HADOOP-16202. Enhanced openFile(): mapreduce and YARN changes. (#2584/2)

These changes ensure that sequential files are opened with the right read policy, and split start/end is passed in. As well as offering opportunities for filesystem clients to choose fetch/cache/seek policies, the settings ensure that processing text files on an s3 bucket where the default policy is "random" will still be processed efficiently. This commit depends on the associated hadoop-common patch, which must be committed first. Contributed by Steve Loughran. Change-Id: Ic6713fd752441cf42ebe8739d05c2293a5db9f94
2022-04-24 17:10:34 +01:00 · 2022-04-24 17:10:34 +01:00 · 6999acf520
parent 1b4dba99b5
commit 6999acf520
10 changed files with 94 additions and 28 deletions
--- a/hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-app/src/main/java/org/apache/hadoop/mapreduce/jobhistory/JobHistoryCopyService.java
+++ b/hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-app/src/main/java/org/apache/hadoop/mapreduce/jobhistory/JobHistoryCopyService.java
@ -35,6 +35,10 @@ import org.apache.hadoop.yarn.exceptions.YarnRuntimeException;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;

+import static org.apache.hadoop.fs.Options.OpenFileOptions.FS_OPTION_OPENFILE_READ_POLICY;
+import static org.apache.hadoop.fs.Options.OpenFileOptions.FS_OPTION_OPENFILE_READ_POLICY_WHOLE_FILE;
+import static org.apache.hadoop.util.functional.FutureIO.awaitFuture;
+
 /**
 * Reads in history events from the JobHistoryFile and sends them out again
 * to be recorded.
@ -118,7 +122,11 @@ public class JobHistoryCopyService extends CompositeService implements HistoryEv
        fc.makeQualified(JobHistoryUtils.getStagingJobHistoryFile(histDirPath,
          jobId, (applicationAttemptId.getAttemptId() - 1)));
    LOG.info("History file is at " + historyFile);
-    in = fc.open(historyFile);
+    in = awaitFuture(
+        fc.openFile(historyFile)
+            .opt(FS_OPTION_OPENFILE_READ_POLICY,
+                FS_OPTION_OPENFILE_READ_POLICY_WHOLE_FILE)
+            .build());
    return in;
  }
  
--- a/hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-core/src/main/java/org/apache/hadoop/mapred/LineRecordReader.java
+++ b/hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-core/src/main/java/org/apache/hadoop/mapred/LineRecordReader.java
@ -28,7 +28,6 @@ import org.apache.hadoop.fs.FSDataInputStream;
 import org.apache.hadoop.fs.FutureDataInputStreamBuilder;
 import org.apache.hadoop.fs.Path;
 import org.apache.hadoop.fs.Seekable;
-import org.apache.hadoop.fs.impl.FutureIOSupport;
 import org.apache.hadoop.io.LongWritable;
 import org.apache.hadoop.io.Text;
 import org.apache.hadoop.io.compress.CodecPool;
@ -41,9 +40,13 @@ import org.apache.hadoop.mapreduce.MRJobConfig;
 import org.apache.hadoop.mapreduce.lib.input.CompressedSplitLineReader;
 import org.apache.hadoop.mapreduce.lib.input.SplitLineReader;
 import org.apache.hadoop.mapreduce.lib.input.UncompressedSplitLineReader;
+import org.apache.hadoop.util.functional.FutureIO;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;

