Java Code Examples for org.apache.hadoop.hdfs.server.datanode.ReplicaInfo#getGenerationStamp()

/**
 * Remove the replica's meta information from the map that matches
 * the input block's id and generation stamp
 * @param bpid block pool id
 * @param block block with its id as the key
 * @return the removed replica's meta information
 * @throws IllegalArgumentException if the input block is null
 */
ReplicaInfo remove(String bpid, Block block) {
  checkBlockPool(bpid);
  checkBlock(block);
  synchronized(mutex) {
    Map<Long, ReplicaInfo> m = map.get(bpid);
    if (m != null) {
      Long key = Long.valueOf(block.getBlockId());
      ReplicaInfo replicaInfo = m.get(key);
      if (replicaInfo != null &&
          block.getGenerationStamp() == replicaInfo.getGenerationStamp()) {
        return m.remove(key);
      } 
    }
  }
  
  return null;
}
 

/**
 * Bump a replica's generation stamp to a new one.
 * Its on-disk meta file name is renamed to be the new one too.
 * 
 * @param replicaInfo a replica
 * @param newGS new generation stamp
 * @throws IOException if rename fails
 */
private void bumpReplicaGS(ReplicaInfo replicaInfo, 
    long newGS) throws IOException { 
  long oldGS = replicaInfo.getGenerationStamp();
  File oldmeta = replicaInfo.getMetaFile();
  replicaInfo.setGenerationStamp(newGS);
  File newmeta = replicaInfo.getMetaFile();

  // rename meta file to new GS
  if (LOG.isDebugEnabled()) {
    LOG.debug("Renaming " + oldmeta + " to " + newmeta);
  }
  try {
    NativeIO.renameTo(oldmeta, newmeta);
  } catch (IOException e) {
    replicaInfo.setGenerationStamp(oldGS); // restore old GS
    throw new IOException("Block " + replicaInfo + " reopen failed. " +
                          " Unable to move meta file  " + oldmeta +
                          " to " + newmeta, e);
  }
}
 

/**
 * Remove the replica's meta information from the map that matches
 * the input block's id and generation stamp
 * @param bpid block pool id
 * @param block block with its id as the key
 * @return the removed replica's meta information
 * @throws IllegalArgumentException if the input block is null
 */
ReplicaInfo remove(String bpid, Block block) {
  checkBlockPool(bpid);
  checkBlock(block);
  synchronized(mutex) {
    Map<Long, ReplicaInfo> m = map.get(bpid);
    if (m != null) {
      Long key = Long.valueOf(block.getBlockId());
      ReplicaInfo replicaInfo = m.get(key);
      if (replicaInfo != null &&
          block.getGenerationStamp() == replicaInfo.getGenerationStamp()) {
        return m.remove(key);
      } 
    }
  }
  
  return null;
}
 

/**
 * Bump a replica's generation stamp to a new one.
 * Its on-disk meta file name is renamed to be the new one too.
 * 
 * @param replicaInfo a replica
 * @param newGS new generation stamp
 * @throws IOException if rename fails
 */
private void bumpReplicaGS(ReplicaInfo replicaInfo, 
    long newGS) throws IOException { 
  long oldGS = replicaInfo.getGenerationStamp();
  File oldmeta = replicaInfo.getMetaFile();
  replicaInfo.setGenerationStamp(newGS);
  File newmeta = replicaInfo.getMetaFile();

  // rename meta file to new GS
  if (LOG.isDebugEnabled()) {
    LOG.debug("Renaming " + oldmeta + " to " + newmeta);
  }
  try {
    NativeIO.renameTo(oldmeta, newmeta);
  } catch (IOException e) {
    replicaInfo.setGenerationStamp(oldGS); // restore old GS
    throw new IOException("Block " + replicaInfo + " reopen failed. " +
                          " Unable to move meta file  " + oldmeta +
                          " to " + newmeta, e);
  }
}
 

/**
 * Get the meta information of the replica that matches both block id 
 * and generation stamp
 * @param bpid block pool id
 * @param block block with its id as the key
 * @return the replica's meta information
 * @throws IllegalArgumentException if the input block or block pool is null
 */
ReplicaInfo get(String bpid, Block block) {
  checkBlockPool(bpid);
  checkBlock(block);
  ReplicaInfo replicaInfo = get(bpid, block.getBlockId());
  if (replicaInfo != null && 
      block.getGenerationStamp() == replicaInfo.getGenerationStamp()) {
    return replicaInfo;
  }
  return null;
}
 

@VisibleForTesting
static ReplicaInfo selectReplicaToDelete(final ReplicaInfo replica1,
    final ReplicaInfo replica2) {
  ReplicaInfo replicaToKeep;
  ReplicaInfo replicaToDelete;

