zhangshenghang commented on code in PR #10078: URL: https://github.com/apache/seatunnel/pull/10078#discussion_r2541006952
########## seatunnel-engine/seatunnel-engine-server/src/main/java/org/apache/seatunnel/engine/server/diagnostic/PendingDiagnosticsCollector.java: ########## @@ -0,0 +1,321 @@ +/* + * 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.seatunnel.engine.server.diagnostic; + +import org.apache.seatunnel.common.utils.ExceptionUtils; +import org.apache.seatunnel.engine.common.utils.concurrent.CompletableFuture; +import org.apache.seatunnel.engine.server.dag.physical.PhysicalPlan; +import org.apache.seatunnel.engine.server.dag.physical.PhysicalVertex; +import org.apache.seatunnel.engine.server.dag.physical.SubPlan; +import org.apache.seatunnel.engine.server.execution.PendingJobInfo; +import org.apache.seatunnel.engine.server.execution.TaskGroupLocation; +import org.apache.seatunnel.engine.server.master.JobMaster; +import org.apache.seatunnel.engine.server.resourcemanager.ResourceManager; +import org.apache.seatunnel.engine.server.resourcemanager.resource.SlotProfile; +import org.apache.seatunnel.engine.server.resourcemanager.resource.SystemLoadInfo; +import org.apache.seatunnel.engine.server.resourcemanager.worker.WorkerProfile; + +import com.hazelcast.cluster.Address; +import lombok.extern.slf4j.Slf4j; + +import java.util.ArrayList; +import java.util.Collections; +import java.util.HashMap; +import java.util.HashSet; +import java.util.List; +import java.util.Map; +import java.util.Optional; +import java.util.Set; +import java.util.concurrent.CompletionException; +import java.util.stream.Collectors; + +@Slf4j +public final class PendingDiagnosticsCollector { + + private static final String REASON_WAITING = "WAITING_SLOT_ASSIGNMENT"; + private static final String REASON_RESOURCE_NOT_ENOUGH = "RESOURCE_NOT_ENOUGH"; + private static final String REASON_REQUEST_FAILED = "REQUEST_FAILED"; + private static final String REASON_REQUEST_CANCELLED = "REQUEST_CANCELLED"; + + private PendingDiagnosticsCollector() {} + + public static PendingJobDiagnostic collectJobDiagnostic( + PendingJobInfo pendingJobInfo, + Map<String, String> tagFilter, + ResourceManager resourceManager) { + if (pendingJobInfo == null) { + return null; + } + JobMaster jobMaster = pendingJobInfo.getJobMaster(); + PendingJobDiagnostic diagnostic = new PendingJobDiagnostic(); + diagnostic.setJobId(jobMaster.getJobId()); + diagnostic.setJobName(jobMaster.getJobImmutableInformation().getJobName()); + diagnostic.setPendingSourceState(pendingJobInfo.getPendingSourceState()); + diagnostic.setJobStatus(jobMaster.getJobStatus()); + diagnostic.setEnqueueTimestamp(pendingJobInfo.getEnqueueTimestamp()); + diagnostic.setCheckTime(System.currentTimeMillis()); + diagnostic.setWaitDurationMs( + diagnostic.getCheckTime() - pendingJobInfo.getEnqueueTimestamp()); + diagnostic.setTagFilter( + tagFilter == null ? Collections.emptyMap() : new HashMap<>(tagFilter)); + Map<TaskGroupLocation, CompletableFuture<SlotProfile>> requestFutures = + Optional.ofNullable(jobMaster.getPhysicalPlan()) + .map(PhysicalPlan::getPreApplyResourceFutures) + .map(HashMap::new) + .orElseGet(HashMap::new); + + buildPipelineDiagnostics(jobMaster, requestFutures, diagnostic); + diagnostic.setTotalTaskGroups( + diagnostic.getPipelines().stream() + .mapToInt(PendingPipelineDiagnostic::getTotalTaskGroups) + .sum()); + diagnostic.setAllocatedTaskGroups( + diagnostic.getPipelines().stream() + .mapToInt(PendingPipelineDiagnostic::getAllocatedTaskGroups) + .sum()); + diagnostic.setLackingTaskGroups( + diagnostic.getPipelines().stream() + .mapToInt(PendingPipelineDiagnostic::getLackingTaskGroups) + .sum()); + + updateFailureReason(diagnostic); + diagnostic.setBlockingJobIds(collectBlockingJobs(resourceManager, jobMaster.getJobId())); + + return