All Downloads are FREE. Search and download functionalities are using the official Maven repository.

org.apache.hadoop.hbase.replication.regionserver.ReplicationSourceWALReader Maven / Gradle / Ivy

There is a newer version: 3.0.0-beta-1
Show newest version
/*
 * Licensed to the Apache Software Foundation (ASF) under one
 * or more contributor license agreements.  See the NOTICE file
 * distributed with this work for additional information
 * regarding copyright ownership.  The ASF licenses this file
 * to you under the Apache License, Version 2.0 (the
 * "License"); you may not use this file except in compliance
 * with the License.  You may obtain a copy of the License at
 *
 *     http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */
package org.apache.hadoop.hbase.replication.regionserver;

import java.io.EOFException;
import java.io.IOException;
import java.util.List;
import java.util.concurrent.BlockingQueue;
import java.util.concurrent.LinkedBlockingQueue;
import java.util.concurrent.PriorityBlockingQueue;
import java.util.concurrent.TimeUnit;
import java.util.concurrent.atomic.AtomicLong;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.hbase.Cell;
import org.apache.hadoop.hbase.CellUtil;
import org.apache.hadoop.hbase.replication.WALEntryFilter;
import org.apache.hadoop.hbase.util.Pair;
import org.apache.hadoop.hbase.util.Threads;
import org.apache.hadoop.hbase.wal.AbstractFSWALProvider;
import org.apache.hadoop.hbase.wal.WAL.Entry;
import org.apache.hadoop.hbase.wal.WALEdit;
import org.apache.hadoop.hbase.wal.WALKey;
import org.apache.yetus.audience.InterfaceAudience;
import org.apache.yetus.audience.InterfaceStability;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;

import org.apache.hadoop.hbase.shaded.protobuf.generated.WALProtos.BulkLoadDescriptor;
import org.apache.hadoop.hbase.shaded.protobuf.generated.WALProtos.StoreDescriptor;

/**
 * Reads and filters WAL entries, groups the filtered entries into batches, and puts the batches
 * onto a queue
 */
@InterfaceAudience.Private
@InterfaceStability.Evolving
class ReplicationSourceWALReader extends Thread {
  private static final Logger LOG = LoggerFactory.getLogger(ReplicationSourceWALReader.class);

  private final ReplicationSourceLogQueue logQueue;
  private final FileSystem fs;
  private final Configuration conf;
  private final WALEntryFilter filter;
  private final ReplicationSource source;

  @InterfaceAudience.Private
  final BlockingQueue entryBatchQueue;
  // max (heap) size of each batch - multiply by number of batches in queue to get total
  private final long replicationBatchSizeCapacity;
  // max count of each batch - multiply by number of batches in queue to get total
  private final int replicationBatchCountCapacity;
  // position in the WAL to start reading at
  private long currentPosition;
  private final long sleepForRetries;
  private final int maxRetriesMultiplier;
  private final boolean eofAutoRecovery;

  // Indicates whether this particular worker is running
  private boolean isReaderRunning = true;

  private AtomicLong totalBufferUsed;
  private long totalBufferQuota;
  private final String walGroupId;