  // it's the same block so don't ever delete it, even if GS or size
  // differs.  caller should keep the one it just discovered on disk
  if (replica1.getBlockFile().equals(replica2.getBlockFile())) {
    return null;
  }
  if (replica1.getGenerationStamp() != replica2.getGenerationStamp()) {
    replicaToKeep = replica1.getGenerationStamp() > replica2.getGenerationStamp()
        ? replica1 : replica2;
  } else if (replica1.getNumBytes() != replica2.getNumBytes()) {
    replicaToKeep = replica1.getNumBytes() > replica2.getNumBytes() ?
        replica1 : replica2;
  } else if (replica1.getVolume().isTransientStorage() &&
             !replica2.getVolume().isTransientStorage()) {
    replicaToKeep = replica2;
  } else {
    replicaToKeep = replica1;
  }

  replicaToDelete = (replicaToKeep == replica1) ? replica2 : replica1;

  if (LOG.isDebugEnabled()) {
    LOG.debug("resolveDuplicateReplicas decide to keep " + replicaToKeep
        + ".  Will try to delete " + replicaToDelete);
  }
  return replicaToDelete;
}
 

@VisibleForTesting
static ReplicaInfo selectReplicaToDelete(final ReplicaInfo replica1,
                                         final ReplicaInfo replica2) {
  ReplicaInfo replicaToKeep;
  ReplicaInfo replicaToDelete;

  // it's the same block so don't ever delete it, even if GS or size
  // differs.  caller should keep the one it just discovered on disk
  if (replica1.getBlockURI().equals(replica2.getBlockURI())) {
    return null;
  }
  if (replica1.getGenerationStamp() != replica2.getGenerationStamp()) {
    replicaToKeep = replica1.getGenerationStamp() > replica2.getGenerationStamp()
        ? replica1 : replica2;
  } else if (replica1.getNumBytes() != replica2.getNumBytes()) {
    replicaToKeep = replica1.getNumBytes() > replica2.getNumBytes() ?
        replica1 : replica2;
  } else if (replica1.getVolume().isTransientStorage() &&
      !replica2.getVolume().isTransientStorage()) {
    replicaToKeep = replica2;
  } else {
    replicaToKeep = replica1;
  }

  replicaToDelete = (replicaToKeep == replica1) ? replica2 : replica1;

  if (LOG.isDebugEnabled()) {
    LOG.debug("resolveDuplicateReplicas decide to keep {}.  Will try to delete {}", replicaToKeep, replicaToDelete);
  }
  return replicaToDelete;
}
 

@VisibleForTesting
static ReplicaInfo selectReplicaToDelete(final ReplicaInfo replica1,
    final ReplicaInfo replica2) {
  ReplicaInfo replicaToKeep;
  ReplicaInfo replicaToDelete;

  // it's the same block so don't ever delete it, even if GS or size
  // differs.  caller should keep the one it just discovered on disk
  if (replica1.getBlockFile().equals(replica2.getBlockFile())) {
    return null;
  }
  if (replica1.getGenerationStamp() != replica2.getGenerationStamp()) {
    replicaToKeep = replica1.getGenerationStamp() > replica2.getGenerationStamp()
        ? replica1 : replica2;
  } else if (replica1.getNumBytes() != replica2.getNumBytes()) {
    replicaToKeep = replica1.getNumBytes() > replica2.getNumBytes() ?
        replica1 : replica2;
  } else if (replica1.getVolume().isTransientStorage() &&
             !replica2.getVolume().isTransientStorage()) {
    replicaToKeep = replica2;
  } else {
    replicaToKeep = replica1;
  }

  replicaToDelete = (replicaToKeep == replica1) ? replica2 : replica1;

  if (LOG.isDebugEnabled()) {
    LOG.debug("resolveDuplicateReplicas decide to keep " + replicaToKeep
        + ".  Will try to delete " + replicaToDelete);
  }
  return replicaToDelete;
}
 

/**
 * Get the meta information of the replica that matches both block id 
 * and generation stamp
 * @param bpid block pool id
 * @param block block with its id as the key
 * @return the replica's meta information
 * @throws IllegalArgumentException if the input block or block pool is null
 */
ReplicaInfo get(String bpid, Block block) {
  checkBlockPool(bpid);
  checkBlock(block);
  ReplicaInfo replicaInfo = get(bpid, block.getBlockId());
  if (replicaInfo != null && 
      block.getGenerationStamp() == replicaInfo.getGenerationStamp()) {
    return replicaInfo;
  }
  return null;
}
 

