sadanand48 commented on code in PR #8477: URL: https://github.com/apache/ozone/pull/8477#discussion_r2133866386
########## hadoop-ozone/ozone-manager/src/main/java/org/apache/hadoop/ozone/om/OMDBCheckPointServletInodeBasedXfer.java: ########## @@ -0,0 +1,207 @@ +/* + * 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.om; + +import static org.apache.hadoop.hdds.utils.Archiver.includeFile; +import static org.apache.hadoop.hdds.utils.Archiver.tar; +import static org.apache.hadoop.ozone.om.OMConfigKeys.OZONE_OM_RATIS_SNAPSHOT_MAX_TOTAL_SST_SIZE_DEFAULT; +import static org.apache.hadoop.ozone.om.OMConfigKeys.OZONE_OM_RATIS_SNAPSHOT_MAX_TOTAL_SST_SIZE_KEY; + +import java.io.File; +import java.io.IOException; +import java.io.OutputStream; +import java.nio.file.Files; +import java.nio.file.Path; +import java.nio.file.Paths; +import java.util.HashMap; +import java.util.HashSet; +import java.util.Map; +import java.util.Set; +import java.util.concurrent.atomic.AtomicLong; +import java.util.stream.Collectors; +import java.util.stream.Stream; +import javax.servlet.http.HttpServletRequest; +import org.apache.commons.compress.archivers.ArchiveOutputStream; +import org.apache.commons.compress.archivers.tar.TarArchiveEntry; +import org.apache.hadoop.hdds.conf.OzoneConfiguration; +import org.apache.hadoop.hdds.utils.db.DBCheckpoint; +import org.apache.hadoop.hdds.utils.db.RDBStore; +import org.apache.hadoop.ozone.OzoneConsts; +import org.apache.hadoop.ozone.om.helpers.SnapshotInfo; +import org.apache.hadoop.ozone.om.snapshot.OmSnapshotUtils; +import org.apache.hadoop.util.Time; +import org.apache.ozone.rocksdiff.RocksDBCheckpointDiffer; + +/** + * Specialized OMDBCheckpointServlet implementation that transfers Ozone Manager + * database checkpoints using inode-based deduplication. + * <p> + * This servlet constructs checkpoint archives by examining file inodes, + * ensuring that files with the same inode (i.e., hardlinks or duplicates) + * are only transferred once. It maintains mappings from inode IDs to file + * paths, manages hardlink information, and enforces snapshot and SST file + * size constraints as needed. + * <p> + * This approach optimizes checkpoint streaming by reducing redundant data + * transfer, especially in environments where RocksDB and snapshotting result + * in multiple hardlinks to the same physical data. + */ +public class OMDBCheckPointServletInodeBasedXfer extends OMDBCheckpointServlet { + + @Override + public void writeDbDataToStream(DBCheckpoint checkpoint, HttpServletRequest request, OutputStream destination, + Set<String> sstFilesToExclude, Path tmpdir) throws IOException, InterruptedException { + + // Key is the InodeID and path is the first encountered file path with this inodeID + // This will be later used to while writing to the tar. + Map<String, Path> copyFiles = new HashMap<>(); + Map<String, Set<Path>> hardlinkFiles = new HashMap<>(); + + OzoneManager om = (OzoneManager) getServletContext().getAttribute(OzoneConsts.OM_CONTEXT_ATTRIBUTE); + OMMetadataManager omMetadataManager = om.getMetadataManager(); + RDBStore rdbStore = (RDBStore) omMetadataManager.getStore(); + boolean includeSnapshotData = includeSnapshotData(request); + + long maxTotalSstSize = getConf().getLong(OZONE_OM_RATIS_SNAPSHOT_MAX_TOTAL_SST_SIZE_KEY, + OZONE_OM_RATIS_SNAPSHOT_MAX_TOTAL_SST_SIZE_DEFAULT); + // Tarball limits are not implemented for processes that don't + // include snapshots. Currently, this is just for recon. + if (!includeSnapshotData) { + maxTotalSstSize = Long.MAX_VALUE; + } + + try (ArchiveOutputStream<TarArchiveEntry> archiveOutputStream = tar(destination)) { + Set<Path> snapshotDbPaths = new HashSet<>(); + if (includeSnapshotData) { + RocksDBCheckpointDiffer differ = getDbStore().getRocksDBCheckpointDiffer(); + Path sstBackupDir = new File(differ.getSSTBackupDir()).toPath(); + Path compactionLogDir = new File(differ.getCompactionLogDir()).toPath(); + SnapshotChainManager snapshotChainManager = new SnapshotChainManager(omMetadataManager); + for (SnapshotChainInfo snapInfo : snapshotChainManager.getGlobalSnapshotChain().values()) { + String snapshotDir = OmSnapshotManager.getSnapshotPath(getConf(), + SnapshotInfo.getCheckpointDirName(snapInfo.getSnapshotId())); + Path path = Paths.get(snapshotDir); + snapshotDbPaths.add(path); + } + snapshotDbPaths.add(sstBackupDir); + snapshotDbPaths.add(compactionLogDir); + } + + boolean shouldContinue = true; + for (Path snapshotDbPath : snapshotDbPaths) { + if (!shouldContinue) { + break; + } + shouldContinue = getFilesForArchive(copyFiles, hardlinkFiles, + sstFilesToExclude, snapshotDbPath, maxTotalSstSize); + } + + if (shouldContinue) { + shouldContinue = getFilesForArchive(copyFiles, hardlinkFiles, + sstFilesToExclude, rdbStore.getDbLocation().toPath(), maxTotalSstSize); + } + + writeFilesToArchive(copyFiles, hardlinkFiles, archiveOutputStream, + shouldContinue, checkpoint.getCheckpointLocation()); + + } catch (Exception e) { + LOG.error("got exception writing to archive " + e); + throw e; + } + } + + private OzoneConfiguration getConf() { + return ((OzoneManager) getServletContext() + .getAttribute(OzoneConsts.OM_CONTEXT_ATTRIBUTE)) + .getConfiguration(); + } + + private void writeFilesToArchive(Map<String, Path> copyFiles, Map<String, Set<Path>> hardlinkFiles, + ArchiveOutputStream<TarArchiveEntry> archiveOutputStream, boolean completed, Path checkpointLocation) + throws IOException { + Map<String, Path> filteredCopyFiles = completed ? copyFiles : + copyFiles.entrySet().stream().filter(e -> e.getValue().getFileName().toString().toLowerCase().endsWith(".sst")) + .collect(Collectors.toMap(Map.Entry::getKey, Map.Entry::getValue)); + + Path metaDirPath = getMetaDirPath(checkpointLocation); + long bytesWritten = 0L; + int filesWritten = 0; + long lastLoggedTime = Time.monotonicNow(); + + // Go through each of the files to be copied and add to archive. + for (Map.Entry<String, Path> entry : filteredCopyFiles.entrySet()) { + Path path = entry.getValue(); + // Confirm the data is in the right place. + if (!path.toString().startsWith(metaDirPath.toString())) { + throw new IOException("tarball file not in metadata dir: " + + path + ": " + metaDirPath); + } + + File file = path.toFile(); + if (!file.isDirectory()) { + filesWritten++; + } + LOG.info("Writing file = {} for inode {}", file.getAbsolutePath(), entry.getKey()); + bytesWritten += includeFile(file, entry.getKey(), archiveOutputStream); + // Log progress every 30 seconds + if (Time.monotonicNow() - lastLoggedTime >= 30000) { + LOG.info("Transferred {} KB, #files {} to checkpoint tarball stream...", + bytesWritten / (1024), filesWritten); + lastLoggedTime = Time.monotonicNow(); + } + } + + /* + if (completed) { + // TODO: + // 1. take an OM db checkpoint , + // 2. hold a snapshotCache lock, + // 3. copy remaining files Review Comment: the idea is to batch each transfer with size maxTotalSstSize. If size exceeds , the remaining files are sent on the next batch. At the last iteration , when we take the checkpoint , the batch size is set to unlimited to transfer all remaining stuff in a single iteration. This is to avoid multiple roundtrips once checkpoint is taken to reduce the gap b/w two OM's. -- 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...@ozone.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@ozone.apache.org For additional commands, e-mail: issues-h...@ozone.apache.org
