Package org.apache.hadoop.hbase.regionserver

Source Code of org.apache.hadoop.hbase.regionserver.SplitTransaction$StoreFileSplitter

/**
*
* 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.regionserver;

import java.io.IOException;
import java.io.InterruptedIOException;
import java.util.ArrayList;
import java.util.List;
import java.util.ListIterator;
import java.util.Map;
import java.util.concurrent.Callable;
import java.util.concurrent.ExecutionException;
import java.util.concurrent.Executors;
import java.util.concurrent.Future;
import java.util.concurrent.ThreadFactory;
import java.util.concurrent.ThreadPoolExecutor;
import java.util.concurrent.TimeUnit;

import org.apache.commons.logging.Log;
import org.apache.commons.logging.LogFactory;
import org.apache.hadoop.classification.InterfaceAudience;
import org.apache.hadoop.hbase.HConstants;
import org.apache.hadoop.hbase.HRegionInfo;
import org.apache.hadoop.hbase.Server;
import org.apache.hadoop.hbase.ServerName;
import org.apache.hadoop.hbase.MetaTableAccessor;
import org.apache.hadoop.hbase.client.HConnection;
import org.apache.hadoop.hbase.client.Mutation;
import org.apache.hadoop.hbase.client.Put;
import org.apache.hadoop.hbase.coordination.BaseCoordinatedStateManager;
import org.apache.hadoop.hbase.coordination.SplitTransactionCoordination;
import org.apache.hadoop.hbase.protobuf.generated.RegionServerStatusProtos.RegionStateTransition.TransitionCode;
import org.apache.hadoop.hbase.util.Bytes;
import org.apache.hadoop.hbase.util.CancelableProgressable;
import org.apache.hadoop.hbase.util.ConfigUtil;
import org.apache.hadoop.hbase.util.EnvironmentEdgeManager;
import org.apache.hadoop.hbase.util.HasThread;
import org.apache.hadoop.hbase.util.PairOfSameType;
import org.apache.zookeeper.KeeperException;

import com.google.common.util.concurrent.ThreadFactoryBuilder;

/**
* Executes region split as a "transaction".  Call {@link #prepare()} to setup
* the transaction, {@link #execute(Server, RegionServerServices)} to run the
* transaction and {@link #rollback(Server, RegionServerServices)} to cleanup if execute fails.
*
* <p>Here is an example of how you would use this class:
* <pre>
*  SplitTransaction st = new SplitTransaction(this.conf, parent, midKey)
*  if (!st.prepare()) return;
*  try {
*    st.execute(server, services);
*  } catch (IOException ioe) {
*    try {
*      st.rollback(server, services);
*      return;
*    } catch (RuntimeException e) {
*      myAbortable.abort("Failed split, abort");
*    }
*  }
* </Pre>
* <p>This class is not thread safe.  Caller needs ensure split is run by
* one thread only.
*/
@InterfaceAudience.Private
public class SplitTransaction {
  private static final Log LOG = LogFactory.getLog(SplitTransaction.class);

  /*
   * Region to split
   */
  private final HRegion parent;
  private HRegionInfo hri_a;
  private HRegionInfo hri_b;
  private long fileSplitTimeout = 30000;
  public SplitTransactionCoordination.SplitTransactionDetails std;
  boolean useZKForAssignment;

  /*
   * Row to split around
   */
  private final byte [] splitrow;

  /**
   * Types to add to the transaction journal.
   * Each enum is a step in the split transaction. Used to figure how much
   * we need to rollback.
   */
  enum JournalEntry {
    /**
     * Set region as in transition, set it into SPLITTING state.
     */
    SET_SPLITTING,
    /**
     * We created the temporary split data directory.
     */
    CREATE_SPLIT_DIR,
    /**
     * Closed the parent region.
     */
    CLOSED_PARENT_REGION,
    /**
     * The parent has been taken out of the server's online regions list.
     */
    OFFLINED_PARENT,
    /**
     * Started in on creation of the first daughter region.
     */
    STARTED_REGION_A_CREATION,
    /**
     * Started in on the creation of the second daughter region.
     */
    STARTED_REGION_B_CREATION,
    /**
     * Point of no return.
     * If we got here, then transaction is not recoverable other than by
     * crashing out the regionserver.
     */
    PONR
  }

