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.event.Logging;
  14 import akka.event.LoggingAdapter;
  15 import akka.japi.Procedure;
  16 import akka.persistence.RecoveryCompleted;
  17 import akka.persistence.SaveSnapshotFailure;
  18 import akka.persistence.SaveSnapshotSuccess;
  19 import akka.persistence.SnapshotOffer;
  20 import akka.persistence.SnapshotSelectionCriteria;
  21 import com.google.common.annotations.VisibleForTesting;
  22 import com.google.common.base.Optional;
  23 import com.google.common.base.Stopwatch;
  24 import com.google.protobuf.ByteString;
  25 import org.opendaylight.controller.cluster.DataPersistenceProvider;
  26 import org.opendaylight.controller.cluster.common.actor.AbstractUntypedPersistentActor;
  27 import org.opendaylight.controller.cluster.raft.base.messages.ApplyLogEntries;
  28 import org.opendaylight.controller.cluster.raft.base.messages.ApplySnapshot;
  29 import org.opendaylight.controller.cluster.raft.base.messages.ApplyState;
  30 import org.opendaylight.controller.cluster.raft.base.messages.CaptureSnapshot;
  31 import org.opendaylight.controller.cluster.raft.base.messages.CaptureSnapshotReply;
  32 import org.opendaylight.controller.cluster.raft.base.messages.Replicate;
  33 import org.opendaylight.controller.cluster.raft.base.messages.SendHeartBeat;
  34 import org.opendaylight.controller.cluster.raft.base.messages.SendInstallSnapshot;
  35 import org.opendaylight.controller.cluster.raft.behaviors.AbstractRaftActorBehavior;
  36 import org.opendaylight.controller.cluster.raft.behaviors.Follower;
  37 import org.opendaylight.controller.cluster.raft.behaviors.RaftActorBehavior;
  38 import org.opendaylight.controller.cluster.raft.client.messages.AddRaftPeer;
  39 import org.opendaylight.controller.cluster.raft.client.messages.FindLeader;
  40 import org.opendaylight.controller.cluster.raft.client.messages.FindLeaderReply;
  41 import org.opendaylight.controller.cluster.raft.client.messages.RemoveRaftPeer;
  42 import org.opendaylight.controller.cluster.raft.messages.AppendEntriesReply;
  43 import org.opendaylight.controller.cluster.raft.protobuff.client.messages.Payload;
  44 import org.opendaylight.controller.protobuff.messages.cluster.raft.AppendEntriesMessages;
  45
  46 import java.io.Serializable;
  47 import java.util.Map;
  48
  49 /**
  50  * RaftActor encapsulates a state machine that needs to be kept synchronized
  51  * in a cluster. It implements the RAFT algorithm as described in the paper
  52  * <a href='https://ramcloud.stanford.edu/wiki/download/attachments/11370504/raft.pdf'>
  53  * In Search of an Understandable Consensus Algorithm</a>
  54  * <p/>
  55  * RaftActor has 3 states and each state has a certain behavior associated
  56  * with it. A Raft actor can behave as,
  57  * <ul>
  58  * <li> A Leader </li>
  59  * <li> A Follower (or) </li>
  60  * <li> A Candidate </li>
  61  * </ul>
  62  * <p/>
  63  * <p/>
  64  * A RaftActor MUST be a Leader in order to accept requests from clients to
  65  * change the state of it's encapsulated state machine. Once a RaftActor becomes
  66  * a Leader it is also responsible for ensuring that all followers ultimately
  67  * have the same log and therefore the same state machine as itself.
  68  * <p/>
  69  * <p/>
  70  * The current behavior of a RaftActor determines how election for leadership
  71  * is initiated and how peer RaftActors react to request for votes.
  72  * <p/>
  73  * <p/>
  74  * Each RaftActor also needs to know the current election term. It uses this
  75  * information for a couple of things. One is to simply figure out who it
  76  * voted for in the last election. Another is to figure out if the message
  77  * it received to update it's state is stale.
  78  * <p/>
  79  * <p/>
  80  * The RaftActor uses akka-persistence to store it's replicated log.