+import static org.apache.hadoop.fs.Options.OpenFileOptions.FS_OPTION_OPENFILE_SPLIT_END;
+import static org.apache.hadoop.fs.Options.OpenFileOptions.FS_OPTION_OPENFILE_SPLIT_START;
+
 /**
 * Treats keys as offset in file and value as line. 
 */
@ -109,10 +112,14 @@ public class LineRecordReader implements RecordReader<LongWritable, Text> {
    // open the file and seek to the start of the split
    final FutureDataInputStreamBuilder builder =
        file.getFileSystem(job).openFile(file);
-    FutureIOSupport.propagateOptions(builder, job,
+    // the start and end of the split may be used to build
+    // an input strategy.
+    builder.opt(FS_OPTION_OPENFILE_SPLIT_START, start)
+        .opt(FS_OPTION_OPENFILE_SPLIT_END, end);
+    FutureIO.propagateOptions(builder, job,
        MRJobConfig.INPUT_FILE_OPTION_PREFIX,
        MRJobConfig.INPUT_FILE_MANDATORY_PREFIX);
-    fileIn = FutureIOSupport.awaitFuture(builder.build());
+    fileIn = FutureIO.awaitFuture(builder.build());
    if (isCompressedInput()) {
      decompressor = CodecPool.getDecompressor(codec);
      if (codec instanceof SplittableCompressionCodec) {
--- a/hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-core/src/main/java/org/apache/hadoop/mapreduce/lib/input/FixedLengthRecordReader.java
+++ b/hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-core/src/main/java/org/apache/hadoop/mapreduce/lib/input/FixedLengthRecordReader.java
@ -28,7 +28,6 @@ import org.apache.hadoop.fs.FSDataInputStream;
 import org.apache.hadoop.fs.FutureDataInputStreamBuilder;
 import org.apache.hadoop.fs.Path;
 import org.apache.hadoop.fs.Seekable;
-import org.apache.hadoop.fs.impl.FutureIOSupport;
 import org.apache.hadoop.io.BytesWritable;
 import org.apache.hadoop.io.LongWritable;
 import org.apache.hadoop.io.compress.CodecPool;
@ -40,6 +39,8 @@ import org.apache.hadoop.mapreduce.InputSplit;
 import org.apache.hadoop.mapreduce.MRJobConfig;
 import org.apache.hadoop.mapreduce.RecordReader;
 import org.apache.hadoop.mapreduce.TaskAttemptContext;
+import org.apache.hadoop.util.functional.FutureIO;
+
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;

@ -94,10 +95,10 @@ public class FixedLengthRecordReader
    // open the file
    final FutureDataInputStreamBuilder builder =
        file.getFileSystem(job).openFile(file);
-    FutureIOSupport.propagateOptions(builder, job,
+    FutureIO.propagateOptions(builder, job,
        MRJobConfig.INPUT_FILE_OPTION_PREFIX,
        MRJobConfig.INPUT_FILE_MANDATORY_PREFIX);
-    fileIn = FutureIOSupport.awaitFuture(builder.build());
+    fileIn = FutureIO.awaitFuture(builder.build());

    CompressionCodec codec = new CompressionCodecFactory(job).getCodec(file);
    if (null != codec) {
--- a/hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-core/src/main/java/org/apache/hadoop/mapreduce/lib/input/LineRecordReader.java
+++ b/hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-core/src/main/java/org/apache/hadoop/mapreduce/lib/input/LineRecordReader.java
@ -27,7 +27,6 @@ import org.apache.hadoop.fs.FSDataInputStream;
 import org.apache.hadoop.fs.FutureDataInputStreamBuilder;
 import org.apache.hadoop.fs.Path;
 import org.apache.hadoop.fs.Seekable;
-import org.apache.hadoop.fs.impl.FutureIOSupport;
 import org.apache.hadoop.io.LongWritable;
 import org.apache.hadoop.io.Text;
 import org.apache.hadoop.io.compress.CodecPool;
@ -40,9 +39,14 @@ import org.apache.hadoop.mapreduce.InputSplit;
 import org.apache.hadoop.mapreduce.MRJobConfig;
 import org.apache.hadoop.mapreduce.RecordReader;
 import org.apache.hadoop.mapreduce.TaskAttemptContext;
+import org.apache.hadoop.util.functional.FutureIO;
+
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;

+import static org.apache.hadoop.fs.Options.OpenFileOptions.FS_OPTION_OPENFILE_SPLIT_END;
+import static org.apache.hadoop.fs.Options.OpenFileOptions.FS_OPTION_OPENFILE_SPLIT_START;
+
 /**
 * Treats keys as offset in file and value as line. 
 */
@ -86,10 +90,14 @@ public class LineRecordReader extends RecordReader<LongWritable, Text> {
    // open the file and seek to the start of the split
    final FutureDataInputStreamBuilder builder =
        file.getFileSystem(job).openFile(file);
-    FutureIOSupport.propagateOptions(builder, job,
+    // the start and end of the split may be used to build
+    // an input strategy.
+    builder.opt(FS_OPTION_OPENFILE_SPLIT_START, start);
+    builder.opt(FS_OPTION_OPENFILE_SPLIT_END, end);
+    FutureIO.propagateOptions(builder, job,
        MRJobConfig.INPUT_FILE_OPTION_PREFIX,
        MRJobConfig.INPUT_FILE_MANDATORY_PREFIX);
-    fileIn = FutureIOSupport.awaitFuture(builder.build());
+    fileIn = FutureIO.awaitFuture(builder.build());
    