  /**
   * Creates a reader worker for a given WAL queue. Reads WAL entries off a given queue, batches the
   * entries, and puts them on a batch queue.
   * @param fs            the files system to use
   * @param conf          configuration to use
   * @param logQueue      The WAL queue to read off of
   * @param startPosition position in the first WAL to start reading from
   * @param filter        The filter to use while reading
   * @param source        replication source
   */
  public ReplicationSourceWALReader(FileSystem fs, Configuration conf,
    ReplicationSourceLogQueue logQueue, long startPosition, WALEntryFilter filter,
    ReplicationSource source, String walGroupId) {
    this.logQueue = logQueue;
    this.currentPosition = startPosition;
    this.fs = fs;
    this.conf = conf;
    this.filter = filter;
    this.source = source;
    this.replicationBatchSizeCapacity =
      this.conf.getLong("replication.source.size.capacity", 1024 * 1024 * 64);
    this.replicationBatchCountCapacity = this.conf.getInt("replication.source.nb.capacity", 25000);
    // memory used will be batchSizeCapacity * (nb.batches + 1)
    // the +1 is for the current thread reading before placing onto the queue
    int batchCount = conf.getInt("replication.source.nb.batches", 1);
    this.totalBufferUsed = source.getSourceManager().getTotalBufferUsed();
    this.totalBufferQuota = source.getSourceManager().getTotalBufferLimit();
    // 1 second
    this.sleepForRetries = this.conf.getLong("replication.source.sleepforretries", 1000);
    // 5 minutes @ 1 sec per
    this.maxRetriesMultiplier = this.conf.getInt("replication.source.maxretriesmultiplier", 300);
    this.eofAutoRecovery = conf.getBoolean("replication.source.eof.autorecovery", false);
    this.entryBatchQueue = new LinkedBlockingQueue<>(batchCount);
    this.walGroupId = walGroupId;
    LOG.info("peerClusterZnode=" + source.getQueueId() + ", ReplicationSourceWALReaderThread : "
      + source.getPeerId() + " inited, replicationBatchSizeCapacity=" + replicationBatchSizeCapacity
      + ", replicationBatchCountCapacity=" + replicationBatchCountCapacity
      + ", replicationBatchQueueCapacity=" + batchCount);
  }

  @Override
  public void run() {
    int sleepMultiplier = 1;
    while (isReaderRunning()) { // we only loop back here if something fatal happened to our stream
      WALEntryBatch batch = null;
      try (WALEntryStream entryStream =
        new WALEntryStream(logQueue, conf, currentPosition, source.getWALFileLengthProvider(),
          source.getServerWALsBelongTo(), source.getSourceMetrics(), walGroupId)) {
        while (isReaderRunning()) { // loop here to keep reusing stream while we can
          batch = null;
          if (!source.isPeerEnabled()) {
            Threads.sleep(sleepForRetries);
            continue;
          }
          if (!checkQuota()) {
            continue;
          }
          batch = tryAdvanceStreamAndCreateWALBatch(entryStream);
          if (batch == null) {
            // got no entries and didn't advance position in WAL
            handleEmptyWALEntryBatch();
            entryStream.reset(); // reuse stream
            continue;
          }
          boolean successAddToQueue = false;
          try {
            // if we have already switched a file, skip reading and put it directly to the ship
            // queue
            if (!batch.isEndOfFile()) {
              readWALEntries(entryStream, batch);
              currentPosition = entryStream.getPosition();
            }
            // need to propagate the batch even it has no entries since it may carry the last
            // sequence id information for serial replication.
            LOG.debug("Read {} WAL entries eligible for replication", batch.getNbEntries());
            entryBatchQueue.put(batch);
            successAddToQueue = true;
            sleepMultiplier = 1;
          } finally {
            if (!successAddToQueue) {
              // batch is not put to ReplicationSourceWALReader#entryBatchQueue,so we should
              // decrease ReplicationSourceWALReader.totalBufferUsed by the byte size which
              // acquired in ReplicationSourceWALReader.acquireBufferQuota.
              this.releaseBufferQuota(batch);
            }
          }
        }
      } catch (WALEntryFilterRetryableException | IOException e) { // stream related
        if (!handleEofException(e, batch)) {
          LOG.warn("Failed to read stream of replication entries", e);
          if (sleepMultiplier < maxRetriesMultiplier) {
            sleepMultiplier++;
          }
          Threads.sleep(sleepForRetries * sleepMultiplier);
        }
      } catch (InterruptedException e) {
        LOG.trace("Interrupted while sleeping between WAL reads or adding WAL batch to ship queue");
        Thread.currentThread().interrupt();
      }
    }
  }

