opendaylight/md-sal/sal-akka-raft/src/main/java/org/opendaylight/controller/cluster/raft/RaftActor.java

   1 /*
   2  * Copyright (c) 2014 Cisco Systems, Inc. and others.  All rights reserved.
   3  *
   4  * This program and the accompanying materials are made available under the
   5  * terms of the Eclipse Public License v1.0 which accompanies this distribution,
   6  * and is available at http://www.eclipse.org/legal/epl-v10.html
   7  */
   8
   9 package org.opendaylight.controller.cluster.raft;
  10
  11 import akka.actor.ActorRef;
  12 import akka.actor.ActorSelection;
  13 import akka.japi.Procedure;
  14 import akka.persistence.SaveSnapshotFailure;
  15 import akka.persistence.SaveSnapshotSuccess;
  16 import akka.persistence.SnapshotSelectionCriteria;
  17 import com.google.common.annotations.VisibleForTesting;
  18 import com.google.common.base.Objects;
  19 import com.google.common.base.Optional;
  20 import com.google.common.collect.ImmutableMap;
  21 import com.google.common.collect.Lists;
  22 import java.io.Serializable;
  23 import java.util.Collection;
  24 import java.util.List;
  25 import java.util.Map;
  26 import java.util.concurrent.TimeUnit;
  27 import javax.annotation.Nonnull;
  28 import org.apache.commons.lang3.time.DurationFormatUtils;
  29 import org.opendaylight.controller.cluster.DataPersistenceProvider;
  30 import org.opendaylight.controller.cluster.DelegatingPersistentDataProvider;
  31 import org.opendaylight.controller.cluster.NonPersistentDataProvider;
  32 import org.opendaylight.controller.cluster.PersistentDataProvider;
  33 import org.opendaylight.controller.cluster.common.actor.AbstractUntypedPersistentActor;
  34 import org.opendaylight.controller.cluster.notifications.LeaderStateChanged;
  35 import org.opendaylight.controller.cluster.notifications.RoleChanged;
  36 import org.opendaylight.controller.cluster.raft.base.messages.ApplyJournalEntries;
  37 import org.opendaylight.controller.cluster.raft.base.messages.ApplySnapshot;
  38 import org.opendaylight.controller.cluster.raft.base.messages.ApplyState;
  39 import org.opendaylight.controller.cluster.raft.base.messages.CaptureSnapshot;
  40 import org.opendaylight.controller.cluster.raft.base.messages.CaptureSnapshotReply;
  41 import org.opendaylight.controller.cluster.raft.base.messages.Replicate;
  42 import org.opendaylight.controller.cluster.raft.behaviors.AbstractLeader;
  43 import org.opendaylight.controller.cluster.raft.behaviors.DelegatingRaftActorBehavior;
  44 import org.opendaylight.controller.cluster.raft.behaviors.Follower;
  45 import org.opendaylight.controller.cluster.raft.behaviors.RaftActorBehavior;
  46 import org.opendaylight.controller.cluster.raft.client.messages.FindLeader;
  47 import org.opendaylight.controller.cluster.raft.client.messages.FindLeaderReply;
  48 import org.opendaylight.controller.cluster.raft.client.messages.FollowerInfo;
  49 import org.opendaylight.controller.cluster.raft.client.messages.GetOnDemandRaftState;
  50 import org.opendaylight.controller.cluster.raft.client.messages.OnDemandRaftState;
  51 import org.opendaylight.controller.cluster.raft.protobuff.client.messages.Payload;
  52 import org.slf4j.Logger;
  53 import org.slf4j.LoggerFactory;
  54
  55 /**
  56  * RaftActor encapsulates a state machine that needs to be kept synchronized
  57  * in a cluster. It implements the RAFT algorithm as described in the paper
  58  * <a href='https://ramcloud.stanford.edu/wiki/download/attachments/11370504/raft.pdf'>
  59  * In Search of an Understandable Consensus Algorithm</a>
  60  * <p/>
  61  * RaftActor has 3 states and each state has a certain behavior associated
  62  * with it. A Raft actor can behave as,
  63  * <ul>
  64  * <li> A Leader </li>
  65  * <li> A Follower (or) </li>
  66  * <li> A Candidate </li>
  67  * </ul>
  68  * <p/>
  69  * <p/>
  70  * A RaftActor MUST be a Leader in order to accept requests from clients to
  71  * change the state of it's encapsulated state machine. Once a RaftActor becomes
  72  * a Leader it is also responsible for ensuring that all followers ultimately
  73  * have the same log and therefore the same state machine as itself.
