeline with minimum storage every time? @whbing Understood. For a fairly balanced cluster, such as a new one, this strategy can work very well, providing similar loads to all DataNodes. However, for a significantly unbalanced cluster, like when adding new nodes, this strategy might be limited, especially in larger clusters. But for the latter case (adding new nodes), we can also balance it using a balancer -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