  /*
   * Journal of how far the split transaction has progressed.
   */
  private final List<JournalEntry> journal = new ArrayList<JournalEntry>();

  /**
   * Constructor
   * @param r Region to split
   * @param splitrow Row to split around
   */
  public SplitTransaction(final HRegion r, final byte [] splitrow) {
    this.parent = r;
    this.splitrow = splitrow;
  }

  /**
   * Does checks on split inputs.
   * @return <code>true</code> if the region is splittable else
   * <code>false</code> if it is not (e.g. its already closed, etc.).
   */
  public boolean prepare() {
    if (!this.parent.isSplittable()) return false;
    // Split key can be null if this region is unsplittable; i.e. has refs.
    if (this.splitrow == null) return false;
    HRegionInfo hri = this.parent.getRegionInfo();
    parent.prepareToSplit();
    // Check splitrow.
    byte [] startKey = hri.getStartKey();
    byte [] endKey = hri.getEndKey();
    if (Bytes.equals(startKey, splitrow) ||
        !this.parent.getRegionInfo().containsRow(splitrow)) {
      LOG.info("Split row is not inside region key range or is equal to " +
          "startkey: " + Bytes.toStringBinary(this.splitrow));
      return false;
    }
    long rid = getDaughterRegionIdTimestamp(hri);
    this.hri_a = new HRegionInfo(hri.getTable(), startKey, this.splitrow, false, rid);
    this.hri_b = new HRegionInfo(hri.getTable(), this.splitrow, endKey, false, rid);
    return true;
  }

  /**
   * Calculate daughter regionid to use.
   * @param hri Parent {@link HRegionInfo}
   * @return Daughter region id (timestamp) to use.
   */
  private static long getDaughterRegionIdTimestamp(final HRegionInfo hri) {
    long rid = EnvironmentEdgeManager.currentTime();
    // Regionid is timestamp.  Can't be less than that of parent else will insert
    // at wrong location in hbase:meta (See HBASE-710).
    if (rid < hri.getRegionId()) {
      LOG.warn("Clock skew; parent regions id is " + hri.getRegionId() +
        " but current time here is " + rid);
      rid = hri.getRegionId() + 1;
    }
    return rid;
  }

  private static IOException closedByOtherException = new IOException(
      "Failed to close region: already closed by another thread");

  /**
   * Prepare the regions and region files.
   * @param server Hosting server instance.  Can be null when testing (won't try
   * and update in zk if a null server)
   * @param services Used to online/offline regions.
   * @throws IOException If thrown, transaction failed.
   *    Call {@link #rollback(Server, RegionServerServices)}
   * @return Regions created
   */
  /* package */PairOfSameType<HRegion> createDaughters(final Server server,
      final RegionServerServices services) throws IOException {
    LOG.info("Starting split of region " + this.parent);
    if ((server != null && server.isStopped()) ||
        (services != null && services.isStopping())) {
      throw new IOException("Server is stopped or stopping");
    }
    assert !this.parent.lock.writeLock().isHeldByCurrentThread():
      "Unsafe to hold write lock while performing RPCs";

    // Coprocessor callback
    if (this.parent.getCoprocessorHost() != null) {
      this.parent.getCoprocessorHost().preSplit();
    }

    // Coprocessor callback
    if (this.parent.getCoprocessorHost() != null) {
      this.parent.getCoprocessorHost().preSplit(this.splitrow);
    }

    // If true, no cluster to write meta edits to or to update znodes in.
    boolean testing = server == null? true:
        server.getConfiguration().getBoolean("hbase.testing.nocluster", false);
    this.fileSplitTimeout = testing ? this.fileSplitTimeout :
        server.getConfiguration().getLong("hbase.regionserver.fileSplitTimeout",
          this.fileSplitTimeout);

    PairOfSameType<HRegion> daughterRegions = stepsBeforePONR(server, services, testing);