  74  * <p/>
  75  * <p/>
  76  * The current behavior of a RaftActor determines how election for leadership
  77  * is initiated and how peer RaftActors react to request for votes.
  78  * <p/>
  79  * <p/>
  80  * Each RaftActor also needs to know the current election term. It uses this
  81  * information for a couple of things. One is to simply figure out who it
  82  * voted for in the last election. Another is to figure out if the message
  83  * it received to update it's state is stale.
  84  * <p/>
  85  * <p/>
  86  * The RaftActor uses akka-persistence to store it's replicated log.
  87  * Furthermore through it's behaviors a Raft Actor determines
  88  * <p/>
  89  * <ul>
  90  * <li> when a log entry should be persisted </li>
  91  * <li> when a log entry should be applied to the state machine (and) </li>
  92  * <li> when a snapshot should be saved </li>
  93  * </ul>
  94  */
  95 public abstract class RaftActor extends AbstractUntypedPersistentActor {
  96
  97     private static final long APPLY_STATE_DELAY_THRESHOLD_IN_NANOS = TimeUnit.MILLISECONDS.toNanos(50L); // 50 millis
  98
  99     private static final String COMMIT_SNAPSHOT = "commit_snapshot";
 100
 101     protected final Logger LOG = LoggerFactory.getLogger(getClass());
 102
 103     /**
 104      * The current state determines the current behavior of a RaftActor
 105      * A Raft Actor always starts off in the Follower State
 106      */
 107     private final DelegatingRaftActorBehavior currentBehavior = new DelegatingRaftActorBehavior();
 108
 109     /**
 110      * This context should NOT be passed directly to any other actor it is
 111      * only to be consumed by the RaftActorBehaviors
 112      */
 113     private final RaftActorContextImpl context;
 114
 115     private final DelegatingPersistentDataProvider delegatingPersistenceProvider = new DelegatingPersistentDataProvider(null);
 116
 117     private final Procedure<Void> createSnapshotProcedure = new CreateSnapshotProcedure();
 118
 119     private RaftActorRecoverySupport raftRecovery;
 120
 121     private final BehaviorStateHolder reusableBehaviorStateHolder = new BehaviorStateHolder();
 122
 123     public RaftActor(String id, Map<String, String> peerAddresses) {
 124         this(id, peerAddresses, Optional.<ConfigParams>absent());
 125     }
 126
 127     public RaftActor(String id, Map<String, String> peerAddresses,
 128          Optional<ConfigParams> configParams) {
 129
 130         context = new RaftActorContextImpl(this.getSelf(),
 131             this.getContext(), id, new ElectionTermImpl(delegatingPersistenceProvider, id, LOG),
 132             -1, -1, peerAddresses,
 133             (configParams.isPresent() ? configParams.get(): new DefaultConfigParamsImpl()), LOG);
 134
 135         context.setReplicatedLog(ReplicatedLogImpl.newInstance(context, delegatingPersistenceProvider, currentBehavior));
 136     }
 137
 138     @Override
 139     public void preStart() throws Exception {
 140         LOG.info("Starting recovery for {} with journal batch size {}", persistenceId(),
 141                 context.getConfigParams().getJournalRecoveryLogBatchSize());
 142
 143         super.preStart();
 144     }
 145
 146     @Override
 147     public void postStop() {
 148         if(currentBehavior != null) {
 149             try {
 150                 currentBehavior.close();
 151             } catch (Exception e) {
 152                 LOG.debug("{}: Error closing behavior {}", persistenceId(), currentBehavior.state());
 153             }
 154         }
 155
 156         super.postStop();
 157     }
 158
 159     @Override
 160     public void handleRecover(Object message) {
 161         if(raftRecovery == null) {
 162             raftRecovery = new RaftActorRecoverySupport(delegatingPersistenceProvider, context, currentBehavior,
 163                     getRaftActorRecoveryCohort());
 164         }
 165
 166         boolean recoveryComplete = raftRecovery.handleRecoveryMessage(message);
 167         if(recoveryComplete) {
 168             if(!persistence().isRecoveryApplicable()) {