  // returns true if we reach the size limit for batch, i.e, we need to finish the batch and return.
  protected final boolean addEntryToBatch(WALEntryBatch batch, Entry entry) {
    WALEdit edit = entry.getEdit();
    if (edit == null || edit.isEmpty()) {
      LOG.trace("Edit null or empty for entry {} ", entry);
      return false;
    }
    LOG.trace("updating TimeStampOfLastAttempted to {}, from entry {}, for source queue: {}",
      entry.getKey().getWriteTime(), entry.getKey(), this.source.getQueueId());
    long entrySize = getEntrySizeIncludeBulkLoad(entry);
    long entrySizeExcludeBulkLoad = getEntrySizeExcludeBulkLoad(entry);
    batch.addEntry(entry, entrySize);
    updateBatchStats(batch, entry, entrySize);
    boolean totalBufferTooLarge = acquireBufferQuota(batch, entrySizeExcludeBulkLoad);

    // Stop if too many entries or too big
    return totalBufferTooLarge || batch.getHeapSize() >= replicationBatchSizeCapacity
      || batch.getNbEntries() >= replicationBatchCountCapacity;
  }

  protected static final boolean switched(WALEntryStream entryStream, Path path) {
    Path newPath = entryStream.getCurrentPath();
    return newPath == null || !path.getName().equals(newPath.getName());
  }

  // We need to get the WALEntryBatch from the caller so we can add entries in there
  // This is required in case there is any exception in while reading entries
  // we do not want to loss the existing entries in the batch
  protected void readWALEntries(WALEntryStream entryStream, WALEntryBatch batch)
    throws IOException, InterruptedException {
    Path currentPath = entryStream.getCurrentPath();
    for (;;) {
      Entry entry = entryStream.next();
      batch.setLastWalPosition(entryStream.getPosition());
      entry = filterEntry(entry);
      if (entry != null) {
        if (addEntryToBatch(batch, entry)) {
          break;
        }
      }
      boolean hasNext = entryStream.hasNext();
      // always return if we have switched to a new file
      if (switched(entryStream, currentPath)) {
        batch.setEndOfFile(true);
        break;
      }
      if (!hasNext) {
        break;
      }
    }
  }

  private void handleEmptyWALEntryBatch() throws InterruptedException {
    LOG.trace("Didn't read any new entries from WAL");
    if (logQueue.getQueue(walGroupId).isEmpty()) {
      // we're done with current queue, either this is a recovered queue, or it is the special group
      // for a sync replication peer and the peer has been transited to DA or S state.
      LOG.debug("Stopping the replication source wal reader");
      setReaderRunning(false);
      // shuts down shipper thread immediately
      entryBatchQueue.put(WALEntryBatch.NO_MORE_DATA);
    } else {
      Thread.sleep(sleepForRetries);
    }
  }

  private WALEntryBatch tryAdvanceStreamAndCreateWALBatch(WALEntryStream entryStream)
    throws IOException {
    Path currentPath = entryStream.getCurrentPath();
    if (!entryStream.hasNext()) {
      // check whether we have switched a file
      if (currentPath != null && switched(entryStream, currentPath)) {
        return WALEntryBatch.endOfFile(currentPath);
      } else {
        return null;
      }
    }
    if (currentPath != null) {
      if (switched(entryStream, currentPath)) {
        return WALEntryBatch.endOfFile(currentPath);
      }
    }
    return createBatch(entryStream);
  }

  /**
   * This is to handle the EOFException from the WAL entry stream. EOFException should be handled
   * carefully because there are chances of data loss because of never replicating the data. Thus we
   * should always try to ship existing batch of entries here. If there was only one log in the
   * queue before EOF, we ship the empty batch here and since reader is still active, in the next
   * iteration of reader we will stop the reader.
   * 

