[SPARK-48586][SS] Remove lock acquisition in doMaintenance() by making a deep copy of file mappings in RocksDBFileManager in load() #46942
There are no files selected for viewing
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
|
|
@@ -133,16 +133,6 @@ class RocksDBFileManager( | |
|
|
||
| import RocksDBImmutableFile._ | ||
|
|
||
riyaverm-db marked this conversation as resolved.
Outdated
Show resolved
Hide resolved
|
||
| private lazy val fm = CheckpointFileManager.create(new Path(dfsRootDir), hadoopConf) | ||
| private val fs = new Path(dfsRootDir).getFileSystem(hadoopConf) | ||
| private val onlyZipFiles = new PathFilter { | ||
|
|
@@ -157,6 +147,29 @@ class RocksDBFileManager( | |
| private def codec = CompressionCodec.createCodec(sparkConf, codecName) | ||
|
|
||
| @volatile private var rootDirChecked: Boolean = false | ||
| @volatile private var fileMappings = RocksDBFileMappings( | ||
| new ConcurrentHashMap[Long, Seq[RocksDBImmutableFile]], | ||
| new ConcurrentHashMap[String, RocksDBImmutableFile] | ||
| ) | ||
|
|
||
| /** | ||
| * Make a deep copy of versionToRocksDBFiles and localFilesToDfsFiles to avoid | ||
| * current task thread from overwriting the file mapping whenever background maintenance | ||
| * thread attempts to upload a snapshot | ||
| */ | ||
| def copyFileMapping() : Unit = { | ||
| val newVersionToRocksDBFiles = new ConcurrentHashMap[Long, Seq[RocksDBImmutableFile]] | ||
| val newLocalFilesToDfsFiles = new ConcurrentHashMap[String, RocksDBImmutableFile] | ||
|
|
||
| newVersionToRocksDBFiles.putAll(fileMappings.versionToRocksDBFiles) | ||
| newLocalFilesToDfsFiles.putAll(fileMappings.localFilesToDfsFiles) | ||
|
|
||
| fileMappings = RocksDBFileMappings(newVersionToRocksDBFiles, newLocalFilesToDfsFiles) | ||
| } | ||
|
|
||
| def captureFileMapReference(): RocksDBFileMappings = { | ||
|
Contributor
There was a problem hiding this comment. I didn't quite get the purpose of this function.
Contributor
Author
There was a problem hiding this comment. It functions as a get method for the RocksDBFileMapping private var. I named it captureFileMapReference for code readability in RocksDB where it is used.
Contributor
There was a problem hiding this comment. I believe Scala has a accessor style guild: https://docs.scala-lang.org/style/naming-conventions.html#accessorsmutators |
||
| fileMappings | ||
| } | ||
|
|
||
| def getChangeLogWriter( | ||
| version: Long, | ||
|
|
@@ -204,11 +217,15 @@ class RocksDBFileManager( | |
| def latestSaveCheckpointMetrics: RocksDBFileManagerMetrics = saveCheckpointMetrics | ||
|
|
||
| /** Save all the files in given local checkpoint directory as a committed version in DFS */ | ||
| def saveCheckpointToDfs( | ||
| checkpointDir: File, | ||
|
Contributor
There was a problem hiding this comment. nit: same here, 2 more spaces |
||
| version: Long, | ||
| numKeys: Long, | ||
| capturedFileMappings: RocksDBFileMappings): Unit = { | ||
| logFilesInDir(checkpointDir, log"Saving checkpoint files " + | ||
| log"for version ${MDC(LogKeys.VERSION_NUM, version)}") | ||
| val (localImmutableFiles, localOtherFiles) = listRocksDBFiles(checkpointDir) | ||
| val rocksDBFiles = saveImmutableFilesToDfs(version, localImmutableFiles, capturedFileMappings) | ||
| val metadata = RocksDBCheckpointMetadata(rocksDBFiles, numKeys) | ||
| val metadataFile = localMetadataFile(checkpointDir) | ||
| metadata.writeToFile(metadataFile) | ||