 169                 // Delete all the messages from the akka journal so that we do not end up with consistency issues
 170                 // Note I am not using the dataPersistenceProvider and directly using the akka api here
 171                 deleteMessages(lastSequenceNr());
 172
 173                 // Delete all the akka snapshots as they will not be needed
 174                 deleteSnapshots(new SnapshotSelectionCriteria(scala.Long.MaxValue(), scala.Long.MaxValue()));
 175             }
 176
 177             onRecoveryComplete();
 178
 179             initializeBehavior();
 180
 181             raftRecovery = null;
 182         }
 183     }
 184
 185     protected void initializeBehavior(){
 186         changeCurrentBehavior(new Follower(context));
 187     }
 188
 189     protected void changeCurrentBehavior(RaftActorBehavior newBehavior){
 190         reusableBehaviorStateHolder.init(getCurrentBehavior());
 191         setCurrentBehavior(newBehavior);
 192         handleBehaviorChange(reusableBehaviorStateHolder, getCurrentBehavior());
 193     }
 194
 195     @Override public void handleCommand(Object message) {
 196         if (message instanceof ApplyState){
 197             ApplyState applyState = (ApplyState) message;
 198
 199             long elapsedTime = (System.nanoTime() - applyState.getStartTime());
 200             if(elapsedTime >= APPLY_STATE_DELAY_THRESHOLD_IN_NANOS){
 201                 LOG.warn("ApplyState took more time than expected. Elapsed Time = {} ms ApplyState = {}",
 202                         TimeUnit.NANOSECONDS.toMillis(elapsedTime), applyState);
 203             }
 204
 205             if(LOG.isDebugEnabled()) {
 206                 LOG.debug("{}: Applying state for log index {} data {}",
 207                     persistenceId(), applyState.getReplicatedLogEntry().getIndex(),
 208                     applyState.getReplicatedLogEntry().getData());
 209             }
 210
 211             applyState(applyState.getClientActor(), applyState.getIdentifier(),
 212                 applyState.getReplicatedLogEntry().getData());
 213
 214         } else if (message instanceof ApplyJournalEntries){
 215             ApplyJournalEntries applyEntries = (ApplyJournalEntries) message;
 216             if(LOG.isDebugEnabled()) {
 217                 LOG.debug("{}: Persisting ApplyLogEntries with index={}", persistenceId(), applyEntries.getToIndex());
 218             }
 219
 220             persistence().persist(applyEntries, NoopProcedure.instance());
 221
 222         } else if(message instanceof ApplySnapshot ) {
 223             Snapshot snapshot = ((ApplySnapshot) message).getSnapshot();
 224
 225             if(LOG.isDebugEnabled()) {
 226                 LOG.debug("{}: ApplySnapshot called on Follower Actor " +
 227                         "snapshotIndex:{}, snapshotTerm:{}", persistenceId(), snapshot.getLastAppliedIndex(),
 228                     snapshot.getLastAppliedTerm()
 229                 );
 230             }
 231
 232             applySnapshot(snapshot.getState());
 233
 234             //clears the followers log, sets the snapshot index to ensure adjusted-index works
 235             context.setReplicatedLog(ReplicatedLogImpl.newInstance(snapshot, context, delegatingPersistenceProvider,
 236                     currentBehavior));
 237             context.setLastApplied(snapshot.getLastAppliedIndex());
 238
 239         } else if (message instanceof FindLeader) {
 240             getSender().tell(
 241                 new FindLeaderReply(getLeaderAddress()),
 242                 getSelf()
 243             );
 244
 245         } else if (message instanceof SaveSnapshotSuccess) {
 246             SaveSnapshotSuccess success = (SaveSnapshotSuccess) message;
 247             LOG.info("{}: SaveSnapshotSuccess received for snapshot", persistenceId());
 248
 249             long sequenceNumber = success.metadata().sequenceNr();
 250
 251             commitSnapshot(sequenceNumber);
 252
 253         } else if (message instanceof SaveSnapshotFailure) {
 254             SaveSnapshotFailure saveSnapshotFailure = (SaveSnapshotFailure) message;