    CompressionCodec codec = new CompressionCodecFactory(job).getCodec(file);
    if (null!=codec) {
--- a/hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-core/src/main/java/org/apache/hadoop/mapreduce/lib/input/NLineInputFormat.java
+++ b/hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-core/src/main/java/org/apache/hadoop/mapreduce/lib/input/NLineInputFormat.java
@ -29,7 +29,6 @@ import org.apache.hadoop.fs.FSDataInputStream;
 import org.apache.hadoop.fs.FileStatus;
 import org.apache.hadoop.fs.FutureDataInputStreamBuilder;
 import org.apache.hadoop.fs.Path;
-import org.apache.hadoop.fs.impl.FutureIOSupport;
 import org.apache.hadoop.io.LongWritable;
 import org.apache.hadoop.io.Text;
 import org.apache.hadoop.mapreduce.InputSplit;
@ -39,6 +38,7 @@ import org.apache.hadoop.mapreduce.MRJobConfig;
 import org.apache.hadoop.mapreduce.RecordReader;
 import org.apache.hadoop.mapreduce.TaskAttemptContext;
 import org.apache.hadoop.util.LineReader;
+import org.apache.hadoop.util.functional.FutureIO;

 /**
 * NLineInputFormat which splits N lines of input as one split.
@ -99,10 +99,10 @@ public class NLineInputFormat extends FileInputFormat<LongWritable, Text> {
    try {
      final FutureDataInputStreamBuilder builder =
          fileName.getFileSystem(conf).openFile(fileName);
-      FutureIOSupport.propagateOptions(builder, conf,
+      FutureIO.propagateOptions(builder, conf,
          MRJobConfig.INPUT_FILE_OPTION_PREFIX,
          MRJobConfig.INPUT_FILE_MANDATORY_PREFIX);
-      FSDataInputStream in  = FutureIOSupport.awaitFuture(builder.build());
+      FSDataInputStream in  = FutureIO.awaitFuture(builder.build());
      lr = new LineReader(in, conf);
      Text line = new Text();
      int numLines = 0;
--- a/hadoop-mapreduce-project/hadoop-mapreduce-examples/src/main/java/org/apache/hadoop/examples/terasort/TeraInputFormat.java
+++ b/hadoop-mapreduce-project/hadoop-mapreduce-examples/src/main/java/org/apache/hadoop/examples/terasort/TeraInputFormat.java
@ -27,6 +27,7 @@ import java.util.List;
 import org.apache.hadoop.conf.Configuration;
 import org.apache.hadoop.fs.FSDataInputStream;
 import org.apache.hadoop.fs.FileSystem;
+import org.apache.hadoop.fs.FutureDataInputStreamBuilder;
 import org.apache.hadoop.fs.Path;
 import org.apache.hadoop.io.Text;
 import org.apache.hadoop.mapreduce.InputSplit;
@ -41,6 +42,12 @@ import org.apache.hadoop.mapreduce.task.TaskAttemptContextImpl;
 import org.apache.hadoop.util.IndexedSortable;
 import org.apache.hadoop.util.QuickSort;
 import org.apache.hadoop.util.StringUtils;
+import org.apache.hadoop.util.functional.FutureIO;
+
+import static org.apache.hadoop.fs.Options.OpenFileOptions.FS_OPTION_OPENFILE_READ_POLICY;
+import static org.apache.hadoop.fs.Options.OpenFileOptions.FS_OPTION_OPENFILE_READ_POLICY_SEQUENTIAL;
+import static org.apache.hadoop.fs.Options.OpenFileOptions.FS_OPTION_OPENFILE_SPLIT_END;
+import static org.apache.hadoop.fs.Options.OpenFileOptions.FS_OPTION_OPENFILE_SPLIT_START;

 /**
 * An input format that reads the first 10 characters of each line as the key
@ -224,12 +231,17 @@ public class TeraInputFormat extends FileInputFormat<Text,Text> {
        throws IOException, InterruptedException {
      Path p = ((FileSplit)split).getPath();
      FileSystem fs = p.getFileSystem(context.getConfiguration());
-      in = fs.open(p);
      long start = ((FileSplit)split).getStart();
      // find the offset to start at a record boundary
      offset = (RECORD_LENGTH - (start % RECORD_LENGTH)) % RECORD_LENGTH;
-      in.seek(start + offset);
      length = ((FileSplit)split).getLength();
+      final FutureDataInputStreamBuilder builder = fs.openFile(p)
+          .opt(FS_OPTION_OPENFILE_SPLIT_START, start)
+          .opt(FS_OPTION_OPENFILE_SPLIT_END, start + length)
+          .opt(FS_OPTION_OPENFILE_READ_POLICY,
+              FS_OPTION_OPENFILE_READ_POLICY_SEQUENTIAL);
+      in = FutureIO.awaitFuture(builder.build());
+      in.seek(start + offset);
    }

    public void close() throws IOException {
--- a/hadoop-tools/hadoop-distcp/src/main/java/org/apache/hadoop/tools/mapred/RetriableFileCopyCommand.java
+++ b/hadoop-tools/hadoop-distcp/src/main/java/org/apache/hadoop/tools/mapred/RetriableFileCopyCommand.java
@ -52,7 +52,10 @@ import org.apache.hadoop.tools.util.ThrottledInputStream;

 import org.apache.hadoop.classification.VisibleForTesting;

