HADOOP-18231. S3A prefetching: fix failing tests & drain stream async. (

apache#4386)


* adds in new test for prefetching input stream
* creates streamStats before opening stream
* updates numBlocks calculation method
* fixes ITestS3AOpenCost.testOpenFileLongerLength
* drains stream async
* fixes failing unit test


Contributed by Ahmar Suhail

hadoop-tools/hadoop-aws/src/main/java/org/apache/hadoop/fs/common/CachingBlockManager.java

@@ -31,6 +31,8 @@
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
+import static org.apache.hadoop.io.IOUtils.cleanupWithLogger;
+
 /**
  * Provides read access to the underlying file one block at a time.
  * Improve read performance by prefetching and locall caching blocks.
@@ -204,7 +206,7 @@ public synchronized void close() {
     // Cancel any prefetches in progress.
     this.cancelPrefetches();
 
-    Io.closeIgnoringIoException(this.cache);
+    cleanupWithLogger(LOG, this.cache);
 
     this.ops.end(op);
     LOG.info(this.ops.getSummary(false));

hadoop-tools/hadoop-aws/src/main/java/org/apache/hadoop/fs/s3a/S3AFileSystem.java

@@ -1524,7 +1524,8 @@ private FSDataInputStream executeOpen(
           new S3PrefetchingInputStream(
               readContext.build(),
               createObjectAttributes(path, fileStatus),
-              createInputStreamCallbacks(auditSpan)));
+              createInputStreamCallbacks(auditSpan),
+              inputStreamStats));
     } else {
       return new FSDataInputStream(
           new S3AInputStream(

hadoop-tools/hadoop-aws/src/main/java/org/apache/hadoop/fs/s3a/read/S3CachingInputStream.java

@@ -31,6 +31,7 @@
 import org.apache.hadoop.fs.s3a.S3AInputStream;
 import org.apache.hadoop.fs.s3a.S3AReadOpContext;
 import org.apache.hadoop.fs.s3a.S3ObjectAttributes;
+import org.apache.hadoop.fs.s3a.statistics.S3AInputStreamStatistics;
 
 /**
  * Provides an {@code InputStream} that allows reading from an S3 file.
@@ -53,6 +54,7 @@ public class S3CachingInputStream extends S3InputStream {
    * @param context read-specific operation context.
    * @param s3Attributes attributes of the S3 object being read.
    * @param client callbacks used for interacting with the underlying S3 client.
+   * @param streamStatistics statistics for this stream.
    *
    * @throws IllegalArgumentException if context is null.
    * @throws IllegalArgumentException if s3Attributes is null.
@@ -61,8 +63,9 @@ public class S3CachingInputStream extends S3InputStream {
   public S3CachingInputStream(
       S3AReadOpContext context,
       S3ObjectAttributes s3Attributes,
-      S3AInputStream.InputStreamCallbacks client) {
-    super(context, s3Attributes, client);
+      S3AInputStream.InputStreamCallbacks client,
+      S3AInputStreamStatistics streamStatistics) {
+    super(context, s3Attributes, client, streamStatistics);
 
     this.numBlocksToPrefetch = this.getContext().getPrefetchBlockCount();
     int bufferPoolSize = this.numBlocksToPrefetch + 1;

hadoop-tools/hadoop-aws/src/main/java/org/apache/hadoop/fs/s3a/read/S3File.java

@@ -19,18 +19,17 @@
 
 package org.apache.hadoop.fs.s3a.read;
 
-import java.io.Closeable;
+
 import java.io.IOException;
 import java.io.InputStream;
-import java.util.ArrayList;
 import java.util.IdentityHashMap;
-import java.util.List;
 import java.util.Map;
 
 import com.amazonaws.services.s3.model.GetObjectRequest;
 import com.amazonaws.services.s3.model.S3Object;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
 
-import org.apache.hadoop.fs.common.Io;
 import org.apache.hadoop.fs.common.Validate;
 import org.apache.hadoop.fs.s3a.Invoker;
 import org.apache.hadoop.fs.s3a.S3AInputStream;
@@ -40,30 +39,56 @@
 import org.apache.hadoop.fs.s3a.statistics.S3AInputStreamStatistics;
 import org.apache.hadoop.fs.statistics.DurationTracker;
 
+import static org.apache.hadoop.fs.statistics.impl.IOStatisticsBinding.invokeTrackingDuration;
+import static org.apache.hadoop.io.IOUtils.cleanupWithLogger;
+
 /**
  * Encapsulates low level interactions with S3 object on AWS.
  */
-public class S3File implements Closeable {
+public class S3File {
+  private static final Logger LOG = LoggerFactory.getLogger(S3File.class);
 
-  // Read-specific operation context.
+  /**
+   * Read-specific operation context.
+   */
   private final S3AReadOpContext context;
 