 255
 256             LOG.error("{}: SaveSnapshotFailure received for snapshot Cause:",
 257                     persistenceId(), saveSnapshotFailure.cause());
 258
 259             context.getSnapshotManager().rollback();
 260
 261         } else if (message instanceof CaptureSnapshot) {
 262             LOG.debug("{}: CaptureSnapshot received by actor: {}", persistenceId(), message);
 263
 264             context.getSnapshotManager().create(createSnapshotProcedure);
 265
 266         } else if (message instanceof CaptureSnapshotReply) {
 267             handleCaptureSnapshotReply(((CaptureSnapshotReply) message).getSnapshot());
 268         } else if(message instanceof GetOnDemandRaftState) {
 269             onGetOnDemandRaftStats();
 270         } else if (message.equals(COMMIT_SNAPSHOT)) {
 271             commitSnapshot(-1);
 272         } else {
 273             reusableBehaviorStateHolder.init(getCurrentBehavior());
 274
 275             setCurrentBehavior(currentBehavior.handleMessage(getSender(), message));
 276
 277             handleBehaviorChange(reusableBehaviorStateHolder, getCurrentBehavior());
 278         }
 279     }
 280
 281     private void onGetOnDemandRaftStats() {
 282         // Debugging message to retrieve raft stats.
 283
 284         OnDemandRaftState.Builder builder = OnDemandRaftState.builder()
 285                 .commitIndex(context.getCommitIndex())
 286                 .currentTerm(context.getTermInformation().getCurrentTerm())
 287                 .inMemoryJournalDataSize(replicatedLog().dataSize())
 288                 .inMemoryJournalLogSize(replicatedLog().size())
 289                 .isSnapshotCaptureInitiated(context.getSnapshotManager().isCapturing())
 290                 .lastApplied(context.getLastApplied())
 291                 .lastIndex(replicatedLog().lastIndex())
 292                 .lastTerm(replicatedLog().lastTerm())
 293                 .leader(getLeaderId())
 294                 .raftState(currentBehavior.state().toString())
 295                 .replicatedToAllIndex(currentBehavior.getReplicatedToAllIndex())
 296                 .snapshotIndex(replicatedLog().getSnapshotIndex())
 297                 .snapshotTerm(replicatedLog().getSnapshotTerm())
 298                 .votedFor(context.getTermInformation().getVotedFor())
 299                 .peerAddresses(ImmutableMap.copyOf(context.getPeerAddresses()));
 300
 301         ReplicatedLogEntry lastLogEntry = getLastLogEntry();
 302         if (lastLogEntry != null) {
 303             builder.lastLogIndex(lastLogEntry.getIndex());
 304             builder.lastLogTerm(lastLogEntry.getTerm());
 305         }
 306
 307         if(getCurrentBehavior() instanceof AbstractLeader) {
 308             AbstractLeader leader = (AbstractLeader)getCurrentBehavior();
 309             Collection<String> followerIds = leader.getFollowerIds();
 310             List<FollowerInfo> followerInfoList = Lists.newArrayListWithCapacity(followerIds.size());
 311             for(String id: followerIds) {
 312                 final FollowerLogInformation info = leader.getFollower(id);
 313                 followerInfoList.add(new FollowerInfo(id, info.getNextIndex(), info.getMatchIndex(),
 314                         info.isFollowerActive(), DurationFormatUtils.formatDurationHMS(info.timeSinceLastActivity())));
 315             }
 316
 317             builder.followerInfoList(followerInfoList);
 318         }
 319
 320         sender().tell(builder.build(), self());
 321
 322     }
 323
 324     private void handleBehaviorChange(BehaviorStateHolder oldBehaviorState, RaftActorBehavior currentBehavior) {
 325         RaftActorBehavior oldBehavior = oldBehaviorState.getBehavior();
 326
 327         if (oldBehavior != currentBehavior){
 328             onStateChanged();
 329         }
 330
 331         String oldBehaviorLeaderId = oldBehavior == null ? null : oldBehaviorState.getLeaderId();
 332         String oldBehaviorStateName = oldBehavior == null ? null : oldBehavior.state().name();
 333
 334         // it can happen that the state has not changed but the leader has changed.