|
|
@@ -243,10 +260,10 @@ class RocksDBFileManager( | |
| // The unique ids of SST files are checked when opening a rocksdb instance. The SST files | ||
| // in larger versions can't be reused even if they have the same size and name because | ||
| // they belong to another rocksdb instance. | ||
| fileMappings.versionToRocksDBFiles.keySet().removeIf(_ >= version) | ||
| val metadata = if (version == 0) { | ||
| if (localDir.exists) Utils.deleteRecursively(localDir) | ||
| fileMappings.localFilesToDfsFiles.clear() | ||
| localDir.mkdirs() | ||
| RocksDBCheckpointMetadata(Seq.empty, 0) | ||
| } else { | ||
|
|
@@ -260,7 +277,7 @@ class RocksDBFileManager( | |
| logInfo(log"Read metadata for version ${MDC(LogKeys.VERSION_NUM, version)}:\n" + | ||
| log"${MDC(LogKeys.METADATA_JSON, metadata.prettyJson)}") | ||
| loadImmutableFilesFromDfs(metadata.immutableFiles, localDir) | ||
| fileMappings.versionToRocksDBFiles.put(version, metadata.immutableFiles) | ||
| metadataFile.delete() | ||
| metadata | ||
| } | ||
|
|
@@ -417,9 +434,9 @@ class RocksDBFileManager( | |
| // Resolve RocksDB files for all the versions and find the max version each file is used | ||
| val fileToMaxUsedVersion = new mutable.HashMap[String, Long] | ||
| sortedSnapshotVersions.foreach { version => | ||
| val files = Option(fileMappings.versionToRocksDBFiles.get(version)).getOrElse { | ||
| val newResolvedFiles = getImmutableFilesFromVersionZip(version) | ||
| fileMappings.versionToRocksDBFiles.put(version, newResolvedFiles) | ||
| newResolvedFiles | ||
| } | ||
| files.foreach(f => fileToMaxUsedVersion(f.dfsFileName) = | ||
|
|
@@ -466,7 +483,7 @@ class RocksDBFileManager( | |
| val versionFile = dfsBatchZipFile(version) | ||
| try { | ||
| fm.delete(versionFile) | ||
| fileMappings.versionToRocksDBFiles.remove(version) | ||
| logDebug(s"Deleted version $version") | ||
| } catch { | ||
| case e: Exception => | ||
|
|
@@ -487,7 +504,8 @@ class RocksDBFileManager( | |
| /** Save immutable files to DFS directory */ | ||
| private def saveImmutableFilesToDfs( | ||
| version: Long, | ||
| localFiles: Seq[File], | ||
| capturedFileMappings: RocksDBFileMappings): Seq[RocksDBImmutableFile] = { | ||
| // Get the immutable files used in previous versions, as some of those uploaded files can be | ||
| // reused for this version | ||
| logInfo(log"Saving RocksDB files to DFS for ${MDC(LogKeys.VERSION_NUM, version)}") | ||
|
|
@@ -497,7 +515,8 @@ class RocksDBFileManager( | |
| var filesReused = 0L | ||
|
|
||
| val immutableFiles = localFiles.map { localFile => | ||
| val existingDfsFile = | ||
| capturedFileMappings.localFilesToDfsFiles.asScala.get(localFile.getName) | ||
| if (existingDfsFile.isDefined && existingDfsFile.get.sizeBytes == localFile.length()) { | ||
| val dfsFile = existingDfsFile.get | ||
| filesReused += 1 | ||
|
|
@@ -521,7 +540,7 @@ class RocksDBFileManager( | |
| bytesCopied += localFileSize | ||
|
|
||
| val immutableDfsFile = RocksDBImmutableFile(localFile.getName, dfsFileName, localFileSize) | ||
| capturedFileMappings.localFilesToDfsFiles.put(localFileName, immutableDfsFile) | ||
|
|
||
| immutableDfsFile | ||
| } | ||
|
|
@@ -530,7 +549,7 @@ class RocksDBFileManager( | |
| log"(${MDC(LogKeys.NUM_BYTES, bytesCopied)} bytes) from local to" + | ||
| log" DFS for version ${MDC(LogKeys.VERSION_NUM, version)}. " + | ||
| log"${MDC(LogKeys.NUM_FILES_REUSED, filesReused)} files reused without copying.") | ||