+import static org.apache.hadoop.fs.Options.OpenFileOptions.FS_OPTION_OPENFILE_READ_POLICY;
+import static org.apache.hadoop.fs.Options.OpenFileOptions.FS_OPTION_OPENFILE_READ_POLICY_SEQUENTIAL;
 import static org.apache.hadoop.tools.mapred.CopyMapper.getFileAttributeSettings;
+import static org.apache.hadoop.util.functional.FutureIO.awaitFuture;

 /**
 * This class extends RetriableCommand to implement the copy of files,
@ -362,7 +365,11 @@ public class RetriableFileCopyCommand extends RetriableCommand {
      FileSystem fs = path.getFileSystem(conf);
      float bandwidthMB = conf.getFloat(DistCpConstants.CONF_LABEL_BANDWIDTH_MB,
              DistCpConstants.DEFAULT_BANDWIDTH_MB);
-      FSDataInputStream in = fs.open(path);
+      // open with sequential read, but not whole-file
+      FSDataInputStream in = awaitFuture(fs.openFile(path)
+          .opt(FS_OPTION_OPENFILE_READ_POLICY,
+              FS_OPTION_OPENFILE_READ_POLICY_SEQUENTIAL)
+          .build());
      return new ThrottledInputStream(in, bandwidthMB * 1024 * 1024);
    }
    catch (IOException e) {
--- a/hadoop-tools/hadoop-streaming/src/main/java/org/apache/hadoop/streaming/mapreduce/StreamInputFormat.java
+++ b/hadoop-tools/hadoop-streaming/src/main/java/org/apache/hadoop/streaming/mapreduce/StreamInputFormat.java
@ -26,7 +26,6 @@ import org.apache.hadoop.fs.FSDataInputStream;
 import org.apache.hadoop.fs.FileSystem;
 import org.apache.hadoop.fs.FutureDataInputStreamBuilder;
 import org.apache.hadoop.fs.Path;
-import org.apache.hadoop.fs.impl.FutureIOSupport;
 import org.apache.hadoop.io.Text;
 import org.apache.hadoop.mapreduce.InputSplit;
 import org.apache.hadoop.mapreduce.MRJobConfig;
@ -35,6 +34,7 @@ import org.apache.hadoop.mapreduce.TaskAttemptContext;
 import org.apache.hadoop.mapreduce.lib.input.FileSplit;
 import org.apache.hadoop.mapreduce.lib.input.KeyValueTextInputFormat;
 import org.apache.hadoop.streaming.StreamUtil;
+import org.apache.hadoop.util.functional.FutureIO;

 /**
 * An input format that selects a RecordReader based on a JobConf property. This
@ -66,10 +66,10 @@ public class StreamInputFormat extends KeyValueTextInputFormat {
    FileSystem fs = path.getFileSystem(conf);
    // open the file
    final FutureDataInputStreamBuilder builder = fs.openFile(path);
-    FutureIOSupport.propagateOptions(builder, conf,
+    FutureIO.propagateOptions(builder, conf,
        MRJobConfig.INPUT_FILE_OPTION_PREFIX,
        MRJobConfig.INPUT_FILE_MANDATORY_PREFIX);
-    FSDataInputStream in = FutureIOSupport.awaitFuture(builder.build());
+    FSDataInputStream in = FutureIO.awaitFuture(builder.build());

    // Factory dispatch based on available params..
    Class readerClass;
--- a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-common/src/main/java/org/apache/hadoop/yarn/logaggregation/AggregatedLogFormat.java
+++ b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-common/src/main/java/org/apache/hadoop/yarn/logaggregation/AggregatedLogFormat.java
@ -56,6 +56,7 @@ import org.apache.hadoop.fs.CreateFlag;
 import org.apache.hadoop.fs.FSDataInputStream;
 import org.apache.hadoop.fs.FSDataOutputStream;
 import org.apache.hadoop.fs.FileContext;
+import org.apache.hadoop.fs.FileStatus;
 import org.apache.hadoop.fs.Options;
 import org.apache.hadoop.fs.Path;
 import org.apache.hadoop.fs.permission.FsPermission;
@ -77,6 +78,11 @@ import org.slf4j.LoggerFactory;
 import org.apache.hadoop.classification.VisibleForTesting;
 import org.apache.hadoop.thirdparty.com.google.common.collect.Iterables;