    List<Mutation> metaEntries = new ArrayList<Mutation>();
    if (this.parent.getCoprocessorHost() != null) {
      if (this.parent.getCoprocessorHost().
          preSplitBeforePONR(this.splitrow, metaEntries)) {
        throw new IOException("Coprocessor bypassing region "
            + this.parent.getRegionNameAsString() + " split.");
      }
      try {
        for (Mutation p : metaEntries) {
          HRegionInfo.parseRegionName(p.getRow());
        }
      } catch (IOException e) {
        LOG.error("Row key of mutation from coprossor is not parsable as region name."
            + "Mutations from coprocessor should only for hbase:meta table.");
        throw e;
      }
    }

    // This is the point of no return.  Adding subsequent edits to .META. as we
    // do below when we do the daughter opens adding each to .META. can fail in
    // various interesting ways the most interesting of which is a timeout
    // BUT the edits all go through (See HBASE-3872).  IF we reach the PONR
    // then subsequent failures need to crash out this regionserver; the
    // server shutdown processing should be able to fix-up the incomplete split.
    // The offlined parent will have the daughters as extra columns.  If
    // we leave the daughter regions in place and do not remove them when we
    // crash out, then they will have their references to the parent in place
    // still and the server shutdown fixup of .META. will point to these
    // regions.
    // We should add PONR JournalEntry before offlineParentInMeta,so even if
    // OfflineParentInMeta timeout,this will cause regionserver exit,and then
    // master ServerShutdownHandler will fix daughter & avoid data loss. (See
    // HBase-4562).
    this.journal.add(JournalEntry.PONR);

    // Edit parent in meta.  Offlines parent region and adds splita and splitb
    // as an atomic update. See HBASE-7721. This update to META makes the region
    // will determine whether the region is split or not in case of failures.
    // If it is successful, master will roll-forward, if not, master will rollback
    // and assign the parent region.
    if (!testing && useZKForAssignment) {
      if (metaEntries == null || metaEntries.isEmpty()) {
        MetaTableAccessor.splitRegion(server.getShortCircuitConnection(),
          parent.getRegionInfo(), daughterRegions.getFirst().getRegionInfo(),
          daughterRegions.getSecond().getRegionInfo(), server.getServerName());
      } else {
        offlineParentInMetaAndputMetaEntries(server.getShortCircuitConnection(),
          parent.getRegionInfo(), daughterRegions.getFirst().getRegionInfo(), daughterRegions
              .getSecond().getRegionInfo(), server.getServerName(), metaEntries);
      }
    } else if (services != null && !useZKForAssignment) {
      if (!services.reportRegionStateTransition(TransitionCode.SPLIT_PONR,
          parent.getRegionInfo(), hri_a, hri_b)) {
        // Passed PONR, let SSH clean it up
        throw new IOException("Failed to notify master that split passed PONR: "
          + parent.getRegionInfo().getRegionNameAsString());
      }
    }
    return daughterRegions;
  }

  public PairOfSameType<HRegion> stepsBeforePONR(final Server server,
      final RegionServerServices services, boolean testing) throws IOException {

    if (useCoordinatedStateManager(server)) {
      if (std == null) {
        std =
            ((BaseCoordinatedStateManager) server.getCoordinatedStateManager())
                .getSplitTransactionCoordination().getDefaultDetails();
      }
      ((BaseCoordinatedStateManager) server.getCoordinatedStateManager())
          .getSplitTransactionCoordination().startSplitTransaction(parent, server.getServerName(),
            hri_a, hri_b);
    } else if (services != null && !useZKForAssignment) {
      if (!services.reportRegionStateTransition(TransitionCode.READY_TO_SPLIT,
          parent.getRegionInfo(), hri_a, hri_b)) {
        throw new IOException("Failed to get ok from master to split "
          + parent.getRegionNameAsString());
      }
    }
    this.journal.add(JournalEntry.SET_SPLITTING);
    if (useCoordinatedStateManager(server)) {
      ((BaseCoordinatedStateManager) server.getCoordinatedStateManager())
          .getSplitTransactionCoordination().waitForSplitTransaction(services, parent, hri_a,
            hri_b, std);
    }

    this.parent.getRegionFileSystem().createSplitsDir();
    this.journal.add(JournalEntry.CREATE_SPLIT_DIR);