/**
 * Move block files from one storage to another storage.
 * @return Returns the Old replicaInfo
 * @throws IOException
 */
@Override
public ReplicaInfo moveBlockAcrossStorage(ExtendedBlock block,
    StorageType targetStorageType) throws IOException {
  ReplicaInfo replicaInfo = getReplicaInfo(block);
  if (replicaInfo.getState() != ReplicaState.FINALIZED) {
    throw new ReplicaNotFoundException(
        ReplicaNotFoundException.UNFINALIZED_REPLICA + block);
  }
  if (replicaInfo.getNumBytes() != block.getNumBytes()) {
    throw new IOException("Corrupted replica " + replicaInfo
        + " with a length of " + replicaInfo.getNumBytes()
        + " expected length is " + block.getNumBytes());
  }
  if (replicaInfo.getVolume().getStorageType() == targetStorageType) {
    throw new ReplicaAlreadyExistsException("Replica " + replicaInfo
        + " already exists on storage " + targetStorageType);
  }

  if (replicaInfo.isOnTransientStorage()) {
    // Block movement from RAM_DISK will be done by LazyPersist mechanism
    throw new IOException("Replica " + replicaInfo
        + " cannot be moved from storageType : "
        + replicaInfo.getVolume().getStorageType());
  }

  try (FsVolumeReference volumeRef = volumes.getNextVolume(
      targetStorageType, block.getNumBytes())) {
    File oldBlockFile = replicaInfo.getBlockFile();
    File oldMetaFile = replicaInfo.getMetaFile();
    FsVolumeImpl targetVolume = (FsVolumeImpl) volumeRef.getVolume();
    // Copy files to temp dir first
    File[] blockFiles = copyBlockFiles(block.getBlockId(),
        block.getGenerationStamp(), oldMetaFile, oldBlockFile,
        targetVolume.getTmpDir(block.getBlockPoolId()),
        replicaInfo.isOnTransientStorage());

    ReplicaInfo newReplicaInfo = new ReplicaInPipeline(
        replicaInfo.getBlockId(), replicaInfo.getGenerationStamp(),
        targetVolume, blockFiles[0].getParentFile(), 0);
    newReplicaInfo.setNumBytes(blockFiles[1].length());
    // Finalize the copied files
    newReplicaInfo = finalizeReplica(block.getBlockPoolId(), newReplicaInfo);

    removeOldReplica(replicaInfo, newReplicaInfo, oldBlockFile, oldMetaFile,
        oldBlockFile.length(), oldMetaFile.length(), block.getBlockPoolId());
  }

  // Replace the old block if any to reschedule the scanning.
  return replicaInfo;
}
 

/** static version of {@link #initReplicaRecovery(RecoveringBlock)}. */
static ReplicaRecoveryInfo initReplicaRecovery(String bpid, ReplicaMap map,
    Block block, long recoveryId, long xceiverStopTimeout) throws IOException {
  final ReplicaInfo replica = map.get(bpid, block.getBlockId());
  LOG.info("initReplicaRecovery: " + block + ", recoveryId=" + recoveryId
      + ", replica=" + replica);

  //check replica
  if (replica == null) {
    return null;
  }

  //stop writer if there is any
  if (replica instanceof ReplicaInPipeline) {
    final ReplicaInPipeline rip = (ReplicaInPipeline)replica;
    rip.stopWriter(xceiverStopTimeout);

    //check replica bytes on disk.
    if (rip.getBytesOnDisk() < rip.getVisibleLength()) {
      throw new IOException("THIS IS NOT SUPPOSED TO HAPPEN:"
          + " getBytesOnDisk() < getVisibleLength(), rip=" + rip);
    }

    //check the replica's files
    checkReplicaFiles(rip);
  }

  //check generation stamp
  if (replica.getGenerationStamp() < block.getGenerationStamp()) {
    throw new IOException(
        "replica.getGenerationStamp() < block.getGenerationStamp(), block="
        + block + ", replica=" + replica);
  }

  //check recovery id
  if (replica.getGenerationStamp() >= recoveryId) {
    throw new IOException("THIS IS NOT SUPPOSED TO HAPPEN:"
        + " replica.getGenerationStamp() >= recoveryId = " + recoveryId
        + ", block=" + block + ", replica=" + replica);
  }

  //check RUR
  final ReplicaUnderRecovery rur;
  if (replica.getState() == ReplicaState.RUR) {
    rur = (ReplicaUnderRecovery)replica;
    if (rur.getRecoveryID() >= recoveryId) {
      throw new RecoveryInProgressException(
          "rur.getRecoveryID() >= recoveryId = " + recoveryId
          + ", block=" + block + ", rur=" + rur);
    }
    final long oldRecoveryID = rur.getRecoveryID();
    rur.setRecoveryID(recoveryId);
    LOG.info("initReplicaRecovery: update recovery id for " + block
        + " from " + oldRecoveryID + " to " + recoveryId);
  }
  else {
    rur = new ReplicaUnderRecovery(replica, recoveryId);
    map.add(bpid, rur);
    LOG.info("initReplicaRecovery: changing replica state for "
        + block + " from " + replica.getState()
        + " to " + rur.getState());
  }
  return rur.createInfo();
}
 

