Package com.alibaba.wasp.master

Source Code of com.alibaba.wasp.master.ActiveMasterManager

/**
* 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 com.alibaba.wasp.master;

import com.alibaba.wasp.DeserializationException;
import com.alibaba.wasp.Server;
import com.alibaba.wasp.ServerName;
import com.alibaba.wasp.ZNodeClearer;
import com.alibaba.wasp.monitoring.MonitoredTask;
import com.alibaba.wasp.zookeeper.ClusterStatusTracker;
import com.alibaba.wasp.zookeeper.MasterAddressTracker;
import com.alibaba.wasp.zookeeper.ZKUtil;
import com.alibaba.wasp.zookeeper.ZooKeeperListener;
import com.alibaba.wasp.zookeeper.ZooKeeperWatcher;
import org.apache.commons.logging.Log;
import org.apache.commons.logging.LogFactory;
import org.apache.zookeeper.KeeperException;

import java.io.IOException;
import java.util.concurrent.atomic.AtomicBoolean;

/**
* Handles everything on master-side related to master election.
*
* <p>
* Listens and responds to ZooKeeper notifications on the master znode, both
* <code>nodeCreated</code> and <code>nodeDeleted</code>.
*
* <p>
* Contains blocking methods which will hold up backup masters, waiting for the
* active master to fail.
*
* <p>
* This class is instantiated in the FMaster constructor and the method
* #blockUntilBecomingActiveMaster() is called to wait until becoming the active
* master of the cluster.
*/
public class ActiveMasterManager extends ZooKeeperListener {
  private static final Log LOG = LogFactory.getLog(ActiveMasterManager.class);

  final AtomicBoolean clusterHasActiveMaster = new AtomicBoolean(false);

  private final ServerName sn;
  private final Server master;

  /**
   * @param watcher
   * @param sn ServerName
   * @param master In an instance of a Master.
   */
  ActiveMasterManager(ZooKeeperWatcher watcher, ServerName sn, Server master) {
    super(watcher);
    this.sn = sn;
    this.master = master;
  }

  @Override
  public void nodeCreated(String path) {
    handle(path);
  }

  @Override
  public void nodeDeleted(String path) {
    handle(path);
  }

  void handle(final String path) {
    if (path.equals(watcher.getMasterAddressZNode()) && !master.isStopped()) {
      handleMasterNodeChange();
    }
  }

  /**
   * Handle a change in the master node. Doesn't matter whether this was called
   * from a nodeCreated or nodeDeleted event because there are no guarantees
   * that the current state of the master node matches the event at the time of
   * our next ZK request.
   *
   * <p>
   * Uses the watchAndCheckExists method which watches the master address node
   * regardless of whether it exists or not. If it does exist (there is an
   * active master), it returns true. Otherwise it returns false.
   *
   * <p>
   * A watcher is set which guarantees that this method will get called again if
   * there is another change in the master node.
   */
  private void handleMasterNodeChange() {
    // Watch the node and check if it exists.
    try {
      synchronized (clusterHasActiveMaster) {
        if (ZKUtil
            .watchAndCheckExists(watcher, watcher.getMasterAddressZNode())) {
          // A master node exists, there is an active master
          LOG.debug("A master is now available");
          clusterHasActiveMaster.set(true);
        } else {
          // Node is no longer there, cluster does not have an active master
          LOG.debug("No master available. Notifying waiting threads");
          clusterHasActiveMaster.set(false);
          // Notify any thread waiting to become the active master
          clusterHasActiveMaster.notifyAll();
        }
      }
    } catch (KeeperException ke) {
      master.abort("Received an unexpected KeeperException, aborting", ke);
    }
  }

  /**
   * Block until becoming the active master.
   *
   * Method blocks until there is not another active master and our attempt to
   * become the new active master is successful.
   *
   * This also makes sure that we are watching the master znode so will be
   * notified if another master dies.
   * @param startupStatus
   * @return True if no issue becoming active master else false if another
   *         master was running or if some other problem (zookeeper, stop flag
   *         has been set on this Master)
   */
  boolean blockUntilBecomingActiveMaster(MonitoredTask startupStatus,
      ClusterStatusTracker clusterStatusTracker) {
    while (true) {
      startupStatus.setStatus("Trying to register in ZK as active master");
      // Try to become the active master, watch if there is another master.
      // Write out our ServerName as versioned bytes.
      try {
        String backupZNode = ZKUtil.joinZNode(
            this.watcher.backupMasterAddressesZNode, this.sn.toString());
        if (MasterAddressTracker.setMasterAddress(this.watcher,
            this.watcher.getMasterAddressZNode(), this.sn)) {