    Map<byte[], List<StoreFile>> hstoreFilesToSplit = null;
    Exception exceptionToThrow = null;
    try{
      hstoreFilesToSplit = this.parent.close(false);
    } catch (Exception e) {
      exceptionToThrow = e;
    }
    if (exceptionToThrow == null && hstoreFilesToSplit == null) {
      // The region was closed by a concurrent thread.  We can't continue
      // with the split, instead we must just abandon the split.  If we
      // reopen or split this could cause problems because the region has
      // probably already been moved to a different server, or is in the
      // process of moving to a different server.
      exceptionToThrow = closedByOtherException;
    }
    if (exceptionToThrow != closedByOtherException) {
      this.journal.add(JournalEntry.CLOSED_PARENT_REGION);
    }
    if (exceptionToThrow != null) {
      if (exceptionToThrow instanceof IOException) throw (IOException)exceptionToThrow;
      throw new IOException(exceptionToThrow);
    }
    if (!testing) {
      services.removeFromOnlineRegions(this.parent, null);
    }
    this.journal.add(JournalEntry.OFFLINED_PARENT);

    // TODO: If splitStoreFiles were multithreaded would we complete steps in
    // less elapsed time?  St.Ack 20100920
    //
    // splitStoreFiles creates daughter region dirs under the parent splits dir
    // Nothing to unroll here if failure -- clean up of CREATE_SPLIT_DIR will
    // clean this up.
    splitStoreFiles(hstoreFilesToSplit);

    // Log to the journal that we are creating region A, the first daughter
    // region.  We could fail halfway through.  If we do, we could have left
    // stuff in fs that needs cleanup -- a storefile or two.  Thats why we
    // add entry to journal BEFORE rather than AFTER the change.
    this.journal.add(JournalEntry.STARTED_REGION_A_CREATION);
    HRegion a = this.parent.createDaughterRegionFromSplits(this.hri_a);

    // Ditto
    this.journal.add(JournalEntry.STARTED_REGION_B_CREATION);
    HRegion b = this.parent.createDaughterRegionFromSplits(this.hri_b);
    return new PairOfSameType<HRegion>(a, b);
  }

  /**
   * Perform time consuming opening of the daughter regions.
   * @param server Hosting server instance.  Can be null when testing
   * @param services Used to online/offline regions.
   * @param a first daughter region
   * @param a second daughter region
   * @throws IOException If thrown, transaction failed.
   *          Call {@link #rollback(Server, RegionServerServices)}
   */
  /* package */void openDaughters(final Server server,
      final RegionServerServices services, HRegion a, HRegion b)
      throws IOException {
    boolean stopped = server != null && server.isStopped();
    boolean stopping = services != null && services.isStopping();
    // TODO: Is this check needed here?
    if (stopped || stopping) {
      LOG.info("Not opening daughters " +
          b.getRegionInfo().getRegionNameAsString() +
          " and " +
          a.getRegionInfo().getRegionNameAsString() +
          " because stopping=" + stopping + ", stopped=" + stopped);
    } else {
      // Open daughters in parallel.
      DaughterOpener aOpener = new DaughterOpener(server, a);
      DaughterOpener bOpener = new DaughterOpener(server, b);
      aOpener.start();
      bOpener.start();
      try {
        aOpener.join();
        bOpener.join();
      } catch (InterruptedException e) {
        throw (InterruptedIOException)new InterruptedIOException().initCause(e);
      }
      if (aOpener.getException() != null) {
        throw new IOException("Failed " +
          aOpener.getName(), aOpener.getException());
      }
      if (bOpener.getException() != null) {
        throw new IOException("Failed " +
          bOpener.getName(), bOpener.getException());
      }
      if (services != null) {
        try {
          if (useZKForAssignment) {
            // add 2nd daughter first (see HBASE-4335)
            services.postOpenDeployTasks(b);
          } else if (!services.reportRegionStateTransition(TransitionCode.SPLIT,
              parent.getRegionInfo(), hri_a, hri_b)) {
            throw new IOException("Failed to report split region to master: "
              + parent.getRegionInfo().getShortNameToLog());
          }
          // Should add it to OnlineRegions
          services.addToOnlineRegions(b);
          if (useZKForAssignment) {
            services.postOpenDeployTasks(a);
          }
          services.addToOnlineRegions(a);
        } catch (KeeperException ke) {
          throw new IOException(ke);
        }
      }
    }
  }

