hbase-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From te...@apache.org
Subject svn commit: r1458546 - in /hbase/trunk/hbase-server/src: main/java/org/apache/hadoop/hbase/regionserver/wal/ main/java/org/apache/hadoop/hbase/util/ test/java/org/apache/hadoop/hbase/regionserver/wal/
Date Tue, 19 Mar 2013 22:17:53 GMT
Author: tedyu
Date: Tue Mar 19 22:17:53 2013
New Revision: 1458546

URL: http://svn.apache.org/r1458546
Log:
HBASE-7878 recoverFileLease does not check return value of recoverLease (Ted Yu)


Modified:
    hbase/trunk/hbase-server/src/main/java/org/apache/hadoop/hbase/regionserver/wal/HLogSplitter.java
    hbase/trunk/hbase-server/src/main/java/org/apache/hadoop/hbase/util/FSHDFSUtils.java
    hbase/trunk/hbase-server/src/test/java/org/apache/hadoop/hbase/regionserver/wal/TestHLog.java
    hbase/trunk/hbase-server/src/test/java/org/apache/hadoop/hbase/regionserver/wal/TestHLogSplit.java

Modified: hbase/trunk/hbase-server/src/main/java/org/apache/hadoop/hbase/regionserver/wal/HLogSplitter.java
URL: http://svn.apache.org/viewvc/hbase/trunk/hbase-server/src/main/java/org/apache/hadoop/hbase/regionserver/wal/HLogSplitter.java?rev=1458546&r1=1458545&r2=1458546&view=diff
==============================================================================
--- hbase/trunk/hbase-server/src/main/java/org/apache/hadoop/hbase/regionserver/wal/HLogSplitter.java
(original)
+++ hbase/trunk/hbase-server/src/main/java/org/apache/hadoop/hbase/regionserver/wal/HLogSplitter.java
Tue Mar 19 22:17:53 2013
@@ -26,6 +26,7 @@ import java.lang.reflect.InvocationTarge
 import java.text.ParseException;
 import java.util.ArrayList;
 import java.util.Collections;
+import java.util.HashSet;
 import java.util.LinkedList;
 import java.util.List;
 import java.util.Map;
@@ -322,10 +323,21 @@ public class HLogSplitter {
       }
       status.setStatus("Log splits complete. Checking for orphaned logs.");
 
-      if (fs.listStatus(srcDir).length > processedLogs.size()
+      FileStatus[] currFiles = fs.listStatus(srcDir);
+      if (currFiles.length > processedLogs.size()
           + corruptedLogs.size()) {
+        Set<Path> fileSet = new HashSet<Path>();
+        for (FileStatus fstat : currFiles) {
+          fileSet.add(fstat.getPath());
+        }
+        for (Path p : processedLogs) {
+          fileSet.remove(p);
+        }
+        for (Path p : corruptedLogs) {
+          fileSet.remove(p);
+        }        
         throw new OrphanHLogAfterSplitException(
-            "Discovered orphan hlog after split. Maybe the "
+          "Discovered orphan hlog after split. " + fileSet.iterator().next() + " Maybe the
"
             + "HRegionServer was not dead when we started");
       }
     } finally {

Modified: hbase/trunk/hbase-server/src/main/java/org/apache/hadoop/hbase/util/FSHDFSUtils.java
URL: http://svn.apache.org/viewvc/hbase/trunk/hbase-server/src/main/java/org/apache/hadoop/hbase/util/FSHDFSUtils.java?rev=1458546&r1=1458545&r2=1458546&view=diff
==============================================================================
--- hbase/trunk/hbase-server/src/main/java/org/apache/hadoop/hbase/util/FSHDFSUtils.java (original)
+++ hbase/trunk/hbase-server/src/main/java/org/apache/hadoop/hbase/util/FSHDFSUtils.java Tue
Mar 19 22:17:53 2013
@@ -54,6 +54,8 @@ public class FSHDFSUtils extends FSUtils
    * in o.a.h.hdfs.protocol.HdfsConstants cause of HDFS-1620.
    */
   public static final long LEASE_SOFTLIMIT_PERIOD = 60 * 1000;
+  
+  public static final String TEST_TRIGGER_DFS_APPEND = "hbase.test.trigger.dfs.append";
 
   @Override
   public void recoverFileLease(final FileSystem fs, final Path p, Configuration conf)
@@ -72,22 +74,37 @@ public class FSHDFSUtils extends FSUtils
 