* If there was more than one log in the queue before EOF, we ship the existing batch and reset * the wal patch and position to the log with EOF, so shipper can remove logs from replication * queue * @return true only the IOE can be handled */ private boolean handleEofException(Exception e, WALEntryBatch batch) { PriorityBlockingQueue queue = logQueue.getQueue(walGroupId); // Dump the log even if logQueue size is 1 if the source is from recovered Source // since we don't add current log to recovered source queue so it is safe to remove. if ( (e instanceof EOFException || e.getCause() instanceof EOFException) && (source.isRecovered() || queue.size() > 1) && this.eofAutoRecovery ) { Path path = queue.peek(); try { if (!fs.exists(path)) { // There is a chance that wal has moved to oldWALs directory, so look there also. path = AbstractFSWALProvider.findArchivedLog(path, conf); // path can be null if unable to locate in archiveDir. } if (path != null && fs.getFileStatus(path).getLen() == 0) { LOG.warn("Forcing removal of 0 length log in queue: {}", path); logQueue.remove(walGroupId); currentPosition = 0; if (batch != null) { // After we removed the WAL from the queue, we should try shipping the existing batch of // entries addBatchToShippingQueue(batch); } return true; } } catch (IOException ioe) { LOG.warn("Couldn't get file length information about log " + path, ioe); } catch (InterruptedException ie) { LOG.trace("Interrupted while adding WAL batch to ship queue"); Thread.currentThread().interrupt(); } } return false; } /** * Update the batch try to ship and return true if shipped * @param batch Batch of entries to ship * @throws InterruptedException throws interrupted exception */ private void addBatchToShippingQueue(WALEntryBatch batch) throws InterruptedException { // need to propagate the batch even it has no entries since it may carry the last // sequence id information for serial replication. LOG.debug("Read {} WAL entries eligible for replication", batch.getNbEntries()); entryBatchQueue.put(batch); } public Path getCurrentPath() { // if we've read some WAL entries, get the Path we read from WALEntryBatch batchQueueHead = entryBatchQueue.peek(); if (batchQueueHead != null) { return batchQueueHead.getLastWalPath(); } // otherwise, we must be currently reading from the head of the log queue return logQueue.getQueue(walGroupId).peek(); } // returns false if we've already exceeded the global quota private boolean checkQuota() { // try not to go over total quota if (totalBufferUsed.get() > totalBufferQuota) { LOG.warn("peer={}, can't read more edits from WAL as buffer usage {}B exceeds limit {}B", this.source.getPeerId(), totalBufferUsed.get(), totalBufferQuota); Threads.sleep(sleepForRetries); return false; } return true; } private WALEntryBatch createBatch(WALEntryStream entryStream) { return new WALEntryBatch(replicationBatchCountCapacity, entryStream.getCurrentPath()); } protected final Entry filterEntry(Entry entry) { Entry filtered = filter.filter(entry); if (entry != null && (filtered == null || filtered.getEdit().size() == 0)) { LOG.trace("Filtered entry for replication: {}", entry); source.getSourceMetrics().incrLogEditsFiltered(); } return filtered; } /** * Retrieves the next batch of WAL entries from the queue, waiting up to the specified time for a * batch to become available * @return A batch of entries, along with the position in the log after reading the batch * @throws InterruptedException if interrupted while waiting */ public WALEntryBatch take() throws InterruptedException { return entryBatchQueue.take(); } public WALEntryBatch poll(long timeout) throws InterruptedException { return entryBatchQueue.poll(timeout, TimeUnit.MILLISECONDS); } private long getEntrySizeIncludeBulkLoad(Entry entry) { WALEdit edit = entry.getEdit(); return