dmvk commented on code in PR #21981: URL: https://github.com/apache/flink/pull/21981#discussion_r1112810683
########## flink-runtime/src/main/java/org/apache/flink/runtime/scheduler/adaptive/allocator/StateLocalitySlotAssigner.java: ########## @@ -0,0 +1,209 @@ +/* + * 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.flink.runtime.scheduler.adaptive.allocator; + +import org.apache.flink.annotation.Internal; +import org.apache.flink.runtime.clusterframework.types.AllocationID; +import org.apache.flink.runtime.executiongraph.ExecutionGraph; +import org.apache.flink.runtime.executiongraph.ExecutionJobVertex; +import org.apache.flink.runtime.executiongraph.ExecutionVertex; +import org.apache.flink.runtime.jobgraph.JobVertexID; +import org.apache.flink.runtime.jobmaster.SlotInfo; +import org.apache.flink.runtime.scheduler.adaptive.allocator.SlotSharingSlotAllocator.ExecutionSlotSharingGroup; +import org.apache.flink.runtime.scheduler.adaptive.allocator.SlotSharingSlotAllocator.ExecutionSlotSharingGroupAndSlot; +import org.apache.flink.runtime.scheduler.strategy.ExecutionVertexID; +import org.apache.flink.runtime.state.KeyGroupRange; +import org.apache.flink.runtime.state.KeyGroupRangeAssignment; +import org.apache.flink.util.Preconditions; + +import java.util.ArrayList; +import java.util.Collection; +import java.util.Comparator; +import java.util.HashMap; +import java.util.Iterator; +import java.util.List; +import java.util.Map; +import java.util.PriorityQueue; +import java.util.stream.StreamSupport; + +import static java.util.function.Function.identity; +import static java.util.stream.Collectors.toMap; + +/** A {@link SlotAssigner} that assigns slots based on the number of local key groups. */ +@Internal +public class StateLocalitySlotAssigner implements SlotAssigner { + + private static class AllocationScore implements Comparable<AllocationScore> { + + private final String group; + private final AllocationID allocationId; + + public AllocationScore(String group, AllocationID allocationId, int score) { + this.group = group; + this.allocationId = allocationId; + this.score = score; + } + + private final int score; + + public String getGroup() { + return group; + } + + public AllocationID getAllocationId() { + return allocationId; + } + + public int getScore() { + return score; + } + + @Override + public int compareTo(StateLocalitySlotAssigner.AllocationScore other) { + int result = Integer.compare(score, other.score); + if (result != 0) { + return result; + } + result = other.allocationId.compareTo(allocationId); + if (result != 0) { + return result; + } + return other.group.compareTo(group); + } + } + + private final Map<AllocationID, Map<JobVertexID, KeyGroupRange>> locality; + private final Map<JobVertexID, Integer> maxParallelism; + + public StateLocalitySlotAssigner(ExecutionGraph archivedExecutionGraph) { + this( + calculateLocalKeyGroups(archivedExecutionGraph), + StreamSupport.stream( + archivedExecutionGraph.getVerticesTopologically().spliterator(), + false) + .collect( + toMap( + ExecutionJobVertex::getJobVertexId, + ExecutionJobVertex::getMaxParallelism))); + } + + public StateLocalitySlotAssigner( + Map<AllocationID, Map<JobVertexID, KeyGroupRange>> locality, + Map<JobVertexID, Integer> maxParallelism) { + this.locality = locality; + this.maxParallelism = maxParallelism; + } + + @Override + public AssignmentResult assignSlots( + Collection<? extends SlotInfo> slots, Collection<ExecutionSlotSharingGroup> groups) { + + final Map<JobVertexID, Integer> parallelism = new HashMap<>(); + groups.forEach( + group -> + group.getContainedExecutionVertices() + .forEach( + evi -> + parallelism.merge( + evi.getJobVertexId(), 1, Integer::sum))); + + PriorityQueue<AllocationScore> scores = new PriorityQueue<>(Comparator.reverseOrder()); + for (ExecutionSlotSharingGroup group : groups) { + calculateScore(group, parallelism) + .forEach( + (allocationId, score) -> + scores.add( + new AllocationScore( + group.getId(), allocationId, score))); + } + Map<String, ExecutionSlotSharingGroup> groupsById = + groups.stream().collect(toMap(ExecutionSlotSharingGroup::getId, identity())); + Map<AllocationID, SlotInfo> slotsById = + slots.stream().collect(toMap(SlotInfo::getAllocationId, identity())); + List<ExecutionSlotSharingGroupAndSlot> result = new ArrayList<>(); + AllocationScore score; + while ((score = scores.poll()) != null) { + SlotInfo slot = slotsById.remove(score.getAllocationId()); + if (slot != null) { + ExecutionSlotSharingGroup group = groupsById.remove(score.getGroup()); + if (group != null) { + result.add(new ExecutionSlotSharingGroupAndSlot(group, slot)); + } + } + } + + // Distribute the remaining slots with no score + Iterator<? extends SlotInfo> remainingSlots = slotsById.values().iterator(); + for (ExecutionSlotSharingGroup group : groupsById.values()) { + result.add(new ExecutionSlotSharingGroupAndSlot(group, remainingSlots.next())); + remainingSlots.remove(); + } + + return AssignmentResult.of(result, remainingSlots); + } + + public Map<AllocationID, Integer> calculateScore( + ExecutionSlotSharingGroup group, Map<JobVertexID, Integer> parallelism) { + final Map<AllocationID, Integer> score = new HashMap<>(); + for (ExecutionVertexID evi : group.getContainedExecutionVertices()) { + if (maxParallelism.containsKey(evi.getJobVertexId())) { + final KeyGroupRange kgr = + KeyGroupRangeAssignment.computeKeyGroupRangeForOperatorIndex( + maxParallelism.get(evi.getJobVertexId()), + parallelism.get(evi.getJobVertexId()), + evi.getSubtaskIndex()); + locality.forEach( + (allocationId, potentials) -> { + KeyGroupRange prev = potentials.get(evi.getJobVertexId()); + if (prev != null) { + int intersection = prev.getIntersection(kgr).getNumberOfKeyGroups(); + if (intersection > 0) { + score.merge(allocationId, intersection, Integer::sum); + } + } + }); + } + } + return score; + } + + private static Map<AllocationID, Map<JobVertexID, KeyGroupRange>> calculateLocalKeyGroups( Review Comment: I'm wondering whether we could somehow filter out stateless operators and ideally weight the operators by state size 🤔 -- 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: issues-unsubscr...@flink.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org