/**
 * Asynchronously attempts to cache a single block via {@link FsDatasetCache}.
 */
private void cacheBlock(String bpid, long blockId) {
  FsVolumeImpl volume;
  String blockFileName;
  long length, genstamp;
  Executor volumeExecutor;

  synchronized (this) {
    ReplicaInfo info = volumeMap.get(bpid, blockId);
    boolean success = false;
    try {
      if (info == null) {
        LOG.warn("Failed to cache block with id " + blockId + ", pool " +
            bpid + ": ReplicaInfo not found.");
        return;
      }
      if (info.getState() != ReplicaState.FINALIZED) {
        LOG.warn("Failed to cache block with id " + blockId + ", pool " +
            bpid + ": replica is not finalized; it is in state " +
            info.getState());
        return;
      }
      try {
        volume = (FsVolumeImpl)info.getVolume();
        if (volume == null) {
          LOG.warn("Failed to cache block with id " + blockId + ", pool " +
              bpid + ": volume not found.");
          return;
        }
      } catch (ClassCastException e) {
        LOG.warn("Failed to cache block with id " + blockId +
            ": volume was not an instance of FsVolumeImpl.");
        return;
      }
      if (volume.isTransientStorage()) {
        LOG.warn("Caching not supported on block with id " + blockId +
            " since the volume is backed by RAM.");
        return;
      }
      success = true;
    } finally {
      if (!success) {
        cacheManager.numBlocksFailedToCache.incrementAndGet();
      }
    }
    blockFileName = info.getBlockFile().getAbsolutePath();
    length = info.getVisibleLength();
    genstamp = info.getGenerationStamp();
    volumeExecutor = volume.getCacheExecutor();
  }
  cacheManager.cacheBlock(blockId, bpid, 
      blockFileName, length, genstamp, volumeExecutor);
}
 

/**
 * We're informed that a block is no longer valid.  We
 * could lazily garbage-collect the block, but why bother?
 * just get rid of it.
 */
@Override // FsDatasetSpi
public void invalidate(String bpid, Block invalidBlks[]) throws IOException {
  final List<String> errors = new ArrayList<String>();
  for (int i = 0; i < invalidBlks.length; i++) {
    final File f;
    final FsVolumeImpl v;
    synchronized (this) {
      final ReplicaInfo info = volumeMap.get(bpid, invalidBlks[i]);
      if (info == null) {
        // It is okay if the block is not found -- it may be deleted earlier.
        LOG.info("Failed to delete replica " + invalidBlks[i]
            + ": ReplicaInfo not found.");
        continue;
      }
      if (info.getGenerationStamp() != invalidBlks[i].getGenerationStamp()) {
        errors.add("Failed to delete replica " + invalidBlks[i]
            + ": GenerationStamp not matched, info=" + info);
        continue;
      }
      f = info.getBlockFile();
      v = (FsVolumeImpl)info.getVolume();
      if (v == null) {
        errors.add("Failed to delete replica " + invalidBlks[i]
            +  ". No volume for this replica, file=" + f);
        continue;
      }
      File parent = f.getParentFile();
      if (parent == null) {
        errors.add("Failed to delete replica " + invalidBlks[i]
            +  ". Parent not found for file " + f);
        continue;
      }
      ReplicaInfo removing = volumeMap.remove(bpid, invalidBlks[i]);
      addDeletingBlock(bpid, removing.getBlockId());
      if (LOG.isDebugEnabled()) {
        LOG.debug("Block file " + removing.getBlockFile().getName()
            + " is to be deleted");
      }
    }

    if (v.isTransientStorage()) {
      RamDiskReplica replicaInfo =
        ramDiskReplicaTracker.getReplica(bpid, invalidBlks[i].getBlockId());
      if (replicaInfo != null) {
        if (!replicaInfo.getIsPersisted()) {
          datanode.getMetrics().incrRamDiskBlocksDeletedBeforeLazyPersisted();
        }
        ramDiskReplicaTracker.discardReplica(replicaInfo.getBlockPoolId(),
          replicaInfo.getBlockId(), true);
      }
    }

    // If a DFSClient has the replica in its cache of short-circuit file
    // descriptors (and the client is using ShortCircuitShm), invalidate it.
    datanode.getShortCircuitRegistry().processBlockInvalidation(
              new ExtendedBlockId(invalidBlks[i].getBlockId(), bpid));

    // If the block is cached, start uncaching it.
    cacheManager.uncacheBlock(bpid, invalidBlks[i].getBlockId());