getEntrySizeExcludeBulkLoad(entry) + sizeOfStoreFilesIncludeBulkLoad(edit); } public static long getEntrySizeExcludeBulkLoad(Entry entry) { WALEdit edit = entry.getEdit(); WALKey key = entry.getKey(); return edit.heapSize() + key.estimatedSerializedSizeOf(); } private void updateBatchStats(WALEntryBatch batch, Entry entry, long entrySize) { WALEdit edit = entry.getEdit(); batch.incrementHeapSize(entrySize); Pair nbRowsAndHFiles = countDistinctRowKeysAndHFiles(edit); batch.incrementNbRowKeys(nbRowsAndHFiles.getFirst()); batch.incrementNbHFiles(nbRowsAndHFiles.getSecond()); } /** * Count the number of different row keys in the given edit because of mini-batching. We assume * that there's at least one Cell in the WALEdit. * @param edit edit to count row keys from * @return number of different row keys and HFiles */ private Pair countDistinctRowKeysAndHFiles(WALEdit edit) { List cells = edit.getCells(); int distinctRowKeys = 1; int totalHFileEntries = 0; Cell lastCell = cells.get(0); int totalCells = edit.size(); for (int i = 0; i < totalCells; i++) { // Count HFiles to be replicated if (CellUtil.matchingQualifier(cells.get(i), WALEdit.BULK_LOAD)) { try { BulkLoadDescriptor bld = WALEdit.getBulkLoadDescriptor(cells.get(i)); List stores = bld.getStoresList(); int totalStores = stores.size(); for (int j = 0; j < totalStores; j++) { totalHFileEntries += stores.get(j).getStoreFileList().size(); } } catch (IOException e) { LOG.error("Failed to deserialize bulk load entry from wal edit. " + "Then its hfiles count will not be added into metric.", e); } } if (!CellUtil.matchingRows(cells.get(i), lastCell)) { distinctRowKeys++; } lastCell = cells.get(i); } Pair result = new Pair<>(distinctRowKeys, totalHFileEntries); return result; } /** * Calculate the total size of all the store files * @param edit edit to count row keys from * @return the total size of the store files */ private int sizeOfStoreFilesIncludeBulkLoad(WALEdit edit) { List cells = edit.getCells(); int totalStoreFilesSize = 0; int totalCells = edit.size(); for (int i = 0; i < totalCells; i++) { if (CellUtil.matchingQualifier(cells.get(i), WALEdit.BULK_LOAD)) { try { BulkLoadDescriptor bld = WALEdit.getBulkLoadDescriptor(cells.get(i)); List stores = bld.getStoresList(); int totalStores = stores.size(); for (int j = 0; j < totalStores; j++) { totalStoreFilesSize = (int) (totalStoreFilesSize + stores.get(j).getStoreFileSizeBytes()); } } catch (IOException e) { LOG.error("Failed to deserialize bulk load entry from wal edit. " + "Size of HFiles part of cell will not be considered in replication " + "request size calculation.", e); } } } return totalStoreFilesSize; } /** * @param size delta size for grown buffer * @return true if we should clear buffer and push all */ private boolean acquireBufferQuota(WALEntryBatch walEntryBatch, long size) { long newBufferUsed = totalBufferUsed.addAndGet(size); // Record the new buffer usage this.source.getSourceManager().getGlobalMetrics().setWALReaderEditsBufferBytes(newBufferUsed); walEntryBatch.incrementUsedBufferSize(size); return newBufferUsed >= totalBufferQuota; } /** * To release the buffer quota of {@link WALEntryBatch} which acquired by * {@link ReplicationSourceWALReader#acquireBufferQuota} */ private void releaseBufferQuota(WALEntryBatch walEntryBatch) { long usedBufferSize = walEntryBatch.getUsedBufferSize(); if (usedBufferSize > 0) { long newBufferUsed = totalBufferUsed.addAndGet(-usedBufferSize); this.source.getSourceManager().getGlobalMetrics().setWALReaderEditsBufferBytes(newBufferUsed); } } /** Returns whether the reader thread is running */ public boolean isReaderRunning() { return isReaderRunning && !isInterrupted(); } /** * @param readerRunning the readerRunning to set */ public void setReaderRunning(boolean readerRunning) { this.isReaderRunning = readerRunning; } }





© 2015 - 2024 Weber Informatics LLC | Privacy Policy