  /**
   * Run the transaction.
   * @param server Hosting server instance.  Can be null when testing
   * @param services Used to online/offline regions.
   * @throws IOException If thrown, transaction failed.
   *          Call {@link #rollback(Server, RegionServerServices)}
   * @return Regions created
   * @throws IOException
   * @see #rollback(Server, RegionServerServices)
   */
  public PairOfSameType<HRegion> execute(final Server server,
      final RegionServerServices services)
  throws IOException {
    useZKForAssignment = server == null ? true :
      ConfigUtil.useZKForAssignment(server.getConfiguration());
    if (useCoordinatedStateManager(server)) {
      std =
          ((BaseCoordinatedStateManager) server.getCoordinatedStateManager())
              .getSplitTransactionCoordination().getDefaultDetails();
    }
    PairOfSameType<HRegion> regions = createDaughters(server, services);
    if (this.parent.getCoprocessorHost() != null) {
      this.parent.getCoprocessorHost().preSplitAfterPONR();
    }
    return stepsAfterPONR(server, services, regions);
  }

  public PairOfSameType<HRegion> stepsAfterPONR(final Server server,
      final RegionServerServices services, PairOfSameType<HRegion> regions)
      throws IOException {
    openDaughters(server, services, regions.getFirst(), regions.getSecond());
    if (useCoordinatedStateManager(server)) {
      ((BaseCoordinatedStateManager) server.getCoordinatedStateManager())
          .getSplitTransactionCoordination().completeSplitTransaction(services, regions.getFirst(),
            regions.getSecond(), std, parent);
    }
    // Coprocessor callback
    if (parent.getCoprocessorHost() != null) {
      parent.getCoprocessorHost().postSplit(regions.getFirst(), regions.getSecond());
    }


    return regions;
  }

  private void offlineParentInMetaAndputMetaEntries(HConnection hConnection,
      HRegionInfo parent, HRegionInfo splitA, HRegionInfo splitB,
      ServerName serverName, List<Mutation> metaEntries) throws IOException {
    List<Mutation> mutations = metaEntries;
    HRegionInfo copyOfParent = new HRegionInfo(parent);
    copyOfParent.setOffline(true);
    copyOfParent.setSplit(true);

    //Put for parent
    Put putParent = MetaTableAccessor.makePutFromRegionInfo(copyOfParent);
    MetaTableAccessor.addDaughtersToPut(putParent, splitA, splitB);
    mutations.add(putParent);
   
    //Puts for daughters
    Put putA = MetaTableAccessor.makePutFromRegionInfo(splitA);
    Put putB = MetaTableAccessor.makePutFromRegionInfo(splitB);

    addLocation(putA, serverName, 1); //these are new regions, openSeqNum = 1 is fine.
    addLocation(putB, serverName, 1);
    mutations.add(putA);
    mutations.add(putB);
    MetaTableAccessor.mutateMetaTable(hConnection, mutations);
  }

  public Put addLocation(final Put p, final ServerName sn, long openSeqNum) {
    p.addImmutable(HConstants.CATALOG_FAMILY, HConstants.SERVER_QUALIFIER,
      Bytes.toBytes(sn.getHostAndPort()));
    p.addImmutable(HConstants.CATALOG_FAMILY, HConstants.STARTCODE_QUALIFIER,
      Bytes.toBytes(sn.getStartcode()));
    p.addImmutable(HConstants.CATALOG_FAMILY, HConstants.SEQNUM_QUALIFIER,
        Bytes.toBytes(openSeqNum));
    return p;
  }

  /*
   * Open daughter region in its own thread.
   * If we fail, abort this hosting server.
   */
  class DaughterOpener extends HasThread {
    private final Server server;
    private final HRegion r;
    private Throwable t = null;