    // Delete the block asynchronously to make sure we can do it fast enough.
    // It's ok to unlink the block file before the uncache operation
    // finishes.
    try {
      asyncDiskService.deleteAsync(v.obtainReference(), f,
          FsDatasetUtil.getMetaFile(f, invalidBlks[i].getGenerationStamp()),
          new ExtendedBlock(bpid, invalidBlks[i]),
          dataStorage.getTrashDirectoryForBlockFile(bpid, f));
    } catch (ClosedChannelException e) {
      LOG.warn("Volume " + v + " is closed, ignore the deletion task for " +
          "block " + invalidBlks[i]);
    }
  }
  if (!errors.isEmpty()) {
    StringBuilder b = new StringBuilder("Failed to delete ")
      .append(errors.size()).append(" (out of ").append(invalidBlks.length)
      .append(") replica(s):");
    for(int i = 0; i < errors.size(); i++) {
      b.append("\n").append(i).append(") ").append(errors.get(i));
    }
    throw new IOException(b.toString());
  }
}
 

@Override // FsDatasetSpi
public ReplicaHandler createTemporary(
    StorageType storageType, ExtendedBlock b) throws IOException {
  long startTimeMs = Time.monotonicNow();
  long writerStopTimeoutMs = datanode.getDnConf().getXceiverStopTimeout();
  ReplicaInfo lastFoundReplicaInfo = null;
  do {
    synchronized (this) {
      ReplicaInfo currentReplicaInfo =
          volumeMap.get(b.getBlockPoolId(), b.getBlockId());
      if (currentReplicaInfo == lastFoundReplicaInfo) {
        if (lastFoundReplicaInfo != null) {
          invalidate(b.getBlockPoolId(), new Block[] { lastFoundReplicaInfo });
        }
        FsVolumeReference ref =
            volumes.getNextVolume(storageType, b.getNumBytes());
        FsVolumeImpl v = (FsVolumeImpl) ref.getVolume();
        // create a temporary file to hold block in the designated volume
        File f;
        try {
          f = v.createTmpFile(b.getBlockPoolId(), b.getLocalBlock());
        } catch (IOException e) {
          IOUtils.cleanup(null, ref);
          throw e;
        }
        ReplicaInPipeline newReplicaInfo =
            new ReplicaInPipeline(b.getBlockId(), b.getGenerationStamp(), v,
                f.getParentFile(), 0);
        volumeMap.add(b.getBlockPoolId(), newReplicaInfo);
        return new ReplicaHandler(newReplicaInfo, ref);
      } else {
        if (!(currentReplicaInfo.getGenerationStamp() < b
            .getGenerationStamp() && currentReplicaInfo instanceof ReplicaInPipeline)) {
          throw new ReplicaAlreadyExistsException("Block " + b
              + " already exists in state " + currentReplicaInfo.getState()
              + " and thus cannot be created.");
        }
        lastFoundReplicaInfo = currentReplicaInfo;
      }
    }

    // Hang too long, just bail out. This is not supposed to happen.
    long writerStopMs = Time.monotonicNow() - startTimeMs;
    if (writerStopMs > writerStopTimeoutMs) {
      LOG.warn("Unable to stop existing writer for block " + b + " after " 
          + writerStopMs + " miniseconds.");
      throw new IOException("Unable to stop existing writer for block " + b
          + " after " + writerStopMs + " miniseconds.");
    }

    // Stop the previous writer
    ((ReplicaInPipeline) lastFoundReplicaInfo)
        .stopWriter(writerStopTimeoutMs);
  } while (true);
}
 

private ReplicaInfo recoverCheck(ExtendedBlock b, long newGS, 
    long expectedBlockLen) throws IOException {
  ReplicaInfo replicaInfo = getReplicaInfo(b.getBlockPoolId(), b.getBlockId());
  
  // check state
  if (replicaInfo.getState() != ReplicaState.FINALIZED &&
      replicaInfo.getState() != ReplicaState.RBW) {
    throw new ReplicaNotFoundException(
        ReplicaNotFoundException.UNFINALIZED_AND_NONRBW_REPLICA + replicaInfo);
  }

  // check generation stamp
  long replicaGenerationStamp = replicaInfo.getGenerationStamp();
  if (replicaGenerationStamp < b.getGenerationStamp() ||
      replicaGenerationStamp > newGS) {
    throw new ReplicaNotFoundException(
        ReplicaNotFoundException.UNEXPECTED_GS_REPLICA + replicaGenerationStamp
        + ". Expected GS range is [" + b.getGenerationStamp() + ", " + 
        newGS + "].");
  }
  
  // stop the previous writer before check a replica's length
  long replicaLen = replicaInfo.getNumBytes();
  if (replicaInfo.getState() == ReplicaState.RBW) {
    ReplicaBeingWritten rbw = (ReplicaBeingWritten)replicaInfo;
    // kill the previous writer
    rbw.stopWriter(datanode.getDnConf().getXceiverStopTimeout());
    rbw.setWriter(Thread.currentThread());
    // check length: bytesRcvd, bytesOnDisk, and bytesAcked should be the same
    if (replicaLen != rbw.getBytesOnDisk() 
        || replicaLen != rbw.getBytesAcked()) {
      throw new ReplicaAlreadyExistsException("RBW replica " + replicaInfo + 
          "bytesRcvd(" + rbw.getNumBytes() + "), bytesOnDisk(" + 
          rbw.getBytesOnDisk() + "), and bytesAcked(" + rbw.getBytesAcked() +
          ") are not the same.");
    }
  }
  
