[
https://issues.apache.org/jira/browse/YARN-11736?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17890854#comment-17890854
]
ASF GitHub Bot commented on YARN-11736:
---------------------------------------
TaoYang526 commented on code in PR #7121:
URL: https://github.com/apache/hadoop/pull/7121#discussion_r1806074629
##########
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/scheduler/placement/policy/MultiComparatorPolicy.java:
##########
@@ -0,0 +1,346 @@
+/**
+ * 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.yarn.server.resourcemanager.scheduler.placement.policy;
+
+import org.apache.hadoop.classification.VisibleForTesting;
+import org.apache.hadoop.conf.Configurable;
+import org.apache.hadoop.conf.Configuration;
+import org.apache.hadoop.yarn.exceptions.ConfigurationException;
+import org.apache.hadoop.yarn.server.resourcemanager.scheduler.SchedulerNode;
+import
org.apache.hadoop.yarn.server.resourcemanager.scheduler.capacity.CapacitySchedulerConfiguration;
+import
org.apache.hadoop.yarn.server.resourcemanager.scheduler.placement.MultiNodeLookupPolicy;
+import org.apache.hadoop.yarn.util.resource.DominantResourceCalculator;
+import org.apache.hadoop.yarn.util.resource.Resources;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+import java.util.ArrayList;
+import java.util.Arrays;
+import java.util.Collection;
+import java.util.Collections;
+import java.util.HashMap;
+import java.util.Iterator;
+import java.util.LinkedHashSet;
+import java.util.List;
+import java.util.Map;
+import java.util.Set;
+import java.util.concurrent.ConcurrentHashMap;
+import java.util.function.Function;
+import java.util.stream.Collectors;
+
+import static
org.apache.hadoop.yarn.server.resourcemanager.scheduler.capacity.CapacitySchedulerConfiguration.DOT;
+
+/**
+ * <p>
+ * This class has the following functionality:
+ *
+ * <p>
+ * MultiComparatorPolicy
+ * - manages some common comparators to help sorting nodes by
+ * allocated/unallocated/total resource, dominant ratio, etc.
+ * - holds sorted nodes list based on the of nodes at given time.
+ * - can be configured with specified comparators.
+ * </p>
+ */
+public class MultiComparatorPolicy<N extends SchedulerNode>
+ implements MultiNodeLookupPolicy<N>, Configurable {
+
+ private static final Logger LOG =
+ LoggerFactory.getLogger(MultiComparatorPolicy.class);
+ // comparators
+ private static final DominantResourceCalculator DOMINANT_RC =
+ new DominantResourceCalculator();
+ private static final Map<ComparatorKey, Function<SchedulerNode, Comparable>>
+ COMPARATOR_CALCULATORS = Collections.unmodifiableMap(
+ new HashMap<ComparatorKey, Function<SchedulerNode, Comparable>>() {{
+ // for vcores
+ put(ComparatorKey.ALLOCATED_VCORES,
+ obj -> obj.getAllocatedResource().getVirtualCores());
+ put(ComparatorKey.UNALLOCATED_VCORES,
+ obj -> obj.getUnallocatedResource().getVirtualCores());
+ put(ComparatorKey.TOTAL_VCORES,
+ obj -> obj.getTotalResource().getVirtualCores());
+ // for memory
+ put(ComparatorKey.ALLOCATED_MEMORY,
+ obj -> obj.getAllocatedResource().getMemorySize());
+ put(ComparatorKey.UNALLOCATED_MEMORY,
+ obj -> obj.getUnallocatedResource().getMemorySize());
+ put(ComparatorKey.TOTAL_MEMORY,
+ obj -> obj.getTotalResource().getMemorySize());
+ // for resource
+ put(ComparatorKey.ALLOCATED_RESOURCE,
+ SchedulerNode::getAllocatedResource);
+ put(ComparatorKey.UNALLOCATED_RESOURCE,
+ SchedulerNode::getUnallocatedResource);
+ put(ComparatorKey.TOTAL_RESOURCE,
+ SchedulerNode::getTotalResource);
+ // for dominant ratio
+ put(ComparatorKey.DOMINANT_ALLOCATED_RATIO, obj -> Resources
+ .ratio(DOMINANT_RC, obj.getAllocatedResource(),
+ obj.getTotalResource()));
+ // for node ID
+ put(ComparatorKey.NODE_ID, SchedulerNode::getNodeID);
+ }});
+ // conf keys and default values