     // Trying recovery
     boolean recovered = false;
+    long recoveryTimeout = conf.getInt("hbase.lease.recovery.timeout", 300000);
+    // conf parameter passed from unit test, indicating whether fs.append() should be triggered
+    boolean triggerAppend = conf.getBoolean(TEST_TRIGGER_DFS_APPEND, false);
+    Exception ex = null;
     while (!recovered) {
       try {
         try {
           DistributedFileSystem dfs = (DistributedFileSystem) fs;
-          DistributedFileSystem.class.getMethod("recoverLease", new Class[] { Path.class
}).invoke(
-              dfs, p);
-        } catch (InvocationTargetException ite) {
-          // function was properly called, but threw it's own exception
-          throw (IOException) ite.getCause();
+          if (triggerAppend) throw new IOException();
+          try {
+            recovered = (Boolean) DistributedFileSystem.class.getMethod(
+              "recoverLease", new Class[] { Path.class }).invoke(dfs, p);
+            if (!recovered) LOG.debug("recoverLease returned false");
+          } catch (InvocationTargetException ite) {
+            // function was properly called, but threw it's own exception
+            throw (IOException) ite.getCause();
+          }
         } catch (Exception e) {
           LOG.debug("Failed fs.recoverLease invocation, " + e.toString() +
-            ", trying fs.append instead");
+              ", trying fs.append instead");
+          ex = e;
+        }
+        if (ex != null || System.currentTimeMillis() - startWaiting > recoveryTimeout)
{
+          LOG.debug("trying fs.append for " + p + " with " + ex);
+          ex = null; // assume the following append() call would succeed
           FSDataOutputStream out = fs.append(p);
           out.close();
+          recovered = true;
+          LOG.debug("fs.append passed");
         }
-        recovered = true;
+        if (recovered) break;
       } catch (IOException e) {
         e = RemoteExceptionHandler.checkIOException(e);
         if (e instanceof AlreadyBeingCreatedException) {
@@ -111,9 +128,9 @@ public class FSHDFSUtils extends FSUtils
       }
       try {
         Thread.sleep(1000);
-      } catch (InterruptedException ex) {
+      } catch (InterruptedException ie) {
         InterruptedIOException iioe = new InterruptedIOException();
-        iioe.initCause(ex);
+        iioe.initCause(ie);
         throw iioe;
       }
     }

Modified: hbase/trunk/hbase-server/src/test/java/org/apache/hadoop/hbase/regionserver/wal/TestHLog.java
URL: http://svn.apache.org/viewvc/hbase/trunk/hbase-server/src/test/java/org/apache/hadoop/hbase/regionserver/wal/TestHLog.java?rev=1458546&r1=1458545&r2=1458546&view=diff
==============================================================================
--- hbase/trunk/hbase-server/src/test/java/org/apache/hadoop/hbase/regionserver/wal/TestHLog.java
(original)
+++ hbase/trunk/hbase-server/src/test/java/org/apache/hadoop/hbase/regionserver/wal/TestHLog.java
Tue Mar 19 22:17:53 2013
@@ -39,6 +39,7 @@ import org.apache.hadoop.hbase.*;
 import org.apache.hadoop.hbase.regionserver.wal.HLog.Reader;
 import org.apache.hadoop.hbase.regionserver.wal.HLogUtil;
 import org.apache.hadoop.hbase.util.Bytes;
+import org.apache.hadoop.hbase.util.FSHDFSUtils;
 import org.apache.hadoop.hbase.util.FSUtils;
 import org.apache.hadoop.hbase.Coprocessor;
 import org.apache.hadoop.hbase.coprocessor.CoprocessorHost;
@@ -49,7 +50,6 @@ import org.apache.hadoop.hdfs.MiniDFSClu
 import org.apache.hadoop.hdfs.protocol.FSConstants;
 import org.apache.hadoop.hdfs.server.datanode.DataNode;
 import org.apache.hadoop.hdfs.server.namenode.LeaseManager;
-import org.apache.hadoop.io.SequenceFile;
 import org.apache.log4j.Level;
 import org.junit.After;
 import org.junit.AfterClass;
@@ -79,7 +79,7 @@ public class TestHLog  {
   private final static HBaseTestingUtility TEST_UTIL = new HBaseTestingUtility();
   private static Path hbaseDir;
   private static Path oldLogDir;
-
+  
   @Before
   public void setUp() throws Exception {
 
@@ -99,6 +99,7 @@ public class TestHLog  {
     // Make block sizes small.
     TEST_UTIL.getConfiguration().setInt("dfs.blocksize", 1024 * 1024);
     // needed for testAppendClose()
+    TEST_UTIL.getConfiguration().setBoolean("dfs.support.broken.append", true);
     TEST_UTIL.getConfiguration().setBoolean("dfs.support.append", true);
     // quicker heartbeat interval for faster DN death notification
     TEST_UTIL.getConfiguration().setInt("heartbeat.recheck.interval", 5000);
@@ -370,18 +371,30 @@ public class TestHLog  {
     }
   }
   