diagnostic; + } + + private static void buildPipelineDiagnostics( + JobMaster jobMaster, + Map<TaskGroupLocation, CompletableFuture<SlotProfile>> requestFutures, + PendingJobDiagnostic diagnostic) { + PhysicalPlan plan = jobMaster.getPhysicalPlan(); + if (plan == null) { + diagnostic.setFailureReason(REASON_WAITING); + diagnostic.setFailureMessage("Job master not initialized"); + return; + } + for (SubPlan subPlan : plan.getPipelineList()) { + PendingPipelineDiagnostic pipelineDiagnostic = new PendingPipelineDiagnostic(); + pipelineDiagnostic.setPipelineId(subPlan.getPipelineId()); + pipelineDiagnostic.setPipelineName(subPlan.getPipelineFullName()); + + List<PhysicalVertex> vertices = new ArrayList<>(); + vertices.addAll(subPlan.getCoordinatorVertexList()); + vertices.addAll(subPlan.getPhysicalVertexList()); + + int allocated = 0; + int lacking = 0; + for (PhysicalVertex vertex : vertices) { + TaskGroupLocation location = vertex.getTaskGroupLocation(); + PendingTaskGroupDiagnostic taskDiagnostic = + buildTaskDiagnostic( + location, vertex.getTaskFullName(), requestFutures.get(location)); + pipelineDiagnostic.getTaskGroupDiagnostics().add(taskDiagnostic); + if (taskDiagnostic.isAllocated()) { + allocated++; + } else { + lacking++; + diagnostic.getLackingTaskGroupDiagnostics().add(taskDiagnostic); + } + } + + pipelineDiagnostic.setTotalTaskGroups(vertices.size()); + pipelineDiagnostic.setAllocatedTaskGroups(allocated); + pipelineDiagnostic.setLackingTaskGroups(lacking); + diagnostic.getPipelines().add(pipelineDiagnostic); + } + } + + private static PendingTaskGroupDiagnostic buildTaskDiagnostic( + TaskGroupLocation location, + String taskFullName, + CompletableFuture<SlotProfile> future) { + PendingTaskGroupDiagnostic diagnostic = new PendingTaskGroupDiagnostic(); + diagnostic.setTaskGroupLocation(location); + diagnostic.setTaskFullName(taskFullName); + + if (future == null) { + diagnostic.setAllocated(false); + diagnostic.setFailureReason(REASON_RESOURCE_NOT_ENOUGH); + diagnostic.setFailureMessage("Slot request future not created"); + return diagnostic; + } + + if (future.isCancelled()) { + diagnostic.setAllocated(false); + diagnostic.setFailureReason(REASON_REQUEST_CANCELLED); + diagnostic.setFailureMessage("Slot request cancelled by resource manager"); + return diagnostic; + } + + if (!future.isDone()) { + diagnostic.setAllocated(false); + diagnostic.setFailureReason(REASON_WAITING); + diagnostic.setFailureMessage("Slot request still pending"); + return diagnostic; + } + try { + SlotProfile slotProfile = future.join(); + if (slotProfile != null) { + diagnostic.setAllocated(true); + return diagnostic; + } + diagnostic.setAllocated(false); + diagnostic.setFailureReason(REASON_RESOURCE_NOT_ENOUGH); + diagnostic.setFailureMessage("No available slot profile"); + } catch (CompletionException e) { + diagnostic.setAllocated(false); + diagnostic.setFailureReason(REASON_REQUEST_FAILED); + diagnostic.setFailureMessage(ExceptionUtils.getMessage(e)); + } + return diagnostic; + } + + private static void updateFailureReason(PendingJobDiagnostic diagnostic) { + if (diagnostic.getLackingTaskGroupDiagnostics().isEmpty()) { + if (diagnostic.getFailureReason() == null) { + diagnostic.setFailureReason(REASON_WAITING); + diagnostic.setFailureMessage("Job is waiting for scheduler to retry"); + } + return; + } + + Map<String, Long> reasonCounter = + diagnostic.getLackingTaskGroupDiagnostics().stream() + .collect( + Collectors.groupingBy( + PendingTaskGroupDiagnostic::getFailureReason, + Collectors.counting())); + String dominantReason = + reasonCounter.entrySet().stream() + .max(Map.Entry.comparingByValue()) + .map(Map.Entry::getKey) + .orElse(REASON_RESOURCE_NOT_ENOUGH); + diagnostic.setFailureReason(dominantReason); + diagnostic.setFailureMessage( + diagnostic.getLackingTaskGroupDiagnostics().stream() + .filter(diag -> dominantReason.equals(diag.getFailureReason())) + .map(PendingTaskGroupDiagnostic::getFailureMessage) + .filter(message -> message != null && !message.isEmpty()) + .distinct() + .collect(Collectors.joining("; "))); + } + + private static List<Long> collectBlockingJobs(ResourceManager resourceManager, long jobId) { + if (resourceManager == null) { + return Collections.emptyList(); + } + List<SlotProfile> assignedSlots = Collections.emptyList(); + try { + assignedSlots = resourceManager.getAssignedSlots(Collections.emptyMap()); + } catch (Exception e) { + log.warn("Collect assigned slots failed: {}", ExceptionUtils.getMessage(e)); + } + Set<Long> blocking = new HashSet<>(); + for (SlotProfile slotProfile : assignedSlots) { + long ownerId = slotProfile.getOwnerJobID(); + if (ownerId > 0 && ownerId != jobId) { + blocking.add(ownerId); + } + } + return new ArrayList<>(blocking); + } + + public static PendingClusterSnapshot collectClusterSnapshot( + ResourceManager resourceManager, Map<String, String> tagFilter) { + PendingClusterSnapshot snapshot = new PendingClusterSnapshot(); + if (resourceManager == null) { + return snapshot; + } + Map<String, String> tags = + tagFilter == null ? Collections.emptyMap() : new HashMap<>(tagFilter); + List<SlotProfile> assignedSlots = Collections.emptyList(); + List<SlotProfile> unassignedSlots = Collections.emptyList(); + try { + assignedSlots = resourceManager.getAssignedSlots(tags); + unassignedSlots = resourceManager.getUnassignedSlots(tags); + } catch (Exception e) { + log.warn("Collect slots info failed: {}", ExceptionUtils.getMessage(e)); + } + snapshot.setAssignedSlots(assignedSlots.size()); + snapshot.setFreeSlots(unassignedSlots.size()); + snapshot.setTotalSlots(assignedSlots.size() + unassignedSlots.size()); + try { + snapshot.setWorkerCount(resourceManager.workerCount(tags)); + } catch (Exception e) { + log.warn("Collect worker count failed: {}", ExceptionUtils.getMessage(e)); + } + snapshot.setWorkers(buildWorkerSnapshots(resourceManager, tags)); + return snapshot; + } + + private static List<WorkerResourceDiagnostic> buildWorkerSnapshots( + ResourceManager resourceManager, Map<String, String> tagFilter) { + if (resourceManager == null) { + return Collections.emptyList(); + } + Map<Address, WorkerProfile> registerWorker = + Optional.ofNullable(resourceManager.getRegisterWorker()) + .map(HashMap::new) + .orElseGet(HashMap::new); + return registerWorker.values().stream() + .map(worker -> convertWorker(worker, tagFilter)) + .collect(Collectors.toList()); + } + + private static WorkerResourceDiagnostic convertWorker( + WorkerProfile workerProfile, Map<String, String> tagFilter) { Review Comment: It might need to be deleted here -- 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]