  // check block length
  if (replicaLen != expectedBlockLen) {
    throw new IOException("Corrupted replica " + replicaInfo + 
        " with a length of " + replicaLen + 
        " expected length is " + expectedBlockLen);
  }
  
  return replicaInfo;
}
 

/**
 * Move block files from one storage to another storage.
 * @return Returns the Old replicaInfo
 * @throws IOException
 */
@Override
public ReplicaInfo moveBlockAcrossStorage(ExtendedBlock block,
    StorageType targetStorageType) throws IOException {
  ReplicaInfo replicaInfo = getReplicaInfo(block);
  if (replicaInfo.getState() != ReplicaState.FINALIZED) {
    throw new ReplicaNotFoundException(
        ReplicaNotFoundException.UNFINALIZED_REPLICA + block);
  }
  if (replicaInfo.getNumBytes() != block.getNumBytes()) {
    throw new IOException("Corrupted replica " + replicaInfo
        + " with a length of " + replicaInfo.getNumBytes()
        + " expected length is " + block.getNumBytes());
  }
  if (replicaInfo.getVolume().getStorageType() == targetStorageType) {
    throw new ReplicaAlreadyExistsException("Replica " + replicaInfo
        + " already exists on storage " + targetStorageType);
  }

  if (replicaInfo.isOnTransientStorage()) {
    // Block movement from RAM_DISK will be done by LazyPersist mechanism
    throw new IOException("Replica " + replicaInfo
        + " cannot be moved from storageType : "
        + replicaInfo.getVolume().getStorageType());
  }

  try (FsVolumeReference volumeRef = volumes.getNextVolume(
      targetStorageType, block.getNumBytes())) {
    File oldBlockFile = replicaInfo.getBlockFile();
    File oldMetaFile = replicaInfo.getMetaFile();
    FsVolumeImpl targetVolume = (FsVolumeImpl) volumeRef.getVolume();
    // Copy files to temp dir first
    File[] blockFiles = copyBlockFiles(block.getBlockId(),
        block.getGenerationStamp(), oldMetaFile, oldBlockFile,
        targetVolume.getTmpDir(block.getBlockPoolId()),
        replicaInfo.isOnTransientStorage());

    ReplicaInfo newReplicaInfo = new ReplicaInPipeline(
        replicaInfo.getBlockId(), replicaInfo.getGenerationStamp(),
        targetVolume, blockFiles[0].getParentFile(), 0);
    newReplicaInfo.setNumBytes(blockFiles[1].length());
    // Finalize the copied files
    newReplicaInfo = finalizeReplica(block.getBlockPoolId(), newReplicaInfo);

    removeOldReplica(replicaInfo, newReplicaInfo, oldBlockFile, oldMetaFile,
        oldBlockFile.length(), oldMetaFile.length(), block.getBlockPoolId());
  }

  // Replace the old block if any to reschedule the scanning.
  return replicaInfo;
}
 

/** static version of {@link #initReplicaRecovery(RecoveringBlock)}. */
static ReplicaRecoveryInfo initReplicaRecovery(String bpid, ReplicaMap map,
    Block block, long recoveryId, long xceiverStopTimeout) throws IOException {
  final ReplicaInfo replica = map.get(bpid, block.getBlockId());
  LOG.info("initReplicaRecovery: " + block + ", recoveryId=" + recoveryId
      + ", replica=" + replica);

  //check replica
  if (replica == null) {
    return null;
  }

  //stop writer if there is any
  if (replica instanceof ReplicaInPipeline) {
    final ReplicaInPipeline rip = (ReplicaInPipeline)replica;
    rip.stopWriter(xceiverStopTimeout);

    //check replica bytes on disk.
    if (rip.getBytesOnDisk() < rip.getVisibleLength()) {
      throw new IOException("THIS IS NOT SUPPOSED TO HAPPEN:"
          + " getBytesOnDisk() < getVisibleLength(), rip=" + rip);
    }

    //check the replica's files
    checkReplicaFiles(rip);
  }

  //check generation stamp
  if (replica.getGenerationStamp() < block.getGenerationStamp()) {
    throw new IOException(
        "replica.getGenerationStamp() < block.getGenerationStamp(), block="
        + block + ", replica=" + replica);
  }

