priyeshkaratha commented on code in PR #9326: URL: https://github.com/apache/ozone/pull/9326#discussion_r2544971681
########## hadoop-ozone/recon/src/main/java/org/apache/hadoop/ozone/recon/api/StorageDistributionEndpoint.java: ########## @@ -0,0 +1,343 @@ +/* + * 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.ozone.recon.api; + +import static org.apache.hadoop.ozone.om.codec.OMDBDefinition.FILE_TABLE; +import static org.apache.hadoop.ozone.om.codec.OMDBDefinition.KEY_TABLE; + +import java.io.IOException; +import java.util.HashMap; +import java.util.List; +import java.util.Map; +import java.util.concurrent.ConcurrentHashMap; +import java.util.stream.Collectors; +import javax.inject.Inject; +import javax.ws.rs.GET; +import javax.ws.rs.Path; +import javax.ws.rs.Produces; +import javax.ws.rs.core.Response; +import org.apache.hadoop.hdds.protocol.DatanodeDetails; +import org.apache.hadoop.hdds.protocol.proto.HddsProtos.DeletedBlocksTransactionSummary; +import org.apache.hadoop.hdds.scm.container.placement.metrics.SCMNodeMetric; +import org.apache.hadoop.hdds.scm.container.placement.metrics.SCMNodeStat; +import org.apache.hadoop.hdds.scm.protocol.StorageContainerLocationProtocol; +import org.apache.hadoop.hdds.scm.server.OzoneStorageContainerManager; +import org.apache.hadoop.ozone.recon.ReconUtils; +import org.apache.hadoop.ozone.recon.api.types.DUResponse; +import org.apache.hadoop.ozone.recon.api.types.DatanodeStorageReport; +import org.apache.hadoop.ozone.recon.api.types.DeletionPendingBytesByComponent; +import org.apache.hadoop.ozone.recon.api.types.GlobalNamespaceReport; +import org.apache.hadoop.ozone.recon.api.types.GlobalStorageReport; +import org.apache.hadoop.ozone.recon.api.types.KeyInsightInfoResponse; +import org.apache.hadoop.ozone.recon.api.types.StorageCapacityDistributionResponse; +import org.apache.hadoop.ozone.recon.api.types.UsedSpaceBreakDown; +import org.apache.hadoop.ozone.recon.scm.ReconNodeManager; +import org.apache.hadoop.ozone.recon.tasks.OmTableInsightTask; +import org.apache.ozone.recon.schema.generated.tables.daos.GlobalStatsDao; +import org.apache.ozone.recon.schema.generated.tables.pojos.GlobalStats; +import org.slf4j.Logger; +import org.slf4j.LoggerFactory; + +/** + * This endpoint handles requests related to storage distribution across + * different datanodes in a Recon instance. It provides detailed reports + * on storage capacity, utilization, and associated metrics. + * <p> + * The data is aggregated from multiple sources, including node manager + * statistics, and is used to construct responses with information + * about global storage and namespace usage, storage usage breakdown, + * and deletion operations in progress. + * <p> + * An instance of {@link ReconNodeManager} is used to fetch detailed + * node-specific statistics required for generating the report. + */ +@Path("/storageDistribution") +@Produces("application/json") +@AdminOnly +public class StorageDistributionEndpoint { + private final ReconNodeManager nodeManager; + private final OMDBInsightEndpoint omdbInsightEndpoint; + private final NSSummaryEndpoint nsSummaryEndpoint; + private final StorageContainerLocationProtocol scmClient; + private static Logger log = LoggerFactory.getLogger(StorageDistributionEndpoint.class); + private Map<DatanodeDetails, Long> blockDeletionMetricsMap = new ConcurrentHashMap<>(); + private GlobalStatsDao globalStatsDao; + + @Inject + public StorageDistributionEndpoint(OzoneStorageContainerManager reconSCM, + OMDBInsightEndpoint omDbInsightEndpoint, + NSSummaryEndpoint nsSummaryEndpoint, + GlobalStatsDao globalStatsDao, + StorageContainerLocationProtocol scmClient) { + this.nodeManager = (ReconNodeManager) reconSCM.getScmNodeManager(); + this.omdbInsightEndpoint = omDbInsightEndpoint; + this.nsSummaryEndpoint = nsSummaryEndpoint; + this.scmClient = scmClient; + this.globalStatsDao = globalStatsDao; + } + + @GET + public Response getStorageDistribution() { + try { + initializeBlockDeletionMetricsMap(); + List<DatanodeStorageReport> nodeStorageReports = collectDatanodeReports(); + GlobalStorageReport