HADOOP-16202. Enhance openFile()

Roll-up of the previous PR

Change-Id: Ib0aec173afcd8aae33f52da3f99ac813bd38c32f

HADOOP-16202. javadocs and style

Change-Id: Id4294ac7034155a10be22fb4631edf43cbadc22b

HADOOP-16202. openFile: read policies

Change "fs.option.openfile.fadvise" to "fs.option.openfile.read.policy"
and expand with "vectored", "parquet" and "orc", all of which map in s3a
to random.

The concept is that by choosing a read policy you can do more than just
change seek policy -it could switch buffering, caching etc.

Change-Id: I2147840f58fb54853c797d2cab5d668c3d1d2541

HADOOP-16202: documentation changes and IOStatistics of open operations

* Thomas Marquardt's suggestions on the docs
* standard action name for file opened
* S3AInputStream measures the count and duration of this, and reports it

Change-Id: I7feacf4eb4d6494bb93b3dfc05b060ad75e52c18

Author: steveloughran

hadoop-common-project/hadoop-common/src/main/java/org/apache/hadoop/fs/AvroFSInput.java

@@ -25,6 +25,11 @@
 import org.apache.hadoop.classification.InterfaceAudience;
 import org.apache.hadoop.classification.InterfaceStability;
 
+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;
+
 /** Adapts an {@link FSDataInputStream} to Avro's SeekableInput interface. */
 @InterfaceAudience.Public
 @InterfaceStability.Stable
@@ -42,7 +47,13 @@ public AvroFSInput(final FSDataInputStream in, final long len) {
   public AvroFSInput(final FileContext fc, final Path p) throws IOException {
     FileStatus status = fc.getFileStatus(p);
     this.len = status.getLen();
-    this.stream = fc.open(p);
+    this.stream = awaitFuture(fc.openFile(p)
+        .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());
+    fc.open(p);
   }
 
   @Override

hadoop-common-project/hadoop-common/src/main/java/org/apache/hadoop/fs/ChecksumFileSystem.java

@@ -24,7 +24,6 @@
 import java.io.InputStream;
 import java.nio.channels.ClosedChannelException;
 import java.util.Arrays;
-import java.util.Collections;
 import java.util.EnumSet;
 import java.util.List;
 import java.util.concurrent.CompletableFuture;
@@ -45,6 +44,7 @@
 import org.apache.hadoop.util.LambdaUtils;
 import org.apache.hadoop.util.Progressable;
 
+import static org.apache.hadoop.fs.Options.OpenFileOptions.FS_OPTION_OPENFILE_STANDARD_OPTIONS;
 import static org.apache.hadoop.fs.impl.PathCapabilitiesSupport.validatePathCapabilityArgs;
 import static org.apache.hadoop.fs.impl.StoreImplementationUtils.isProbeForSyncable;
 
@@ -889,7 +889,7 @@ protected CompletableFuture<FSDataInputStream> openFileWithOptions(
       final OpenFileParameters parameters) throws IOException {
     AbstractFSBuilderImpl.rejectUnknownMandatoryKeys(
         parameters.getMandatoryKeys(),
-        Collections.emptySet(),
+        FS_OPTION_OPENFILE_STANDARD_OPTIONS,
         "for " + path);
     return LambdaUtils.eval(
         new CompletableFuture<>(),

hadoop-common-project/hadoop-common/src/main/java/org/apache/hadoop/fs/FSBuilder.java

@@ -61,6 +61,13 @@ public interface FSBuilder<S, B extends FSBuilder<S, B>> {
    */
   B opt(@Nonnull String key, float value);
 
+  /**
+   * Set optional long parameter for the Builder.
+   *
+   * @see #opt(String, String)
+   */
+  B opt(@Nonnull String key, long value);
+
   /**
    * Set optional double parameter for the Builder.
    *
@@ -104,6 +111,13 @@ public interface FSBuilder<S, B extends FSBuilder<S, B>> {
    */
   B must(@Nonnull String key, float value);
 
+  /**
+   * Set mandatory long option.
+   *
+   * @see #must(String, String)
+   */
+  B must(@Nonnull String key, long value);
+
   /**
    * Set mandatory double option.
    *

hadoop-common-project/hadoop-common/src/main/java/org/apache/hadoop/fs/FileContext.java

@@ -71,7 +71,12 @@
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
+import static org.apache.hadoop.fs.Options.OpenFileOptions.FS_OPTION_OPENFILE_BUFFER_SIZE;
+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.fs.impl.PathCapabilitiesSupport.validatePathCapabilityArgs;
+import static org.apache.hadoop.util.functional.FutureIO.awaitFuture;
 