-  // S3 object attributes.
+  /**
+   * S3 object attributes.
+   */
   private final S3ObjectAttributes s3Attributes;
 
-  // Callbacks used for interacting with the underlying S3 client.
+  /**
+   * Callbacks used for interacting with the underlying S3 client.
+   */
   private final S3AInputStream.InputStreamCallbacks client;
 
-  // Used for reporting input stream access statistics.
+  /**
+   * Used for reporting input stream access statistics.
+   */
   private final S3AInputStreamStatistics streamStatistics;
 
-  // Enforces change tracking related policies.
+  /**
+   * Enforces change tracking related policies.
+   */
   private final ChangeTracker changeTracker;
 
-  // Maps a stream returned by openForRead() to the associated S3 object.
-  // That allows us to close the object when closing the stream.
+  /**
+   * Maps a stream returned by openForRead() to the associated S3 object.
+   * That allows us to close the object when closing the stream.
+   */
   private Map<InputStream, S3Object> s3Objects;
 
+  /**
+   * uri of the object being read.
+   */
+  private final String uri;
+
+  /**
+   * size of a buffer to create when draining the stream.
+   */
+  private static final int DRAIN_BUFFER_SIZE = 16384;
+
   /**
    * Initializes a new instance of the {@code S3File} class.
    *
@@ -97,7 +122,8 @@ public S3File(
     this.client = client;
     this.streamStatistics = streamStatistics;
     this.changeTracker = changeTracker;
-    this.s3Objects = new IdentityHashMap<InputStream, S3Object>();
+    this.s3Objects = new IdentityHashMap<>();
+    this.uri = this.getPath();
   }
 
   /**
@@ -169,7 +195,6 @@ public InputStream openForRead(long offset, int size) throws IOException {
         .withRange(offset, offset + size - 1);
     this.changeTracker.maybeApplyConstraint(request);
 
-    String uri = this.getPath();
     String operation = String.format(
         "%s %s at %d", S3AInputStream.OPERATION_OPEN, uri, offset);
     DurationTracker tracker = streamStatistics.initiateGetRequest();
@@ -193,18 +218,7 @@ public InputStream openForRead(long offset, int size) throws IOException {
     return stream;
   }
 
-  /**
-   * Closes this stream and releases all acquired resources.
-   */
-  @Override
-  public synchronized void close() {
-    List<InputStream> streams = new ArrayList<InputStream>(this.s3Objects.keySet());
-    for (InputStream stream : streams) {
-      this.close(stream);
-    }
-  }
-
-  void close(InputStream inputStream) {
+  void close(InputStream inputStream, int numRemainingBytes) {
     S3Object obj;
     synchronized (this.s3Objects) {
       obj = this.s3Objects.get(inputStream);
@@ -214,7 +228,91 @@ void close(InputStream inputStream) {
       this.s3Objects.remove(inputStream);
     }
 
-    Io.closeIgnoringIoException(inputStream);
-    Io.closeIgnoringIoException(obj);
+    if (numRemainingBytes <= this.context.getAsyncDrainThreshold()) {
+      // don't bother with async io.
+      drain(false, "close() operation", numRemainingBytes, obj, inputStream);
+    } else {
+      LOG.debug("initiating asynchronous drain of {} bytes", numRemainingBytes);
+      // schedule an async drain/abort with references to the fields so they
+      // can be reused
+      client.submit(() -> drain(false, "close() operation", numRemainingBytes, obj, inputStream));
+    }
+  }
+
+  /**
+   * drain the stream. This method is intended to be
+   * used directly or asynchronously, and measures the
+   * duration of the operation in the stream statistics.
+   *
+   * @param shouldAbort   force an abort; used if explicitly requested.
+   * @param reason        reason for stream being closed; used in messages
+   * @param remaining     remaining bytes
+   * @param requestObject http request object;
+   * @param inputStream   stream to close.
+   * @return was the stream aborted?
+   */
+  private boolean drain(
+      final boolean shouldAbort,
+      final String reason,
+      final long remaining,
+      final S3Object requestObject,
+      final InputStream inputStream) {
+
+    try {
+      return invokeTrackingDuration(streamStatistics.initiateInnerStreamClose(shouldAbort),
+          () -> drainOrAbortHttpStream(shouldAbort, reason, remaining, requestObject, inputStream));
+    } catch (IOException e) {
+      // this is only here because invokeTrackingDuration() has it in its
+      // signature
+      return shouldAbort;
+    }
+  }
+
+  /**
+   * Drain or abort the inner stream.
+   * Exceptions are swallowed.
+   * If a close() is attempted and fails, the operation escalates to
+   * an abort.
+   *
+   * @param shouldAbort   force an abort; used if explicitly requested.
+   * @param reason        reason for stream being closed; used in messages
+   * @param remaining     remaining bytes
+   * @param requestObject http request object
+   * @param inputStream   stream to close.
+   * @return was the stream aborted?
+   */
+  private boolean drainOrAbortHttpStream(
+      boolean shouldAbort,
+      final String reason,
+      final long remaining,
+      final S3Object requestObject,
+      final InputStream inputStream) {
+
+    if (!shouldAbort && remaining > 0) {
+      try {
+        long drained = 0;
+        byte[] buffer = new byte[DRAIN_BUFFER_SIZE];
+        while (true) {
+          final int count = inputStream.read(buffer);
+          if (count < 0) {
+            // no more data is left
+            break;
+          }
+          drained += count;
+        }
+        LOG.debug("Drained stream of {} bytes", drained);
+      } catch (Exception e) {
+        // exception escalates to an abort
+        LOG.debug("When closing {} stream for {}, will abort the stream", uri, reason, e);
+        shouldAbort = true;
+      }
+    }
+    cleanupWithLogger(LOG, inputStream);
+    cleanupWithLogger(LOG, requestObject);
+    streamStatistics.streamClose(shouldAbort, remaining);
+
+    LOG.debug("Stream {} {}: {}; remaining={}", uri, (shouldAbort ? "aborted" : "closed"), reason,
+        remaining);
+    return shouldAbort;
   }
 }