globalStorageReport = calculateGlobalStorageReport(); + + Map<String, Long> namespaceMetrics = new HashMap<>(); + try { + namespaceMetrics = calculateNamespaceMetrics(); + } catch (Exception e) { + log.error("Error calculating namespace metrics", e); + // Initialize with default values + namespaceMetrics.put("totalUsedNamespace", 0L); + namespaceMetrics.put("totalOpenKeySize", 0L); + namespaceMetrics.put("totalCommittedSize", 0L); + namespaceMetrics.put("pendingDirectorySize", 0L); + namespaceMetrics.put("pendingKeySize", 0L); + namespaceMetrics.put("totalKeys", 0L); + } + + StorageCapacityDistributionResponse response = buildStorageDistributionResponse( + nodeStorageReports, globalStorageReport, namespaceMetrics); + return Response.ok(response).build(); + } catch (Exception e) { + log.error("Error getting storage distribution", e); + return Response.status(Response.Status.INTERNAL_SERVER_ERROR) + .entity("Error retrieving storage distribution: " + e.getMessage()) + .build(); + } + } + + private GlobalStorageReport calculateGlobalStorageReport() { + try { + SCMNodeStat stats = nodeManager.getStats(); + if (stats == null) { + log.warn("Node manager stats are null, returning default values"); + return new GlobalStorageReport(0L, 0L, 0L); + } + + long scmUsed = stats.getScmUsed() != null ? stats.getScmUsed().get() : 0L; + long remaining = stats.getRemaining() != null ? stats.getRemaining().get() : 0L; + long capacity = stats.getCapacity() != null ? stats.getCapacity().get() : 0L; + + return new GlobalStorageReport(scmUsed, remaining, capacity); + } catch (Exception e) { + log.error("Error calculating global storage report", e); + return new GlobalStorageReport(0L, 0L, 0L); + } + } + + private Map<String, Long> calculateNamespaceMetrics() { + Map<String, Long> metrics = new HashMap<>(); + Map<String, Long> totalPendingAtOmSide = calculatePendingSizes(); + long totalOpenKeySize = calculateOpenKeySizes(); + long totalCommittedSize = calculateCommittedSize(); + long pendingDirectorySize = totalPendingAtOmSide.getOrDefault("pendingDirectorySize", 0L); + long pendingKeySize = totalPendingAtOmSide.getOrDefault("pendingKeySize", 0L); + long totalUsedNamespace = pendingDirectorySize + pendingKeySize + totalOpenKeySize + totalCommittedSize; + + long totalKeys = 0L; + // Keys from OBJECT_STORE buckets. + GlobalStats keyRecord = globalStatsDao.findById( + OmTableInsightTask.getTableCountKeyFromTable(KEY_TABLE)); + // Keys from FILE_SYSTEM_OPTIMIZED buckets + GlobalStats fileRecord = globalStatsDao.findById( + OmTableInsightTask.getTableCountKeyFromTable(FILE_TABLE)); + if (keyRecord != null) { + totalKeys += keyRecord.getValue(); + } + if (fileRecord != null) { + totalKeys += fileRecord.getValue(); + } + + metrics.put("pendingDirectorySize", pendingDirectorySize); + metrics.put("pendingKeySize", pendingKeySize); + metrics.put("totalOpenKeySize", totalOpenKeySize); + metrics.put("totalCommittedSize", totalCommittedSize); + metrics.put("totalUsedNamespace", totalUsedNamespace); + metrics.put("totalKeys", totalKeys); + return metrics; + } + + private StorageCapacityDistributionResponse buildStorageDistributionResponse( + List<DatanodeStorageReport> nodeStorageReports, + GlobalStorageReport storageMetrics, + Map<String, Long> namespaceMetrics) { + DeletedBlocksTransactionSummary scmSummary = null; + try { + scmSummary = scmClient.getDeletedBlockSummary(); + } catch (IOException e) { + log.error("Failed to get deleted block summary from SCM", e); + } + + long totalPendingAtDnSide = 0L; + try { + totalPendingAtDnSide = blockDeletionMetricsMap.values().stream().reduce(0L, Long::sum); + } catch (Exception e) { + log.error("Error calculating pending deletion metrics", e); + } + + DeletionPendingBytesByComponent deletionPendingBytesByStage = + createDeletionPendingBytesByStage( + namespaceMetrics.getOrDefault("pendingDirectorySize", 0L), + namespaceMetrics.getOrDefault("pendingKeySize", 0L), + scmSummary != null ? scmSummary.getTotalBlockReplicatedSize() : 0L, + totalPendingAtDnSide); + + // Safely get values from namespaceMetrics