From 4cf7f8441a288cefd44f126d60dae1998239892a Mon Sep 17 00:00:00 2001
From: Kihwal Lee <kihwal@apache.org>
Date: Thu, 15 Oct 2015 11:26:14 -0500
Subject: [PATCH] HDFS-9220. Reading small file (< 512 bytes) that is open for
 append fails due to incorrect checksum. Contributed by Jing Zhao. (cherry
 picked from commit c7c36cbd6218f46c33d7fb2f60cd52cb29e6d720)

---
 hadoop-hdfs-project/hadoop-hdfs/CHANGES.txt   |  3 ++
 .../hdfs/server/datanode/BlockReceiver.java   |  5 +--
 .../apache/hadoop/hdfs/TestFileAppend2.java   | 37 +++++++++++++++++++
 3 files changed, 42 insertions(+), 3 deletions(-)

diff --git a/hadoop-hdfs-project/hadoop-hdfs/CHANGES.txt b/hadoop-hdfs-project/hadoop-hdfs/CHANGES.txt
index b828e791930..d7884e85f7a 100644
--- a/hadoop-hdfs-project/hadoop-hdfs/CHANGES.txt
+++ b/hadoop-hdfs-project/hadoop-hdfs/CHANGES.txt
@@ -69,6 +69,9 @@ Release 2.7.2 - UNRELEASED
     HDFS-8676. Delayed rolling upgrade finalization can cause heartbeat
     expiration. (Walter Su via kihwal)
 
+    HDFS-9220. Reading small file (< 512 bytes) that is open for append fails
+    due to incorrect checksum (Jing Zhao via kihwal)
+
 Release 2.7.1 - 2015-07-06
 
   INCOMPATIBLE CHANGES
diff --git a/hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/datanode/BlockReceiver.java b/hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/datanode/BlockReceiver.java
index 3be8d4c5e92..27783e247d1 100644
--- a/hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/datanode/BlockReceiver.java
+++ b/hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/datanode/BlockReceiver.java
@@ -747,11 +747,10 @@ class BlockReceiver implements Closeable {
             final int offset = checksumBuf.arrayOffset() +
                 checksumBuf.position() + skip;
             final int end = offset + checksumLen - skip;
-            // If offset > end, there is no more checksum to write.
+            // If offset >= end, there is no more checksum to write.
             // I.e. a partial chunk checksum rewrite happened and there is no
             // more to write after that.
-            if (offset > end) {
-              assert crcBytes != null;
+            if (offset >= end && doCrcRecalc) {
               lastCrc = crcBytes;
             } else {
               final int remainingBytes = checksumLen - skip;
diff --git a/hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/TestFileAppend2.java b/hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/TestFileAppend2.java
index dd4fe144406..22dd4a0137a 100644
--- a/hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/TestFileAppend2.java
+++ b/hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/TestFileAppend2.java
@@ -31,6 +31,7 @@ import java.util.List;
 import org.apache.commons.logging.impl.Log4JLogger;
 import org.apache.hadoop.conf.Configuration;
 import org.apache.hadoop.fs.CreateFlag;
+import org.apache.hadoop.fs.FSDataInputStream;
 import org.apache.hadoop.fs.FSDataOutputStream;
 import org.apache.hadoop.fs.FileSystem;
 import org.apache.hadoop.fs.Path;
@@ -549,4 +550,40 @@ public class TestFileAppend2 {
   public void testComplexAppend2() throws IOException {
     testComplexAppend(true);
   }
+
+  /**
+   * Make sure when the block length after appending is less than 512 bytes, the
+   * checksum re-calculation and overwrite are performed correctly.
+   */
+  @Test
+  public void testAppendLessThanChecksumChunk() throws Exception {
+    final byte[] buf = new byte[1024];
+    final MiniDFSCluster cluster = new MiniDFSCluster
+        .Builder(new HdfsConfiguration()).numDataNodes(1).build();
+    cluster.waitActive();
+
+    try (DistributedFileSystem fs = cluster.getFileSystem()) {
+      final int len1 = 200;
+      final int len2 = 300;
+      final Path p = new Path("/foo");
+
+      FSDataOutputStream out = fs.create(p);
+      out.write(buf, 0, len1);
+      out.close();
+
+      out = fs.append(p);
+      out.write(buf, 0, len2);
+      // flush but leave open
+      out.hflush();
+
+      // read data to verify the replica's content and checksum are correct
+      FSDataInputStream in = fs.open(p);
+      final int length = in.read(0, buf, 0, len1 + len2);
+      assertTrue(length > 0);
+      in.close();
+      out.close();
+    } finally {
+      cluster.shutdown();
+    }
+  }
 }