 335         Optional<ActorRef> roleChangeNotifier = getRoleChangeNotifier();
 336         if(!Objects.equal(oldBehaviorLeaderId, currentBehavior.getLeaderId())) {
 337             if(roleChangeNotifier.isPresent()) {
 338                 roleChangeNotifier.get().tell(new LeaderStateChanged(getId(), currentBehavior.getLeaderId()), getSelf());
 339             }
 340
 341             onLeaderChanged(oldBehaviorLeaderId, currentBehavior.getLeaderId());
 342         }
 343
 344         if (roleChangeNotifier.isPresent() &&
 345                 (oldBehavior == null || (oldBehavior.state() != currentBehavior.state()))) {
 346             roleChangeNotifier.get().tell(new RoleChanged(getId(), oldBehaviorStateName ,
 347                     currentBehavior.state().name()), getSelf());
 348         }
 349     }
 350
 351     /**
 352      * When a derived RaftActor needs to persist something it must call
 353      * persistData.
 354      *
 355      * @param clientActor
 356      * @param identifier
 357      * @param data
 358      */
 359     protected void persistData(final ActorRef clientActor, final String identifier,
 360         final Payload data) {
 361
 362         ReplicatedLogEntry replicatedLogEntry = new ReplicatedLogImplEntry(
 363             context.getReplicatedLog().lastIndex() + 1,
 364             context.getTermInformation().getCurrentTerm(), data);
 365
 366         if(LOG.isDebugEnabled()) {
 367             LOG.debug("{}: Persist data {}", persistenceId(), replicatedLogEntry);
 368         }
 369
 370         final RaftActorContext raftContext = getRaftActorContext();
 371
 372         replicatedLog().appendAndPersist(replicatedLogEntry, new Procedure<ReplicatedLogEntry>() {
 373             @Override
 374             public void apply(ReplicatedLogEntry replicatedLogEntry) throws Exception {
 375                 if(!hasFollowers()){
 376                     // Increment the Commit Index and the Last Applied values
 377                     raftContext.setCommitIndex(replicatedLogEntry.getIndex());
 378                     raftContext.setLastApplied(replicatedLogEntry.getIndex());
 379
 380                     // Apply the state immediately
 381                     applyState(clientActor, identifier, data);
 382
 383                     // Send a ApplyJournalEntries message so that we write the fact that we applied
 384                     // the state to durable storage
 385                     self().tell(new ApplyJournalEntries(replicatedLogEntry.getIndex()), self());
 386
 387                     context.getSnapshotManager().trimLog(context.getLastApplied(), currentBehavior);
 388
 389                 } else if (clientActor != null) {
 390                     // Send message for replication
 391                     currentBehavior.handleMessage(getSelf(),
 392                             new Replicate(clientActor, identifier, replicatedLogEntry));
 393                 }
 394             }
 395         });
 396     }
 397
 398     private ReplicatedLog replicatedLog() {
 399         return context.getReplicatedLog();
 400     }
 401
 402     protected String getId() {
 403         return context.getId();
 404     }
 405
 406     @VisibleForTesting
 407     void setCurrentBehavior(RaftActorBehavior behavior) {
 408         currentBehavior.setDelegate(behavior);
 409     }
 410
 411     protected RaftActorBehavior getCurrentBehavior() {
 412         return currentBehavior.getDelegate();
 413     }
 414
 415     /**
 416      * Derived actors can call the isLeader method to check if the current
 417      * RaftActor is the Leader or not
 418      *
 419      * @return true it this RaftActor is a Leader false otherwise
 420      */
 421     protected boolean isLeader() {
 422         return context.getId().equals(currentBehavior.getLeaderId());
 423     }
 424
 425     /**
 426      * Derived actor can call getLeader if they need a reference to the Leader.