    DaughterOpener(final Server s, final HRegion r) {
      super((s == null? "null-services": s.getServerName()) +
        "-daughterOpener=" + r.getRegionInfo().getEncodedName());
      setDaemon(true);
      this.server = s;
      this.r = r;
    }

    /**
     * @return Null if open succeeded else exception that causes us fail open.
     * Call it after this thread exits else you may get wrong view on result.
     */
    Throwable getException() {
      return this.t;
    }

    @Override
    public void run() {
      try {
        openDaughterRegion(this.server, r);
      } catch (Throwable t) {
        this.t = t;
      }
    }
  }

  /**
   * Open daughter regions, add them to online list and update meta.
   * @param server
   * @param daughter
   * @throws IOException
   * @throws KeeperException
   */
  void openDaughterRegion(final Server server, final HRegion daughter)
  throws IOException, KeeperException {
    HRegionInfo hri = daughter.getRegionInfo();
    LoggingProgressable reporter = server == null ? null
        : new LoggingProgressable(hri, server.getConfiguration().getLong(
            "hbase.regionserver.split.daughter.open.log.interval", 10000));
    daughter.openHRegion(reporter);
  }

  static class LoggingProgressable implements CancelableProgressable {
    private final HRegionInfo hri;
    private long lastLog = -1;
    private final long interval;

    LoggingProgressable(final HRegionInfo hri, final long interval) {
      this.hri = hri;
      this.interval = interval;
    }

    @Override
    public boolean progress() {
      long now = System.currentTimeMillis();
      if (now - lastLog > this.interval) {
        LOG.info("Opening " + this.hri.getRegionNameAsString());
        this.lastLog = now;
      }
      return true;
    }
  }

  private boolean useCoordinatedStateManager(final Server server) {
    return server != null && useZKForAssignment && server.getCoordinatedStateManager() != null;
  }

  private void splitStoreFiles(final Map<byte[], List<StoreFile>> hstoreFilesToSplit)
      throws IOException {
    if (hstoreFilesToSplit == null) {
      // Could be null because close didn't succeed -- for now consider it fatal
      throw new IOException("Close returned empty list of StoreFiles");
    }
    // The following code sets up a thread pool executor with as many slots as
    // there's files to split. It then fires up everything, waits for
    // completion and finally checks for any exception
    int nbFiles = hstoreFilesToSplit.size();
    if (nbFiles == 0) {
      // no file needs to be splitted.
      return;
    }
    ThreadFactoryBuilder builder = new ThreadFactoryBuilder();
    builder.setNameFormat("StoreFileSplitter-%1$d");
    ThreadFactory factory = builder.build();
    ThreadPoolExecutor threadPool =
      (ThreadPoolExecutor) Executors.newFixedThreadPool(nbFiles, factory);
    List<Future<Void>> futures = new ArrayList<Future<Void>>(nbFiles);

    // Split each store file.
    for (Map.Entry<byte[], List<StoreFile>> entry: hstoreFilesToSplit.entrySet()) {
      for (StoreFile sf: entry.getValue()) {
        StoreFileSplitter sfs = new StoreFileSplitter(entry.getKey(), sf);
        futures.add(threadPool.submit(sfs));
      }
    }
    // Shutdown the pool
    threadPool.shutdown();

    // Wait for all the tasks to finish
    try {
      boolean stillRunning = !threadPool.awaitTermination(
          this.fileSplitTimeout, TimeUnit.MILLISECONDS);
      if (stillRunning) {
        threadPool.shutdownNow();
        // wait for the thread to shutdown completely.
        while (!threadPool.isTerminated()) {
          Thread.sleep(50);
        }
        throw new IOException("Took too long to split the" +
            " files and create the references, aborting split");
      }
    } catch (InterruptedException e) {
      throw (InterruptedIOException)new InterruptedIOException().initCause(e);
    }

    // Look for any exception
    for (Future<Void> future: futures) {
      try {
        future.get();
      } catch (InterruptedException e) {
        throw (InterruptedIOException)new InterruptedIOException().initCause(e);
      } catch (ExecutionException e) {
        throw new IOException(e);
      }
    }
  }

  private void splitStoreFile(final byte[] family, final StoreFile sf) throws IOException {
    HRegionFileSystem fs = this.parent.getRegionFileSystem();
    String familyName = Bytes.toString(family);
    fs.splitStoreFile(this.hri_a, familyName, sf, this.splitrow, false);
    fs.splitStoreFile(this.hri_b, familyName, sf, this.splitrow, true);
  }