+import static org.apache.hadoop.fs.Options.OpenFileOptions.FS_OPTION_OPENFILE_READ_POLICY;
+import static org.apache.hadoop.fs.Options.OpenFileOptions.FS_OPTION_OPENFILE_READ_POLICY_SEQUENTIAL;
+import static org.apache.hadoop.fs.Options.OpenFileOptions.FS_OPTION_OPENFILE_LENGTH;
+import static org.apache.hadoop.util.functional.FutureIO.awaitFuture;
+
@Public
@Evolving
 public class AggregatedLogFormat {
@ -576,9 +582,16 @@ public class AggregatedLogFormat {
      try {
        FileContext fileContext =
            FileContext.getFileContext(remoteAppLogFile.toUri(), conf);
-        this.fsDataIStream = fileContext.open(remoteAppLogFile);
+        FileStatus status = fileContext.getFileStatus(remoteAppLogFile);
+        this.fsDataIStream = awaitFuture(
+            fileContext.openFile(remoteAppLogFile)
+                .opt(FS_OPTION_OPENFILE_READ_POLICY,
+                    FS_OPTION_OPENFILE_READ_POLICY_SEQUENTIAL)
+                .opt(FS_OPTION_OPENFILE_LENGTH,
+                    status.getLen())   // file length hint for object stores
+                .build());
        reader = new TFile.Reader(this.fsDataIStream,
-            fileContext.getFileStatus(remoteAppLogFile).getLen(), conf);
+            status.getLen(), conf);
        this.scanner = reader.createScanner();
      } catch (IOException ioe) {
        close();
--- a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-common/src/main/java/org/apache/hadoop/yarn/util/FSDownload.java
+++ b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-common/src/main/java/org/apache/hadoop/yarn/util/FSDownload.java
@ -60,7 +60,11 @@ import org.apache.hadoop.thirdparty.com.google.common.cache.LoadingCache;
 import org.apache.hadoop.thirdparty.com.google.common.util.concurrent.Futures;
 import org.apache.hadoop.yarn.exceptions.YarnException;

-/**
+import static org.apache.hadoop.fs.Options.OpenFileOptions.FS_OPTION_OPENFILE_READ_POLICY;
+import static org.apache.hadoop.fs.Options.OpenFileOptions.FS_OPTION_OPENFILE_READ_POLICY_WHOLE_FILE;
+import static org.apache.hadoop.util.functional.FutureIO.awaitFuture;
+
+ /**
 * Download a single URL to the local disk.
 *
 */
@ -285,23 +289,25 @@ public class FSDownload implements Callable<Path> {
      }
    }

-    downloadAndUnpack(sCopy, destination);
+    downloadAndUnpack(sCopy, sStat, destination);
  }

  /**
   * Copy source path to destination with localization rules.
-   * @param source source path to copy. Typically HDFS
+   * @param source source path to copy. Typically HDFS or an object store.
+   * @param sourceStatus status of source
   * @param destination destination path. Typically local filesystem
   * @exception YarnException Any error has occurred
   */
-  private void downloadAndUnpack(Path source, Path destination)
+  private void downloadAndUnpack(Path source,
+      FileStatus sourceStatus,  Path destination)
      throws YarnException {
    try {
      FileSystem sourceFileSystem = source.getFileSystem(conf);
      FileSystem destinationFileSystem = destination.getFileSystem(conf);
-      if (sourceFileSystem.getFileStatus(source).isDirectory()) {
+      if (sourceStatus.isDirectory()) {
        FileUtil.copy(
-            sourceFileSystem, source,
+            sourceFileSystem, sourceStatus,
            destinationFileSystem, destination, false,
            true, conf);
      } else {
@ -329,7 +335,11 @@ public class FSDownload implements Callable<Path> {
                      FileSystem sourceFileSystem,
                      FileSystem destinationFileSystem)
      throws IOException, InterruptedException, ExecutionException {
-    try (InputStream inputStream = sourceFileSystem.open(source)) {
+    try (InputStream inputStream = awaitFuture(
+        sourceFileSystem.openFile(source)
+            .opt(FS_OPTION_OPENFILE_READ_POLICY,
+                FS_OPTION_OPENFILE_READ_POLICY_WHOLE_FILE)
+            .build())) {
      File dst = new File(destination.toUri());
      String lowerDst = StringUtils.toLowerCase(dst.getName());
      switch (resource.getType()) {