HBASE-4510 Check and workaround usage of internal HDFS APIs in HBase

git-svn-id: https://svn.apache.org/repos/asf/hbase/trunk@1186992 13f79535-47bb-0310-9956-ffa450edef68

CHANGES.txt

@@ -381,6 +381,8 @@ Release 0.92.0 - Unreleased
    HBASE-4430  Disable TestSlabCache and TestSingleSizedCache temporarily to
                see if these are cause of build box failure though all tests
                pass (Li Pi)
+   HBASE-4510  Check and workaround usage of internal HDFS APIs in HBase
+               (Harsh)
 
   TESTS
    HBASE-4450  test for number of blocks read: to serve as baseline for expected

src/main/java/org/apache/hadoop/hbase/util/FSHDFSUtils.java

@@ -32,7 +32,6 @@ import org.apache.hadoop.fs.Path;
 import org.apache.hadoop.hbase.RemoteExceptionHandler;
 import org.apache.hadoop.hdfs.DistributedFileSystem;
 import org.apache.hadoop.hdfs.protocol.AlreadyBeingCreatedException;
-import org.apache.hadoop.hdfs.protocol.FSConstants;
 import org.apache.hadoop.hdfs.server.namenode.LeaseExpiredException;
 
 
@@ -42,6 +41,16 @@ import org.apache.hadoop.hdfs.server.namenode.LeaseExpiredException;
 public class FSHDFSUtils extends FSUtils{
   private static final Log LOG = LogFactory.getLog(FSHDFSUtils.class);
 
+  /**
+   * Lease timeout constant, sourced from HDFS upstream.
+   * The upstream constant is defined in a private interface, so we
+   * can't reuse for compatibility reasons.
+   * NOTE: On versions earlier than Hadoop 0.23, the constant is in
+   * o.a.h.hdfs.protocol.FSConstants, while for 0.23 and above it is
+   * in o.a.h.hdfs.protocol.HdfsConstants cause of HDFS-1620.
+   */
+  public static final long LEASE_SOFTLIMIT_PERIOD = 60 * 1000;
+
   public void recoverFileLease(final FileSystem fs, final Path p, Configuration conf)
   throws IOException{
     if (!isAppendSupported(conf)) {
@@ -86,7 +95,7 @@ public class FSHDFSUtils extends FSUtils{
           // that the RS is holding onto the file even though it lost its
           // znode. We could potentially abort after some time here.
           long waitedFor = System.currentTimeMillis() - startWaiting;
-          if (waitedFor > FSConstants.LEASE_SOFTLIMIT_PERIOD) {
+          if (waitedFor > LEASE_SOFTLIMIT_PERIOD) {
             LOG.warn("Waited " + waitedFor + "ms for lease recovery on " + p +
               ":" + e.getMessage());
           }

src/main/java/org/apache/hadoop/hbase/util/FSUtils.java

@@ -29,6 +29,7 @@ import org.apache.hadoop.fs.FileStatus;
 import org.apache.hadoop.fs.FileSystem;
 import org.apache.hadoop.fs.Path;
 import org.apache.hadoop.fs.PathFilter;
+import org.apache.hadoop.fs.permission.FsPermission;
 import org.apache.hadoop.hbase.HConstants;
 import org.apache.hadoop.hbase.HDFSBlocksDistribution;
 import org.apache.hadoop.hbase.HRegionInfo;
@@ -38,6 +39,7 @@ import org.apache.hadoop.hbase.master.HMaster;
 import org.apache.hadoop.hbase.regionserver.HRegion;
 import org.apache.hadoop.hdfs.DistributedFileSystem;
 import org.apache.hadoop.hdfs.protocol.FSConstants;
+import org.apache.hadoop.hdfs.server.namenode.SafeModeException;
 import org.apache.hadoop.io.SequenceFile;
 import org.apache.hadoop.util.ReflectionUtils;
 import org.apache.hadoop.util.StringUtils;
@@ -150,22 +152,37 @@ public abstract class FSUtils {
     throw io;
   }
 
+  /**
+   * Utility to check if provided FS is in safemode.
+   * @return true if dfs is in safemode, false otherwise.
+   *
+   */
+  private static boolean isInSafeMode(FileSystem fs) throws IOException {
+    // Refactored safe-mode check for HBASE-4510
+    if (fs instanceof DistributedFileSystem) {
+      Path rootPath = new Path("/");
+      FsPermission rootPerm = fs.getFileStatus(rootPath).getPermission();
+      try {
+        // Should be harmless to set back the path we retrieved.
+        // The first check server-side is the safemode, so if
+        // other exceptions are spewed out, we're not interested.
+        fs.setPermission(rootPath, rootPerm);
+      } catch (SafeModeException e) {
+        return true;
+      }
+    }
+    return false;
+  }
+
   /**
    * Check whether dfs is in safemode. 
-   * @param conf 
+   * @param conf Configuration to use
-   * @return true if dfs is in safemode.
+   * @throws IOException if dfs is in safemode
-   * @throws IOException
    */
   public static void checkDfsSafeMode(final Configuration conf) 
   throws IOException {
-    boolean isInSafeMode = false;
     FileSystem fs = FileSystem.get(conf);
-    if (fs instanceof DistributedFileSystem) {
+    if (isInSafeMode(fs)) {
-      DistributedFileSystem dfs = (DistributedFileSystem)fs;
-      // Check whether dfs is on safemode.
-      isInSafeMode = dfs.setSafeMode(FSConstants.SafeModeAction.SAFEMODE_GET);
-    }
-    if (isInSafeMode) {
       throw new IOException("File system is in safemode, it can't be written now");
     }
   }
@@ -436,10 +453,8 @@ public abstract class FSUtils {
     final long wait)
   throws IOException {
     FileSystem fs = FileSystem.get(conf);
-    if (!(fs instanceof DistributedFileSystem)) return;
-    DistributedFileSystem dfs = (DistributedFileSystem)fs;
     // Make sure dfs is not in safe mode
-    while (dfs.setSafeMode(FSConstants.SafeModeAction.SAFEMODE_GET)) {
+    while (isInSafeMode(fs)) {
       LOG.info("Waiting for dfs to exit safe mode...");
       try {
         Thread.sleep(wait);