          // If we were a backup master before, delete our ZNode from the backup
          // master directory since we are the active now)
          if (ZKUtil.checkExists(this.watcher, backupZNode) != -1) {
            LOG.info("Deleting ZNode for " + backupZNode
                + " from backup master directory");
            ZKUtil.deleteNodeFailSilent(this.watcher, backupZNode);
          }
          // Save the znode in a file, this will allow to check if we crash in
          // the launch scripts
          ZNodeClearer.writeMyEphemeralNodeOnDisk(this.sn.toString());

          // We are the master, return
          startupStatus.setStatus("Successfully registered as active master.");
          this.clusterHasActiveMaster.set(true);
          LOG.info("Master=" + this.sn);
          return true;
        }

        // There is another active master running elsewhere or this is a restart
        // and the master ephemeral node has not expired yet.
        this.clusterHasActiveMaster.set(true);

        /*
         * Add a ZNode for ourselves in the backup master directory since we are
         * not the active master.
         *
         * If we become the active master later, ActiveMasterManager will delete
         * this node explicitly. If we crash before then, ZooKeeper will delete
         * this node for us since it is ephemeral.
         */
        LOG.info("Adding ZNode for " + backupZNode
            + " in backup master directory");
        MasterAddressTracker.setMasterAddress(this.watcher, backupZNode,
            this.sn);

        String msg;
        byte[] bytes = ZKUtil.getDataAndWatch(this.watcher,
            this.watcher.getMasterAddressZNode());
        if (bytes == null) {
          msg = ("A master was detected, but went down before its address "
              + "could be read.  Attempting to become the next active master");
        } else {
          ServerName currentMaster;
          try {
            currentMaster = ServerName.parseFrom(bytes);
          } catch (DeserializationException e) {
            LOG.warn("Failed parse", e);
            // Hopefully next time around we won't fail the parse. Dangerous.
            continue;
          }
          if (ServerName.isSameHostnameAndPort(currentMaster, this.sn)) {
            msg = ("Current master has this master's address, " + currentMaster + "; master was restarted? Deleting node.");
            // Hurry along the expiration of the znode.
            ZKUtil.deleteNode(this.watcher,
                this.watcher.getMasterAddressZNode());

            // We may have failed to delete the znode at the previous step, but
            // we delete the file anyway: a second attempt to delete the znode
            // is likely to fail again.
            ZNodeClearer.deleteMyEphemeralNodeOnDisk();
          } else {
            msg = "Another master is the active master, " + currentMaster
                + "; waiting to become the next active master";
          }
        }
        LOG.info(msg);
        startupStatus.setStatus(msg);
      } catch (KeeperException ke) {
        master.abort("Received an unexpected KeeperException, aborting", ke);
        return false;
      }
      synchronized (this.clusterHasActiveMaster) {
        while (this.clusterHasActiveMaster.get() && !this.master.isStopped()) {
          try {
            this.clusterHasActiveMaster.wait();
          } catch (InterruptedException e) {
            // We expect to be interrupted when a master dies,
            // will fall out if so
            LOG.debug("Interrupted waiting for master to die", e);
          }
        }
        if (!clusterStatusTracker.isClusterUp()) {
          this.master
              .stop("Cluster went down before this master became active");
        }
        if (this.master.isStopped()) {
          return false;
        }
        // there is no active master so we can try to become active master again
      }
    }
  }

  /**
   * @return True if cluster has an active master.
   */
  public boolean isActiveMaster() {
    try {
      if (ZKUtil.checkExists(watcher, watcher.getMasterAddressZNode()) >= 0) {
        return true;
      }
    }
catch (KeeperException ke) {
      LOG.info("Received an unexpected KeeperException when checking "
          + "isActiveMaster : " + ke);
    }
    return false;
  }

  public void stop() {
    try {
      // If our address is in ZK, delete it on our way out
      ServerName activeMaster = null;
      try {
        activeMaster = MasterAddressTracker.getMasterAddress(this.watcher);
      } catch (IOException e) {
        LOG.warn("Failed get of master address: " + e.toString());
      }
      if (activeMaster != null && activeMaster.equals(this.sn)) {
        ZKUtil.deleteNode(watcher, watcher.getMasterAddressZNode());
        // We may have failed to delete the znode at the previous step, but
        // we delete the file anyway: a second attempt to delete the znode is
        // likely to fail again.
        ZNodeClearer.deleteMyEphemeralNodeOnDisk();
      }
    } catch (KeeperException e) {
      LOG.error(
          this.watcher.prefix("Error deleting our own master address node"), e);
    }
  }
}
TOP

Related Classes of com.alibaba.wasp.master.ActiveMasterManager

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.