[ https://issues.apache.org/jira/browse/HDDS-1577?focusedWorklogId=306080&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-306080 ]
ASF GitHub Bot logged work on HDDS-1577: ---------------------------------------- Author: ASF GitHub Bot Created on: 04/Sep/19 06:26 Start Date: 04/Sep/19 06:26 Worklog Time Spent: 10m Work Description: timmylicheng commented on pull request #1366: HDDS-1577. Add default pipeline placement policy implementation. URL: https://github.com/apache/hadoop/pull/1366#discussion_r320589153 ########## File path: hadoop-hdds/server-scm/src/main/java/org/apache/hadoop/hdds/scm/pipeline/PipelinePlacementPolicy.java ########## @@ -0,0 +1,291 @@ +/** + * 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.hadoop.hdds.scm.pipeline; + +import com.google.common.annotations.VisibleForTesting; +import org.apache.hadoop.conf.Configuration; +import org.apache.hadoop.hdds.protocol.DatanodeDetails; +import org.apache.hadoop.hdds.protocol.proto.HddsProtos; +import org.apache.hadoop.hdds.scm.ScmConfigKeys; +import org.apache.hadoop.hdds.scm.container.placement.algorithms.SCMCommonPolicy; +import org.apache.hadoop.hdds.scm.container.placement.metrics.SCMNodeMetric; +import org.apache.hadoop.hdds.scm.exceptions.SCMException; +import org.apache.hadoop.hdds.scm.net.NetworkTopology; +import org.apache.hadoop.hdds.scm.net.Node; +import org.apache.hadoop.hdds.scm.node.NodeManager; +import org.slf4j.Logger; +import org.slf4j.LoggerFactory; + +import java.util.ArrayList; +import java.util.Collection; +import java.util.List; +import java.util.stream.Collectors; + +/** + * Pipeline placement policy that choose datanodes based on load balancing + * and network topology to supply pipeline creation. + * <p> + * 1. get a list of healthy nodes + * 2. filter out nodes that are not too heavily engaged in other pipelines + * 3. Choose an anchor node among the viable nodes which follows the algorithm + * described @SCMContainerPlacementCapacity + * 4. Choose other nodes around the anchor node based on network topology + */ +public final class PipelinePlacementPolicy extends SCMCommonPolicy { + @VisibleForTesting + static final Logger LOG = + LoggerFactory.getLogger(PipelinePlacementPolicy.class); + private final NodeManager nodeManager; + private final Configuration conf; + private final int heavyNodeCriteria; + + /** + * Constructs a Container Placement with considering only capacity. + * That is this policy tries to place containers based on node weight. + * + * @param nodeManager Node Manager + * @param conf Configuration + */ + public PipelinePlacementPolicy(final NodeManager nodeManager, + final Configuration conf) { + super(nodeManager, conf); + this.nodeManager = nodeManager; + this.conf = conf; + heavyNodeCriteria = conf.getInt( + ScmConfigKeys.OZONE_DATANODE_MAX_PIPELINE_ENGAGEMENT, + ScmConfigKeys.OZONE_DATANODE_MAX_PIPELINE_ENGAGEMENT_DEFAULT); + } + + /** + * Returns true if this node meets the criteria. + * + * @param datanodeDetails DatanodeDetails + * @return true if we have enough space. + */ + @VisibleForTesting + boolean meetCriteria(DatanodeDetails datanodeDetails, + long heavyNodeLimit) { + return (nodeManager.getPipelinesCount(datanodeDetails) <= heavyNodeLimit); + } + + /** + * Filter out viable nodes based on + * 1. nodes that are healthy + * 2. nodes that are not too heavily engaged in other pipelines + * + * @param excludedNodes - excluded nodes + * @param nodesRequired - number of datanodes required. + * @return a list of viable nodes + * @throws SCMException when viable nodes are not enough in numbers + */ + List<DatanodeDetails> filterViableNodes( + List<DatanodeDetails> excludedNodes, int nodesRequired) + throws SCMException { + // get nodes in HEALTHY state + List<DatanodeDetails> healthyNodes = + nodeManager.getNodes(HddsProtos.NodeState.HEALTHY); + if (excludedNodes != null) { + healthyNodes.removeAll(excludedNodes); + } + String msg; + if (healthyNodes.size() == 0) { + msg = "No healthy node found to allocate container."; + LOG.error(msg); + throw new SCMException(msg, SCMException.ResultCodes + .FAILED_TO_FIND_HEALTHY_NODES); + } + + if (healthyNodes.size() < nodesRequired) { + msg = String.format("Not enough healthy nodes to allocate container. %d " + + " datanodes required. Found %d", + nodesRequired, healthyNodes.size()); + LOG.error(msg); + throw new SCMException(msg, + SCMException.ResultCodes.FAILED_TO_FIND_SUITABLE_NODE); + } + + // filter nodes that meet the size and pipeline engagement criteria. + // Pipeline placement doesn't take node space left into account. + List<DatanodeDetails> healthyList = healthyNodes.stream().filter(d -> + meetCriteria(d, heavyNodeCriteria)).collect(Collectors.toList()); + + if (healthyList.size() < nodesRequired) { + msg = String.format("Unable to find enough nodes that meet " + + "the criteria that cannot engage in more than %d pipelines." + + " Nodes required: %d Found: %d", + heavyNodeCriteria, nodesRequired, healthyList.size()); + LOG.error(msg); + throw new SCMException(msg, + SCMException.ResultCodes.FAILED_TO_FIND_NODES_WITH_SPACE); Review comment: Switched to FAILED_TO_FIND_SUITABLE_NODE ---------------------------------------------------------------- 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org Issue Time Tracking ------------------- Worklog Id: (was: 306080) Time Spent: 4h 40m (was: 4.5h) > Add default pipeline placement policy implementation > ---------------------------------------------------- > > Key: HDDS-1577 > URL: https://issues.apache.org/jira/browse/HDDS-1577 > Project: Hadoop Distributed Data Store > Issue Type: Sub-task > Components: SCM > Reporter: Siddharth Wagle > Assignee: Li Cheng > Priority: Major > Labels: pull-request-available > Time Spent: 4h 40m > Remaining Estimate: 0h > > This is a simpler implementation of the PipelinePlacementPolicy that can be > utilized if no network topology is defined for the cluster. We try to form > pipelines from existing HEALTHY datanodes randomly, as long as they satisfy > PipelinePlacementCriteria. -- This message was sent by Atlassian Jira (v8.3.2#803003) --------------------------------------------------------------------- To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org