 /**
  * The FileContext class provides an interface for users of the Hadoop
@@ -2204,7 +2209,12 @@ public boolean copy(final Path src, final Path dst, boolean deleteSource,
         EnumSet<CreateFlag> createFlag = overwrite ? EnumSet.of(
             CreateFlag.CREATE, CreateFlag.OVERWRITE) :
             EnumSet.of(CreateFlag.CREATE);
-        InputStream in = open(qSrc);
+        InputStream in = awaitFuture(openFile(qSrc)
+            .opt(FS_OPTION_OPENFILE_READ_POLICY,
+                FS_OPTION_OPENFILE_READ_POLICY_SEQUENTIAL)
+            .opt(FS_OPTION_OPENFILE_LENGTH,
+                fs.getLen())   // file length hint for object stores
+            .build());
         try (OutputStream out = create(qDst, createFlag)) {
           IOUtils.copyBytes(in, out, conf, true);
         } finally {
@@ -2936,9 +2946,13 @@ public CompletableFuture<FSDataInputStream> build() throws IOException {
       final Path absF = fixRelativePart(getPath());
       OpenFileParameters parameters = new OpenFileParameters()
           .withMandatoryKeys(getMandatoryKeys())
+          .withOptionalKeys(getOptionalKeys())
           .withOptions(getOptions())
           .withBufferSize(getBufferSize())
           .withStatus(getStatus());
+      parameters.withBufferSize(
+          getOptions().getInt(FS_OPTION_OPENFILE_BUFFER_SIZE,
+              getBufferSize()));
       return new FSLinkResolver<CompletableFuture<FSDataInputStream>>() {
         @Override
         public CompletableFuture<FSDataInputStream> next(

hadoop-common-project/hadoop-common/src/main/java/org/apache/hadoop/fs/FileSystem.java

@@ -89,6 +89,7 @@
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
+import static org.apache.hadoop.fs.Options.OpenFileOptions.FS_OPTION_OPENFILE_BUFFER_SIZE;
 import static org.apache.hadoop.thirdparty.com.google.common.base.Preconditions.checkArgument;
 import static org.apache.hadoop.fs.CommonConfigurationKeysPublic.*;
 import static org.apache.hadoop.fs.impl.PathCapabilitiesSupport.validatePathCapabilityArgs;
@@ -4617,7 +4618,7 @@ protected CompletableFuture<FSDataInputStream> openFileWithOptions(
       final OpenFileParameters parameters) throws IOException {
     AbstractFSBuilderImpl.rejectUnknownMandatoryKeys(
         parameters.getMandatoryKeys(),
-        Collections.emptySet(),
+        Options.OpenFileOptions.FS_OPTION_OPENFILE_STANDARD_OPTIONS,
         "for " + path);
     return LambdaUtils.eval(
         new CompletableFuture<>(), () ->
@@ -4645,7 +4646,7 @@ protected CompletableFuture<FSDataInputStream> openFileWithOptions(
       final OpenFileParameters parameters) throws IOException {
     AbstractFSBuilderImpl.rejectUnknownMandatoryKeys(
         parameters.getMandatoryKeys(),
-        Collections.emptySet(), "");
+        Options.OpenFileOptions.FS_OPTION_OPENFILE_STANDARD_OPTIONS, "");
     CompletableFuture<FSDataInputStream> result = new CompletableFuture<>();
     try {
       result.complete(open(pathHandle, parameters.getBufferSize()));
@@ -4752,9 +4753,13 @@ public CompletableFuture<FSDataInputStream> build() throws IOException {
       Optional<Path> optionalPath = getOptionalPath();
       OpenFileParameters parameters = new OpenFileParameters()
           .withMandatoryKeys(getMandatoryKeys())
+          .withOptionalKeys(getOptionalKeys())
           .withOptions(getOptions())
-          .withBufferSize(getBufferSize())
           .withStatus(super.getStatus());  // explicit to avoid IDE warnings
+      // buffer size can be configured
+      parameters.withBufferSize(
+          getOptions().getInt(FS_OPTION_OPENFILE_BUFFER_SIZE,
+              getBufferSize()));
       if(optionalPath.isPresent()) {
         return getFS().openFileWithOptions(optionalPath.get(),
             parameters);

hadoop-common-project/hadoop-common/src/main/java/org/apache/hadoop/fs/FileUtil.java