  /**
   * Utility class used to do the file splitting / reference writing
   * in parallel instead of sequentially.
   */
  class StoreFileSplitter implements Callable<Void> {
    private final byte[] family;
    private final StoreFile sf;

    /**
     * Constructor that takes what it needs to split
     * @param family Family that contains the store file
     * @param sf which file
     */
    public StoreFileSplitter(final byte[] family, final StoreFile sf) {
      this.sf = sf;
      this.family = family;
    }

    public Void call() throws IOException {
      splitStoreFile(family, sf);
      return null;
    }
  }

  /**
   * @param server Hosting server instance (May be null when testing).
   * @param services
   * @throws IOException If thrown, rollback failed.  Take drastic action.
   * @return True if we successfully rolled back, false if we got to the point
   * of no return and so now need to abort the server to minimize damage.
   */
  @SuppressWarnings("deprecation")
  public boolean rollback(final Server server, final RegionServerServices services)
  throws IOException {
    // Coprocessor callback
    if (this.parent.getCoprocessorHost() != null) {
      this.parent.getCoprocessorHost().preRollBackSplit();
    }

    boolean result = true;
    ListIterator<JournalEntry> iterator =
      this.journal.listIterator(this.journal.size());
    // Iterate in reverse.
    while (iterator.hasPrevious()) {
      JournalEntry je = iterator.previous();
      switch(je) {

      case SET_SPLITTING:
        if (useCoordinatedStateManager(server) && server instanceof HRegionServer) {
          ((BaseCoordinatedStateManager) server.getCoordinatedStateManager())
              .getSplitTransactionCoordination().clean(this.parent.getRegionInfo());
        } else if (services != null && !useZKForAssignment
            && !services.reportRegionStateTransition(TransitionCode.SPLIT_REVERTED,
                parent.getRegionInfo(), hri_a, hri_b)) {
          return false;
        }
        break;

      case CREATE_SPLIT_DIR:
        this.parent.writestate.writesEnabled = true;
        this.parent.getRegionFileSystem().cleanupSplitsDir();
        break;

      case CLOSED_PARENT_REGION:
        try {
          // So, this returns a seqid but if we just closed and then reopened, we
          // should be ok. On close, we flushed using sequenceid obtained from
          // hosting regionserver so no need to propagate the sequenceid returned
          // out of initialize below up into regionserver as we normally do.
          // TODO: Verify.
          this.parent.initialize();
        } catch (IOException e) {
          LOG.error("Failed rollbacking CLOSED_PARENT_REGION of region " +
            this.parent.getRegionNameAsString(), e);
          throw new RuntimeException(e);
        }
        break;

      case STARTED_REGION_A_CREATION:
        this.parent.getRegionFileSystem().cleanupDaughterRegion(this.hri_a);
        break;

      case STARTED_REGION_B_CREATION:
        this.parent.getRegionFileSystem().cleanupDaughterRegion(this.hri_b);
        break;

      case OFFLINED_PARENT:
        if (services != null) services.addToOnlineRegions(this.parent);
        break;

      case PONR:
        // We got to the point-of-no-return so we need to just abort. Return
        // immediately.  Do not clean up created daughter regions.  They need
        // to be in place so we don't delete the parent region mistakenly.
        // See HBASE-3872.
        return false;

      default:
        throw new RuntimeException("Unhandled journal entry: " + je);
      }
    }
    // Coprocessor callback
    if (this.parent.getCoprocessorHost() != null) {
      this.parent.getCoprocessorHost().postRollBackSplit();
    }
    return result;
  }

  HRegionInfo getFirstDaughter() {
    return hri_a;
  }

  HRegionInfo getSecondDaughter() {
    return hri_b;
  }

}
TOP

Related Classes of org.apache.hadoop.hbase.regionserver.SplitTransaction$StoreFileSplitter

TOP
Copyright © 2018 www.massapi.com. All rights reserved.
All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.