| capturedFileMappings.versionToRocksDBFiles.put(version, immutableFiles) | ||
|
|
||
| // Cleanup locally deleted files from the localFilesToDfsFiles map | ||
| // Locally, SST Files can be deleted due to RocksDB compaction. These files need | ||
|
|
@@ -570,7 +589,7 @@ class RocksDBFileManager( | |
| .foreach { existingFile => | ||
| val existingFileSize = existingFile.length() | ||
| val requiredFile = requiredFileNameToFileDetails.get(existingFile.getName) | ||
| val prevDfsFile = fileMappings.localFilesToDfsFiles.asScala.get(existingFile.getName) | ||
| val isSameFile = if (requiredFile.isDefined && prevDfsFile.isDefined) { | ||
| requiredFile.get.dfsFileName == prevDfsFile.get.dfsFileName && | ||
| existingFile.length() == requiredFile.get.sizeBytes | ||
|
|
@@ -580,7 +599,7 @@ class RocksDBFileManager( | |
|
|
||
| if (!isSameFile) { | ||
| existingFile.delete() | ||
| fileMappings.localFilesToDfsFiles.remove(existingFile.getName) | ||
| logInfo(log"Deleted local file ${MDC(LogKeys.FILE_NAME, existingFile)} " + | ||
| log"with size ${MDC(LogKeys.NUM_BYTES, existingFileSize)} mapped" + | ||
| log" to previous dfsFile ${MDC(LogKeys.DFS_FILE, prevDfsFile.getOrElse("null"))}") | ||
|
|
@@ -612,7 +631,7 @@ class RocksDBFileManager( | |
| } | ||
| filesCopied += 1 | ||
| bytesCopied += localFileSize | ||
| fileMappings.localFilesToDfsFiles.put(localFileName, file) | ||
| logInfo(log"Copied ${MDC(LogKeys.DFS_FILE, dfsFile)} to " + | ||
| log"${MDC(LogKeys.FILE_NAME, localFile)} - " + | ||
| log"${MDC(LogKeys.NUM_BYTES, localFileSize)} bytes") | ||
|
|
@@ -633,13 +652,13 @@ class RocksDBFileManager( | |
| private def removeLocallyDeletedSSTFilesFromDfsMapping(localFiles: Seq[File]): Unit = { | ||
| // clean up deleted SST files from the localFilesToDfsFiles Map | ||
| val currentLocalFiles = localFiles.map(_.getName).toSet | ||
| val mappingsToClean = fileMappings.localFilesToDfsFiles.asScala | ||
| .keys | ||
| .filterNot(currentLocalFiles.contains) | ||
|
|
||
| mappingsToClean.foreach { f => | ||
| logInfo(log"cleaning ${MDC(LogKeys.FILE_NAME, f)} from the localFilesToDfsFiles map") | ||
| fileMappings.localFilesToDfsFiles.remove(f) | ||
| } | ||
| } | ||
|
|
||
|
|
@@ -749,6 +768,20 @@ class RocksDBFileManager( | |
| } | ||
| } | ||
|
|
||
| /** | ||
| * Track file mappings in RocksDB across local and remote directories | ||
| * @param versionToRocksDBFiles Mapping of RocksDB files used across versions for maintenance | ||
| * @param localFilesToDfsFiles Mapping of the exact Dfs file used to create a local SST file | ||
| * The reason localFilesToDfsFiles is a separate map because versionToRocksDBFiles can contain | ||
| * multiple similar SST files to a particular local file (for example 1.sst can map to 1-UUID1.sst | ||
| * in v1 and 1-UUID2.sst in v2). We need to capture the exact file used to ensure Version ID | ||
| * compatibility across SST files and RocksDB manifest. | ||
| */ | ||
|
|
||
| case class RocksDBFileMappings( | ||
| versionToRocksDBFiles: ConcurrentHashMap[Long, Seq[RocksDBImmutableFile]], | ||
|
|
||
| localFilesToDfsFiles: ConcurrentHashMap[String, RocksDBImmutableFile]) | ||
|
|
||
| /** | ||
| * Metrics regarding RocksDB file sync between local and DFS. | ||
| */ | ||
|
|
||
Oops, something went wrong.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
nit: 2 more spaces (4 spaces for params in multi-lines)