  //check recovery id
  if (replica.getGenerationStamp() >= recoveryId) {
    throw new IOException("THIS IS NOT SUPPOSED TO HAPPEN:"
        + " replica.getGenerationStamp() >= recoveryId = " + recoveryId
        + ", block=" + block + ", replica=" + replica);
  }

  //check RUR
  final ReplicaUnderRecovery rur;
  if (replica.getState() == ReplicaState.RUR) {
    rur = (ReplicaUnderRecovery)replica;
    if (rur.getRecoveryID() >= recoveryId) {
      throw new RecoveryInProgressException(
          "rur.getRecoveryID() >= recoveryId = " + recoveryId
          + ", block=" + block + ", rur=" + rur);
    }
    final long oldRecoveryID = rur.getRecoveryID();
    rur.setRecoveryID(recoveryId);
    LOG.info("initReplicaRecovery: update recovery id for " + block
        + " from " + oldRecoveryID + " to " + recoveryId);
  }
  else {
    rur = new ReplicaUnderRecovery(replica, recoveryId);
    map.add(bpid, rur);
    LOG.info("initReplicaRecovery: changing replica state for "
        + block + " from " + replica.getState()
        + " to " + rur.getState());
  }
  return rur.createInfo();
}
 

/**
 * We're informed that a block is no longer valid.  We
 * could lazily garbage-collect the block, but why bother?
 * just get rid of it.
 */
@Override // FsDatasetSpi
public void invalidate(String bpid, Block invalidBlks[]) throws IOException {
  final List<String> errors = new ArrayList<String>();
  for (int i = 0; i < invalidBlks.length; i++) {
    final File f;
    final FsVolumeImpl v;
    synchronized (this) {
      final ReplicaInfo info = volumeMap.get(bpid, invalidBlks[i]);
      if (info == null) {
        // It is okay if the block is not found -- it may be deleted earlier.
        LOG.info("Failed to delete replica " + invalidBlks[i]
            + ": ReplicaInfo not found.");
        continue;
      }
      if (info.getGenerationStamp() != invalidBlks[i].getGenerationStamp()) {
        errors.add("Failed to delete replica " + invalidBlks[i]
            + ": GenerationStamp not matched, info=" + info);
        continue;
      }
      f = info.getBlockFile();
      v = (FsVolumeImpl)info.getVolume();
      if (v == null) {
        errors.add("Failed to delete replica " + invalidBlks[i]
            +  ". No volume for this replica, file=" + f);
        continue;
      }
      File parent = f.getParentFile();
      if (parent == null) {
        errors.add("Failed to delete replica " + invalidBlks[i]
            +  ". Parent not found for file " + f);
        continue;
      }
      ReplicaInfo removing = volumeMap.remove(bpid, invalidBlks[i]);
      addDeletingBlock(bpid, removing.getBlockId());
      if (LOG.isDebugEnabled()) {
        LOG.debug("Block file " + removing.getBlockFile().getName()
            + " is to be deleted");
      }
    }

    if (v.isTransientStorage()) {
      RamDiskReplica replicaInfo =
        ramDiskReplicaTracker.getReplica(bpid, invalidBlks[i].getBlockId());
      if (replicaInfo != null) {
        if (!replicaInfo.getIsPersisted()) {
          datanode.getMetrics().incrRamDiskBlocksDeletedBeforeLazyPersisted();
        }
        ramDiskReplicaTracker.discardReplica(replicaInfo.getBlockPoolId(),
          replicaInfo.getBlockId(), true);
      }
    }

    // If a DFSClient has the replica in its cache of short-circuit file
    // descriptors (and the client is using ShortCircuitShm), invalidate it.
    datanode.getShortCircuitRegistry().processBlockInvalidation(
              new ExtendedBlockId(invalidBlks[i].getBlockId(), bpid));

    // If the block is cached, start uncaching it.
    cacheManager.uncacheBlock(bpid, invalidBlks[i].getBlockId());

    // Delete the block asynchronously to make sure we can do it fast enough.
    // It's ok to unlink the block file before the uncache operation
    // finishes.
    try {
      asyncDiskService.deleteAsync(v.obtainReference(), f,
          FsDatasetUtil.getMetaFile(f, invalidBlks[i].getGenerationStamp()),
          new ExtendedBlock(bpid, invalidBlks[i]),
          dataStorage.getTrashDirectoryForBlockFile(bpid, f));
    } catch (ClosedChannelException e) {
      LOG.warn("Volume " + v + " is closed, ignore the deletion task for " +
          "block " + invalidBlks[i]);
    }
  }
  if (!errors.isEmpty()) {
    StringBuilder b = new StringBuilder("Failed to delete ")
      .append(errors.size()).append(" (out of ").append(invalidBlks.length)
      .append(") replica(s):");
    for(int i = 0; i < errors.size(); i++) {
      b.append("\n").append(i).append(") ").append(errors.get(i));
    }
    throw new IOException(b.toString());
  }
}
 