-  // For this test to pass, requires:
-  // 1. HDFS-200 (append support)
-  // 2. HDFS-988 (SafeMode should freeze file operations
-  //              [FSNamesystem.nextGenerationStampForBlock])
-  // 3. HDFS-142 (on restart, maintain pendingCreates)
+  /*
+   * We pass different values to recoverFileLease() so that different code paths are covered
+   * 
+   * For this test to pass, requires:
+   * 1. HDFS-200 (append support)
+   * 2. HDFS-988 (SafeMode should freeze file operations
+   *              [FSNamesystem.nextGenerationStampForBlock])
+   * 3. HDFS-142 (on restart, maintain pendingCreates)
+   */
   @Test
   public void testAppendClose() throws Exception {
+    testAppendClose(true);
+    testAppendClose(false);
+  }
+  
+  /*
+   * @param triggerDirectAppend whether to trigger direct call of fs.append()
+   */
+  public void testAppendClose(final boolean triggerDirectAppend) throws Exception {
     byte [] tableName = Bytes.toBytes(getName());
     HRegionInfo regioninfo = new HRegionInfo(tableName,
              HConstants.EMPTY_START_ROW, HConstants.EMPTY_END_ROW, false);
 
-    HLog wal = HLogFactory.createHLog(fs, dir, "hlogdir", 
+    HLog wal = HLogFactory.createHLog(fs, dir, "hlogdir" + triggerDirectAppend, 
         "hlogdir_archive", conf);
     final int total = 20;
 
@@ -456,6 +469,7 @@ public class TestHLog  {
       public Exception exception = null;
       public void run() {
           try {
+            rlConf.setBoolean(FSHDFSUtils.TEST_TRIGGER_DFS_APPEND, triggerDirectAppend);
             FSUtils.getInstance(fs, rlConf)
               .recoverFileLease(recoveredFs, walPath, rlConf);
           } catch (IOException e) {

Modified: hbase/trunk/hbase-server/src/test/java/org/apache/hadoop/hbase/regionserver/wal/TestHLogSplit.java
URL: http://svn.apache.org/viewvc/hbase/trunk/hbase-server/src/test/java/org/apache/hadoop/hbase/regionserver/wal/TestHLogSplit.java?rev=1458546&r1=1458545&r2=1458546&view=diff
==============================================================================
--- hbase/trunk/hbase-server/src/test/java/org/apache/hadoop/hbase/regionserver/wal/TestHLogSplit.java
(original)
+++ hbase/trunk/hbase-server/src/test/java/org/apache/hadoop/hbase/regionserver/wal/TestHLogSplit.java
Tue Mar 19 22:17:53 2013
@@ -137,6 +137,8 @@ public class TestHLogSplit {
     FSUtils.setRootDir(TEST_UTIL.getConfiguration(), HBASEDIR);
     TEST_UTIL.getConfiguration().setClass("hbase.regionserver.hlog.writer.impl",
       InstrumentedSequenceFileLogWriter.class, HLog.Writer.class);
+    TEST_UTIL.getConfiguration().setBoolean("dfs.support.broken.append", true);
+    TEST_UTIL.getConfiguration().setBoolean("dfs.support.append", true);    
     // This is how you turn off shortcircuit read currently.  TODO: Fix.  Should read config.
     System.setProperty("hbase.tests.use.shortcircuit.reads", "false");
     // Create fake maping user to group and set it to the conf.
@@ -715,14 +717,17 @@ public class TestHLogSplit {
     fs.initialize(fs.getUri(), conf);
     Thread zombie = new ZombieNewLogWriterRegionServer(stop);
 
+    List<Path> splits = null;
     try {
       zombie.start();
       try {
         HLogSplitter logSplitter = HLogSplitter.createLogSplitter(conf,
             HBASEDIR, HLOGDIR, OLDLOGDIR, fs);
-        logSplitter.splitLog();
+        splits = logSplitter.splitLog();
       } catch (IOException ex) {/* expected */}
-      int logFilesNumber = fs.listStatus(HLOGDIR).length;
+      FileStatus[] files = fs.listStatus(HLOGDIR);
+      if (files == null) fail("no files in " + HLOGDIR + " with splits " + splits);
+      int logFilesNumber = files.length;
 
       assertEquals("Log files should not be archived if there's an extra file after split",
               NUM_WRITERS + 1, logFilesNumber);



Mime
View raw message