+ public static final String COMPARATORS_CONF_KEY = "comparators";
+ protected static final List<Comparator> DEFAULT_COMPARATORS = Collections
+ .unmodifiableList(Arrays.asList(
+ new Comparator(ComparatorKey.DOMINANT_ALLOCATED_RATIO,
+ OrderDirection.ASC, COMPARATOR_CALCULATORS
+ .get(ComparatorKey.DOMINANT_ALLOCATED_RATIO)),
+ new Comparator(ComparatorKey.NODE_ID, OrderDirection.ASC,
+ COMPARATOR_CALCULATORS.get(ComparatorKey.NODE_ID))));
+
+ protected Map<String, Set<N>> nodesPerPartition = new ConcurrentHashMap<>();
+ protected List<Comparator> comparators;
+ private Configuration conf;
+
+ public MultiComparatorPolicy() {
+ }
+
+ @Override
+ public void setConf(Configuration conf) {
+ // init comparators
+ this.comparators = DEFAULT_COMPARATORS;
+ if (conf == null) {
+ return;
+ }
+ this.conf = conf;
+ String policyName = conf.get(
+ CapacitySchedulerConfiguration.MULTI_NODE_SORTING_POLICY_CURRENT_NAME);
Review Comment:
Currently`MULTI_NODE_SORTING_POLICY_NAME` is the prefix of all the
configurations of multi-node policy:
`yarn.scheduler.capacity.multi-node-sorting.policy`, it's not a proper name, I
would like rename it as MULTI_NODE_SORTING_POLICY in next update.
`MULTI_NODE_SORTING_POLICY_CURRENT_NAME` is used to transfer the policyName to
policy instance.
> Enhance MultiNodeLookupPolicy to allow configuration of extended comparators
> for better usability.
> --------------------------------------------------------------------------------------------------
>
> Key: YARN-11736
> URL: https://issues.apache.org/jira/browse/YARN-11736
> Project: Hadoop YARN
> Issue Type: Improvement
> Components: multi-node-placement
> Reporter: Tao Yang
> Assignee: Tao Yang
> Priority: Major
> Labels: pull-request-available
>
> Currently when multi-nodes is enabled, there is only 1 implementation of
> _MultiNodeLookupPolicy_ interface: {_}ResourceUsageMultiNodeLookupPolicy{_},
> which will sort nodes by allocated resources in ascending order.
> If cluster has nodes with different resource-spec, the resource utilization
> of smaller nodes will be significantly high, while larger nodes will
> experience low resource utilization. This may rise the hotspot risk and
> reduce the scheduling effectiveness.
> So I propose to add a new policy called _MultiComparatorPolicy_ to meet
> requirements from some complex scenarios, which should contains serveral
> inherit comparators and can be extended later, and supports configuring
> specified comparators for different policy instances.
>
> {*}Implementation Details{*}:
> 1. _MultiNodeSorter#initPolicy_ will pass the policyConf which is cloned from
> scheduler configuration and attached the name of current policy, so that we
> can fetch the specified configuration for this policy inside the
> implementations of MultiNodeLookupPolicy.
> 2. new implementation of {_}MultiNodeLookupPolicy{_}: _MultiComparatorPolicy_
> 2.1) contains several inherit comparators and can be extendable later.
> comparator keys: _ALLOCATED_RESOURCE / UNALLOCATED_RESOURCE /
> DOMINANT_RESOURCE_RATIO / NODE_ID,_ order-directions: _ASC / DESC._
> 2.2) supports configuring specified comparators with
> order-direction(ASC/DESC) for different policy instances via conf-key:
> {+}{color:#172b4d}yarn.scheduler.capacity.multi-node-sorting.policy.<policy-name>.comparators{color}{+},
> value format is
> "<comparator_key_1>[:<order_direction_1>],<comparator_key_2>[:<order_direction_2>],...".
> For example, "DOMINANT_ALLOCATED_RATIO,NODE_ID:DESC" means that for policy
> test, nodes should be sorted by dominant-resource-ratio in ascending order,
> by nodeID desc in descending order.
> 3. Refactor variable names in
> AbstractCSQueue/CSQueue/FiCaSchedulerApp/AppPlacementAllocator after
> supporting multiple policy instances with the same policy class.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