 427      * This would be useful for example in forwarding a request to an actor
 428      * which is the leader
 429      *
 430      * @return A reference to the leader if known, null otherwise
 431      */
 432     protected ActorSelection getLeader(){
 433         String leaderAddress = getLeaderAddress();
 434
 435         if(leaderAddress == null){
 436             return null;
 437         }
 438
 439         return context.actorSelection(leaderAddress);
 440     }
 441
 442     /**
 443      *
 444      * @return the current leader's id
 445      */
 446     protected String getLeaderId(){
 447         return currentBehavior.getLeaderId();
 448     }
 449
 450     protected RaftState getRaftState() {
 451         return currentBehavior.state();
 452     }
 453
 454     protected ReplicatedLogEntry getLastLogEntry() {
 455         return replicatedLog().last();
 456     }
 457
 458     protected Long getCurrentTerm(){
 459         return context.getTermInformation().getCurrentTerm();
 460     }
 461
 462     protected Long getCommitIndex(){
 463         return context.getCommitIndex();
 464     }
 465
 466     protected Long getLastApplied(){
 467         return context.getLastApplied();
 468     }
 469
 470     protected RaftActorContext getRaftActorContext() {
 471         return context;
 472     }
 473
 474     protected void updateConfigParams(ConfigParams configParams) {
 475         context.setConfigParams(configParams);
 476     }
 477
 478     public final DataPersistenceProvider persistence() {
 479         return delegatingPersistenceProvider.getDelegate();
 480     }
 481
 482     public void setPersistence(DataPersistenceProvider provider) {
 483         delegatingPersistenceProvider.setDelegate(provider);
 484     }
 485
 486     protected void setPersistence(boolean persistent) {
 487         if(persistent) {
 488             setPersistence(new PersistentDataProvider(this));
 489         } else {
 490             setPersistence(new NonPersistentDataProvider() {
 491                 /**
 492                  * The way snapshotting works is,
 493                  * <ol>
 494                  * <li> RaftActor calls createSnapshot on the Shard
 495                  * <li> Shard sends a CaptureSnapshotReply and RaftActor then calls saveSnapshot
 496                  * <li> When saveSnapshot is invoked on the akka-persistence API it uses the SnapshotStore to save
 497                  * the snapshot. The SnapshotStore sends SaveSnapshotSuccess or SaveSnapshotFailure. When the
 498                  * RaftActor gets SaveSnapshot success it commits the snapshot to the in-memory journal. This
 499                  * commitSnapshot is mimicking what is done in SaveSnapshotSuccess.
 500                  * </ol>
 501                  */
 502                 @Override
 503                 public void saveSnapshot(Object o) {
 504                     // Make saving Snapshot successful
 505                     // Committing the snapshot here would end up calling commit in the creating state which would
 506                     // be a state violation. That's why now we send a message to commit the snapshot.
 507                     self().tell(COMMIT_SNAPSHOT, self());
 508                 }
 509             });
 510         }
 511     }
 512
 513     /**
 514      * setPeerAddress sets the address of a known peer at a later time.
 515      * <p>
 516      * This is to account for situations where a we know that a peer
 517      * exists but we do not know an address up-front. This may also be used in
 518      * situations where a known peer starts off in a different location and we
 519      * need to change it's address
 520      * <p>
 521      * Note that if the peerId does not match the list of peers passed to
 522      * this actor during construction an IllegalStateException will be thrown.
 523      *
 524      * @param peerId
 525      * @param peerAddress
 526      */
 527     protected void setPeerAddress(String peerId, String peerAddress){
 528         context.setPeerAddress(peerId, peerAddress);
 529     }
 530
 531     protected void commitSnapshot(long sequenceNumber) {
 532         context.getSnapshotManager().commit(persistence(), sequenceNumber);
 533     }
 534
 535     /**
 536      * The applyState method will be called by the RaftActor when some data
 537      * needs to be applied to the actor's state
 538      *
 539      * @param clientActor A reference to the client who sent this message. This
 540      *                    is the same reference that was passed to persistData
 541      *                    by the derived actor. clientActor may be null when
 542      *                    the RaftActor is behaving as a follower or during
 543      *                    recovery.
 544      * @param identifier  The identifier of the persisted data. This is also
 545      *                    the same identifier that was passed to persistData by
 546      *                    the derived actor. identifier may be null when
 547      *                    the RaftActor is behaving as a follower or during
 548      *                    recovery
 549      * @param data        A piece of data that was persisted by the persistData call.
 550      *                    This should NEVER be null.
 551      */
 552     protected abstract void applyState(ActorRef clientActor, String identifier,
 553         Object data);
 554
 555     /**
 556      * Returns the RaftActorRecoveryCohort to participate in persistence recovery.