@@ -71,6 +71,11 @@
 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_SEQUENTIAL;
+import static org.apache.hadoop.fs.Options.OpenFileOptions.FS_OPTION_OPENFILE_LENGTH;
+import static org.apache.hadoop.util.functional.FutureIO.awaitFuture;
+
 /**
  * A collection of file-processing util methods
  */
@@ -390,7 +395,32 @@ public static boolean copy(FileSystem srcFS, Path src,
     return copy(srcFS, fileStatus, dstFS, dst, deleteSource, overwrite, conf);
   }
 
-  /** Copy files between FileSystems. */
+  /**
+   * Copy a file/directory tree within/between filesystems.
+   * <p></p>
+   * returns true if the operation succeeded. When deleteSource is true,
+   * this means "after the copy, delete(source) returned true"
+   * If the destination is a directory, and mkdirs (dest) fails,
+   * the operation will return false rather than raise any exception.
+   * <p></p>
+   * The overwrite flag is about overwriting files; it has no effect about
+   * handing an attempt to copy a file atop a directory (expect an IOException),
+   * or a directory over a path which contains a file (mkdir will fail, so
+   * "false").
+   * <p></p>
+   * The operation is recursive, and the deleteSource operation takes place
+   * as each subdirectory is copied. Therefore, if an operation fails partway
+   * through, the source tree may be partially deleted.
+   * @param srcFS source filesystem
+   * @param srcStatus status of source
+   * @param dstFS destination filesystem
+   * @param dst path of source
+   * @param deleteSource delete the source?
+   * @param overwrite overwrite files at destination?
+   * @param conf configuration to use when opening files
+   * @return true if the operation succeeded.
+   * @throws IOException failure
+   */
   public static boolean copy(FileSystem srcFS, FileStatus srcStatus,
                              FileSystem dstFS, Path dst,
                              boolean deleteSource,
@@ -409,22 +439,27 @@ public static boolean copy(FileSystem srcFS, FileStatus srcStatus,
       if (!dstFS.mkdirs(dst)) {
         return false;
       }
-      FileStatus contents[] = srcFS.listStatus(src);
-      for (int i = 0; i < contents.length; i++) {
-        copy(srcFS, contents[i], dstFS,
-             new Path(dst, contents[i].getPath().getName()),
-             deleteSource, overwrite, conf);
+      RemoteIterator<FileStatus> contents = srcFS.listStatusIterator(src);
+      while (contents.hasNext()) {
+        FileStatus next = contents.next();
+        copy(srcFS, next, dstFS,
+            new Path(dst, next.getPath().getName()),
+            deleteSource, overwrite, conf);
       }
     } else {
-      InputStream in=null;
+      InputStream in = null;
       OutputStream out = null;
       try {
-        in = srcFS.open(src);
+        in = awaitFuture(srcFS.openFile(src)
+            .opt(FS_OPTION_OPENFILE_READ_POLICY,
+                FS_OPTION_OPENFILE_READ_POLICY_SEQUENTIAL)
+            .opt(FS_OPTION_OPENFILE_LENGTH,
+                srcStatus.getLen())   // file length hint for object stores
+            .build());
         out = dstFS.create(dst, overwrite);
         IOUtils.copyBytes(in, out, conf, true);
       } catch (IOException e) {
-        IOUtils.closeStream(out);
-        IOUtils.closeStream(in);
+        IOUtils.cleanupWithLogger(LOG, in, out);
         throw e;
       }
     }

hadoop-common-project/hadoop-common/src/main/java/org/apache/hadoop/fs/FutureDataInputStreamBuilder.java

@@ -17,6 +17,7 @@
  */
 package org.apache.hadoop.fs;
 
+import javax.annotation.Nullable;
 import java.io.IOException;
 import java.util.concurrent.CompletableFuture;
 
@@ -34,7 +35,7 @@
  * options accordingly, for example:
  *
  * If the option is not related to the file system, the option will be ignored.
- * If the option is must, but not supported by the file system, a
+ * If the option is must, but not supported/known by the file system, an
  * {@link IllegalArgumentException} will be thrown.
  *
  */
@@ -51,10 +52,11 @@ CompletableFuture<FSDataInputStream> build()
   /**
    * A FileStatus may be provided to the open request.
    * It is up to the implementation whether to use this or not.
-   * @param status status.
+   * @param status status: may be null
    * @return the builder.
    */
-  default FutureDataInputStreamBuilder withFileStatus(FileStatus status) {
+  default FutureDataInputStreamBuilder withFileStatus(
+      @Nullable FileStatus status) {
     return this;
   }