@Override // FsDatasetSpi
public ReplicaHandler createTemporary(
    StorageType storageType, ExtendedBlock b) throws IOException {
  long startTimeMs = Time.monotonicNow();
  long writerStopTimeoutMs = datanode.getDnConf().getXceiverStopTimeout();
  ReplicaInfo lastFoundReplicaInfo = null;
  do {
    synchronized (this) {
      ReplicaInfo currentReplicaInfo =
          volumeMap.get(b.getBlockPoolId(), b.getBlockId());
      if (currentReplicaInfo == lastFoundReplicaInfo) {
        if (lastFoundReplicaInfo != null) {
          invalidate(b.getBlockPoolId(), new Block[] { lastFoundReplicaInfo });
        }
        FsVolumeReference ref =
            volumes.getNextVolume(storageType, b.getNumBytes());
        FsVolumeImpl v = (FsVolumeImpl) ref.getVolume();
        // create a temporary file to hold block in the designated volume
        File f;
        try {
          f = v.createTmpFile(b.getBlockPoolId(), b.getLocalBlock());
        } catch (IOException e) {
          IOUtils.cleanup(null, ref);
          throw e;
        }
        ReplicaInPipeline newReplicaInfo =
            new ReplicaInPipeline(b.getBlockId(), b.getGenerationStamp(), v,
                f.getParentFile(), 0);
        volumeMap.add(b.getBlockPoolId(), newReplicaInfo);
        return new ReplicaHandler(newReplicaInfo, ref);
      } else {
        if (!(currentReplicaInfo.getGenerationStamp() < b
            .getGenerationStamp() && currentReplicaInfo instanceof ReplicaInPipeline)) {
          throw new ReplicaAlreadyExistsException("Block " + b
              + " already exists in state " + currentReplicaInfo.getState()
              + " and thus cannot be created.");
        }
        lastFoundReplicaInfo = currentReplicaInfo;
      }
    }

    // Hang too long, just bail out. This is not supposed to happen.
    long writerStopMs = Time.monotonicNow() - startTimeMs;
    if (writerStopMs > writerStopTimeoutMs) {
      LOG.warn("Unable to stop existing writer for block " + b + " after " 
          + writerStopMs + " miniseconds.");
      throw new IOException("Unable to stop existing writer for block " + b
          + " after " + writerStopMs + " miniseconds.");
    }

    // Stop the previous writer
    ((ReplicaInPipeline) lastFoundReplicaInfo)
        .stopWriter(writerStopTimeoutMs);
  } while (true);
}
 

private ReplicaInfo recoverCheck(ExtendedBlock b, long newGS, 
    long expectedBlockLen) throws IOException {
  ReplicaInfo replicaInfo = getReplicaInfo(b.getBlockPoolId(), b.getBlockId());
  
  // check state
  if (replicaInfo.getState() != ReplicaState.FINALIZED &&
      replicaInfo.getState() != ReplicaState.RBW) {
    throw new ReplicaNotFoundException(
        ReplicaNotFoundException.UNFINALIZED_AND_NONRBW_REPLICA + replicaInfo);
  }

  // check generation stamp
  long replicaGenerationStamp = replicaInfo.getGenerationStamp();
  if (replicaGenerationStamp < b.getGenerationStamp() ||
      replicaGenerationStamp > newGS) {
    throw new ReplicaNotFoundException(
        ReplicaNotFoundException.UNEXPECTED_GS_REPLICA + replicaGenerationStamp
        + ". Expected GS range is [" + b.getGenerationStamp() + ", " + 
        newGS + "].");
  }
  
  // stop the previous writer before check a replica's length
  long replicaLen = replicaInfo.getNumBytes();
  if (replicaInfo.getState() == ReplicaState.RBW) {
    ReplicaBeingWritten rbw = (ReplicaBeingWritten)replicaInfo;
    // kill the previous writer
    rbw.stopWriter(datanode.getDnConf().getXceiverStopTimeout());
    rbw.setWriter(Thread.currentThread());
    // check length: bytesRcvd, bytesOnDisk, and bytesAcked should be the same
    if (replicaLen != rbw.getBytesOnDisk() 
        || replicaLen != rbw.getBytesAcked()) {
      throw new ReplicaAlreadyExistsException("RBW replica " + replicaInfo + 
          "bytesRcvd(" + rbw.getNumBytes() + "), bytesOnDisk(" + 
          rbw.getBytesOnDisk() + "), and bytesAcked(" + rbw.getBytesAcked() +
          ") are not the same.");
    }
  }
  
  // check block length
  if (replicaLen != expectedBlockLen) {
    throw new IOException("Corrupted replica " + replicaInfo + 
        " with a length of " + replicaLen + 
        " expected length is " + expectedBlockLen);
  }
  
  return replicaInfo;
}