 557      */
 558     @Nonnull
 559     protected abstract RaftActorRecoveryCohort getRaftActorRecoveryCohort();
 560
 561     /**
 562      * This method is called when recovery is complete.
 563      */
 564     protected abstract void onRecoveryComplete();
 565
 566     /**
 567      * This method will be called by the RaftActor when a snapshot needs to be
 568      * created. The derived actor should respond with its current state.
 569      * <p/>
 570      * During recovery the state that is returned by the derived actor will
 571      * be passed back to it by calling the applySnapshot  method
 572      *
 573      * @return The current state of the actor
 574      */
 575     protected abstract void createSnapshot();
 576
 577     /**
 578      * This method can be called at any other point during normal
 579      * operations when the derived actor is out of sync with it's peers
 580      * and the only way to bring it in sync is by applying a snapshot
 581      *
 582      * @param snapshotBytes A snapshot of the state of the actor
 583      */
 584     protected abstract void applySnapshot(byte[] snapshotBytes);
 585
 586     /**
 587      * This method will be called by the RaftActor when the state of the
 588      * RaftActor changes. The derived actor can then use methods like
 589      * isLeader or getLeader to do something useful
 590      */
 591     protected abstract void onStateChanged();
 592
 593     /**
 594      * Notifier Actor for this RaftActor to notify when a role change happens
 595      * @return ActorRef - ActorRef of the notifier or Optional.absent if none.
 596      */
 597     protected abstract Optional<ActorRef> getRoleChangeNotifier();
 598
 599     protected void onLeaderChanged(String oldLeader, String newLeader){};
 600
 601     private String getLeaderAddress(){
 602         if(isLeader()){
 603             return getSelf().path().toString();
 604         }
 605         String leaderId = currentBehavior.getLeaderId();
 606         if (leaderId == null) {
 607             return null;
 608         }
 609         String peerAddress = context.getPeerAddress(leaderId);
 610         if(LOG.isDebugEnabled()) {
 611             LOG.debug("{}: getLeaderAddress leaderId = {} peerAddress = {}",
 612                     persistenceId(), leaderId, peerAddress);
 613         }
 614
 615         return peerAddress;
 616     }
 617
 618     private void handleCaptureSnapshotReply(byte[] snapshotBytes) {
 619         LOG.debug("{}: CaptureSnapshotReply received by actor: snapshot size {}", persistenceId(), snapshotBytes.length);
 620
 621         context.getSnapshotManager().persist(persistence(), snapshotBytes, currentBehavior, context.getTotalMemory());
 622     }
 623
 624     protected boolean hasFollowers(){
 625         return getRaftActorContext().hasFollowers();
 626     }
 627
 628     static class DeleteEntries implements Serializable {
 629         private static final long serialVersionUID = 1L;
 630         private final int fromIndex;
 631
 632         public DeleteEntries(int fromIndex) {
 633             this.fromIndex = fromIndex;
 634         }
 635
 636         public int getFromIndex() {
 637             return fromIndex;
 638         }
 639     }
 640
 641     static class UpdateElectionTerm implements Serializable {
 642         private static final long serialVersionUID = 1L;
 643         private final long currentTerm;
 644         private final String votedFor;
 645
 646         public UpdateElectionTerm(long currentTerm, String votedFor) {
 647             this.currentTerm = currentTerm;
 648             this.votedFor = votedFor;
 649         }
 650
 651         public long getCurrentTerm() {
 652             return currentTerm;
 653         }
 654
 655         public String getVotedFor() {
 656             return votedFor;
 657         }
 658     }
 659
 660     private class CreateSnapshotProcedure implements Procedure<Void> {
 661
 662         @Override
 663         public void apply(Void aVoid) throws Exception {
 664             createSnapshot();
 665         }
 666     }
 667
 668     private static class BehaviorStateHolder {
 669         private RaftActorBehavior behavior;
 670         private String leaderId;
 671
 672         void init(RaftActorBehavior behavior) {
 673             this.behavior = behavior;
 674             this.leaderId = behavior != null ? behavior.getLeaderId() : null;
 675         }
 676
 677         RaftActorBehavior getBehavior() {
 678             return behavior;
 679         }
 680
 681         String getLeaderId() {
 682             return leaderId;
 683         }
 684     }
 685 }