hadoop-tools/hadoop-aws/src/main/java/org/apache/hadoop/fs/s3a/read/S3InMemoryInputStream.java

@@ -29,6 +29,7 @@
 import org.apache.hadoop.fs.s3a.S3AInputStream;
 import org.apache.hadoop.fs.s3a.S3AReadOpContext;
 import org.apache.hadoop.fs.s3a.S3ObjectAttributes;
+import org.apache.hadoop.fs.s3a.statistics.S3AInputStreamStatistics;
 
 /**
  * Provides an {@code InputStream} that allows reading from an S3 file.
@@ -48,6 +49,7 @@ public class S3InMemoryInputStream extends S3InputStream {
    * @param context read-specific operation context.
    * @param s3Attributes attributes of the S3 object being read.
    * @param client callbacks used for interacting with the underlying S3 client.
+   * @param streamStatistics statistics for this stream.
    *
    * @throws IllegalArgumentException if context is null.
    * @throws IllegalArgumentException if s3Attributes is null.
@@ -56,8 +58,9 @@ public class S3InMemoryInputStream extends S3InputStream {
   public S3InMemoryInputStream(
       S3AReadOpContext context,
       S3ObjectAttributes s3Attributes,
-      S3AInputStream.InputStreamCallbacks client) {
-    super(context, s3Attributes, client);
+      S3AInputStream.InputStreamCallbacks client,
+      S3AInputStreamStatistics streamStatistics) {
+    super(context, s3Attributes, client, streamStatistics);
     int fileSize = (int) s3Attributes.getLen();
     this.buffer = ByteBuffer.allocate(fileSize);
     LOG.debug("Created in-memory input stream for {} (size = {})", this.getName(), fileSize);

hadoop-tools/hadoop-aws/src/main/java/org/apache/hadoop/fs/s3a/read/S3InputStream.java

@@ -44,6 +44,8 @@
 import org.apache.hadoop.fs.statistics.IOStatistics;
 import org.apache.hadoop.fs.statistics.IOStatisticsSource;
 
+import static java.util.Objects.requireNonNull;
+
 /**
  * Provides an {@link InputStream} that allows reading from an S3 file.
  */
@@ -96,6 +98,7 @@ public abstract class S3InputStream
    * @param context read-specific operation context.
    * @param s3Attributes attributes of the S3 object being read.
    * @param client callbacks used for interacting with the underlying S3 client.
+   * @param streamStatistics statistics for this stream.
    *
    * @throws IllegalArgumentException if context is null.
    * @throws IllegalArgumentException if s3Attributes is null.
@@ -104,16 +107,13 @@ public abstract class S3InputStream
   public S3InputStream(
       S3AReadOpContext context,
       S3ObjectAttributes s3Attributes,
-      S3AInputStream.InputStreamCallbacks client) {
-
-    Validate.checkNotNull(context, "context");
-    Validate.checkNotNull(s3Attributes, "s3Attributes");
-    Validate.checkNotNull(client, "client");
+      S3AInputStream.InputStreamCallbacks client,
+      S3AInputStreamStatistics streamStatistics) {
 
-    this.context = context;
-    this.s3Attributes = s3Attributes;
-    this.client = client;
-    this.streamStatistics = context.getS3AStatisticsContext().newInputStreamStatistics();
+    this.context = requireNonNull(context);
+    this.s3Attributes = requireNonNull(s3Attributes);
+    this.client = requireNonNull(client);
+    this.streamStatistics = requireNonNull(streamStatistics);
     this.ioStatistics = streamStatistics.getIOStatistics();
     this.name = S3File.getPath(s3Attributes);
     this.changeTracker = new ChangeTracker(