with null checks + Long totalUsedNamespace = namespaceMetrics.get("totalUsedNamespace"); + Long totalOpenKeySize = namespaceMetrics.get("totalOpenKeySize"); + Long totalCommittedSize = namespaceMetrics.get("totalCommittedSize"); + Long totalKeys = namespaceMetrics.get("totalKeys"); + Long totalContainerPreAllocated = nodeStorageReports.stream() + .map(report -> report.getCommitted()) + .reduce(0L, Long::sum); + + return StorageCapacityDistributionResponse.newBuilder() + .setDataNodeUsage(nodeStorageReports) + .setGlobalStorage(storageMetrics) + .setGlobalNamespace(new GlobalNamespaceReport( + totalUsedNamespace != null ? totalUsedNamespace : 0L, + totalKeys != null ? totalKeys : 0L)) + .setUsedSpaceBreakDown(new UsedSpaceBreakDown( + totalOpenKeySize != null ? totalOpenKeySize : 0L, + totalCommittedSize != null ? totalCommittedSize : 0L, + totalContainerPreAllocated != null ? totalContainerPreAllocated : 0L, + deletionPendingBytesByStage)) + .build(); + } + + private List<DatanodeStorageReport> collectDatanodeReports() { + return nodeManager.getAllNodes().stream() + .map(this::getStorageReport) + .filter(report -> report != null) // Filter out null reports + .collect(Collectors.toList()); + } + + private Map<String, Long> calculatePendingSizes() { + Map<String, Long> result = new HashMap<>(); + KeyInsightInfoResponse response = (KeyInsightInfoResponse) + omdbInsightEndpoint.getDeletedDirInfo(-1, "").getEntity(); + + Map<String, Long> pendingKeySize = new HashMap<>(); + omdbInsightEndpoint.createKeysSummaryForDeletedKey(pendingKeySize); + result.put("pendingDirectorySize", response.getReplicatedDataSize()); + result.put("pendingKeySize", pendingKeySize.getOrDefault("totalReplicatedDataSize", 0L)); + return result; + } + + private long calculateOpenKeySizes() { + Map<String, Long> openKeySummary = new HashMap<>(); + omdbInsightEndpoint.createKeysSummaryForOpenKey(openKeySummary); + Map<String, Long> openKeyMPUSummary = new HashMap<>(); + omdbInsightEndpoint.createKeysSummaryForOpenMPUKey(openKeyMPUSummary); + long openKeyDataSize = openKeySummary.getOrDefault("totalReplicatedDataSize", 0L); + long totalMPUKeySize = openKeyMPUSummary.getOrDefault("totalReplicatedDataSize", 0L); + return openKeyDataSize + totalMPUKeySize; + } + + private long calculateCommittedSize() { + try { + Response rootResponse = nsSummaryEndpoint.getDiskUsage("/", false, true, false); + if (rootResponse.getStatus() != Response.Status.OK.getStatusCode()) { + log.warn("Failed to get disk usage, status: {}", rootResponse.getStatus()); + return 0L; + } + DUResponse duRootRes = (DUResponse) rootResponse.getEntity(); + return duRootRes != null ? duRootRes.getSizeWithReplica() : 0L; + } catch (IOException e) { + log.error("IOException while calculating committed size", e); + return 0L; + } + } + + private DeletionPendingBytesByComponent createDeletionPendingBytesByStage(long pendingDirectorySize, + long pendingKeySize, + long totalPendingAtScmSide, + long totalPendingAtDnSide) { + long totalPending = pendingDirectorySize + pendingKeySize + totalPendingAtScmSide + totalPendingAtDnSide; + Map<String, Map<String, Long>> stageItems = new HashMap<>(); + Map<String, Long> omMap = new HashMap<>(); + omMap.put("totalBytes", pendingDirectorySize + pendingKeySize); + omMap.put("pendingDirectoryBytes", pendingDirectorySize); + omMap.put("pendingKeyBytes", pendingKeySize); + Map<String, Long> scmMap = new HashMap<>(); + scmMap.put("pendingBytes", totalPendingAtScmSide); + Map<String, Long> dnMap = new HashMap<>(); + dnMap.put("pendingBytes", totalPendingAtDnSide); + stageItems.put("OM", omMap); + stageItems.put("SCM", scmMap); + stageItems.put("DN", dnMap); + return new DeletionPendingBytesByComponent(totalPending, stageItems); + } + + private void initializeBlockDeletionMetricsMap() { + nodeManager.getNodeStats().keySet().parallelStream().forEach(nodeId -> { Review Comment: Thanks @devmadhuu for the detailed suggestion. I have already a JIRA available to handle this performance part by caching data - HDDS-13845 -- 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]