  81  * Furthermore through it's behaviors a Raft Actor determines
  82  * <p/>
  83  * <ul>
  84  * <li> when a log entry should be persisted </li>
  85  * <li> when a log entry should be applied to the state machine (and) </li>
  86  * <li> when a snapshot should be saved </li>
  87  * </ul>
  88  */
  89 public abstract class RaftActor extends AbstractUntypedPersistentActor {
  90     protected final LoggingAdapter LOG =
  91         Logging.getLogger(getContext().system(), this);
  92
  93     /**
  94      * The current state determines the current behavior of a RaftActor
  95      * A Raft Actor always starts off in the Follower State
  96      */
  97     private RaftActorBehavior currentBehavior;
  98
  99     /**
 100      * This context should NOT be passed directly to any other actor it is
 101      * only to be consumed by the RaftActorBehaviors
 102      */
 103     private final RaftActorContext context;
 104
 105     /**
 106      * The in-memory journal
 107      */
 108     private ReplicatedLogImpl replicatedLog = new ReplicatedLogImpl();
 109
 110     private CaptureSnapshot captureSnapshot = null;
 111
 112     private volatile boolean hasSnapshotCaptureInitiated = false;
 113
 114     private Stopwatch recoveryTimer;
 115
 116     private int currentRecoveryBatchCount;
 117
 118     public RaftActor(String id, Map<String, String> peerAddresses) {
 119         this(id, peerAddresses, Optional.<ConfigParams>absent());
 120     }
 121
 122     public RaftActor(String id, Map<String, String> peerAddresses,
 123          Optional<ConfigParams> configParams) {
 124
 125         context = new RaftActorContextImpl(this.getSelf(),
 126             this.getContext(), id, new ElectionTermImpl(),
 127             -1, -1, replicatedLog, peerAddresses,
 128             (configParams.isPresent() ? configParams.get(): new DefaultConfigParamsImpl()),
 129             LOG);
 130     }
 131
 132     private void initRecoveryTimer() {
 133         if(recoveryTimer == null) {
 134             recoveryTimer = new Stopwatch();
 135             recoveryTimer.start();
 136         }
 137     }
 138
 139     @Override
 140     public void preStart() throws Exception {
 141         LOG.info("Starting recovery for {} with journal batch size {}", persistenceId(),
 142                 context.getConfigParams().getJournalRecoveryLogBatchSize());
 143
 144         super.preStart();
 145     }
 146
 147     @Override
 148     public void handleRecover(Object message) {
 149         if(persistence().isRecoveryApplicable()) {
 150             if (message instanceof SnapshotOffer) {
 151                 onRecoveredSnapshot((SnapshotOffer) message);
 152             } else if (message instanceof ReplicatedLogEntry) {
 153                 onRecoveredJournalLogEntry((ReplicatedLogEntry) message);
 154             } else if (message instanceof ApplyLogEntries) {
 155                 onRecoveredApplyLogEntries((ApplyLogEntries) message);
 156             } else if (message instanceof DeleteEntries) {
 157                 replicatedLog.removeFrom(((DeleteEntries) message).getFromIndex());
 158             } else if (message instanceof UpdateElectionTerm) {
 159                 context.getTermInformation().update(((UpdateElectionTerm) message).getCurrentTerm(),
 160                         ((UpdateElectionTerm) message).getVotedFor());
 161             } else if (message instanceof RecoveryCompleted) {
 162                 onRecoveryCompletedMessage();
 163             }
 164         } else {
 165             if (message instanceof RecoveryCompleted) {
 166                 // Delete all the messages from the akka journal so that we do not end up with consistency issues
 167                 // Note I am not using the dataPersistenceProvider and directly using the akka api here
 168                 deleteMessages(lastSequenceNr());
 169
 170                 // Delete all the akka snapshots as they will not be needed
 171                 deleteSnapshots(new SnapshotSelectionCriteria(scala.Long.MaxValue(), scala.Long.MaxValue()));
 172
 173                 onRecoveryComplete();
 174                 currentBehavior = new Follower(context);
 175                 onStateChanged();
 176             }
 177         }
 178     }
 179
 180     private void onRecoveredSnapshot(SnapshotOffer offer) {
 181         if(LOG.isDebugEnabled()) {
 182             LOG.debug("SnapshotOffer called..");
 183         }
 184
 185         initRecoveryTimer();
 186
 187         Snapshot snapshot = (Snapshot) offer.snapshot();
 188
 189         // Create a replicated log with the snapshot information
 190         // The replicated log can be used later on to retrieve this snapshot
 191         // when we need to install it on a peer
 192         replicatedLog = new ReplicatedLogImpl(snapshot);
 193
 194         context.setReplicatedLog(replicatedLog);
 195         context.setLastApplied(snapshot.getLastAppliedIndex());
 196         context.setCommitIndex(snapshot.getLastAppliedIndex());
 197
 198         Stopwatch timer = new Stopwatch();
 199         timer.start();
 200
 201         // Apply the snapshot to the actors state
 202         applyRecoverySnapshot(ByteString.copyFrom(snapshot.getState()));
 203
 204         timer.stop();
 205         LOG.info("Recovery snapshot applied for {} in {}: snapshotIndex={}, snapshotTerm={}, journal-size=" +
 206                 replicatedLog.size(), persistenceId(), timer.toString(),
 207                 replicatedLog.snapshotIndex, replicatedLog.snapshotTerm);
 208     }
 209
 210     private void onRecoveredJournalLogEntry(ReplicatedLogEntry logEntry) {
 211         if(LOG.isDebugEnabled()) {
 212             LOG.debug("Received ReplicatedLogEntry for recovery: {}", logEntry.getIndex());
 213         }
 214
 215         replicatedLog.append(logEntry);
 216     }
 217
 218     private void onRecoveredApplyLogEntries(ApplyLogEntries ale) {
 219         if(LOG.isDebugEnabled()) {
 220             LOG.debug("Received ApplyLogEntries for recovery, applying to state: {} to {}",
 221                     context.getLastApplied() + 1, ale.getToIndex());
 222         }
 223
 224         for (long i = context.getLastApplied() + 1; i <= ale.getToIndex(); i++) {
 225             batchRecoveredLogEntry(replicatedLog.get(i));
 226         }
 227
 228         context.setLastApplied(ale.getToIndex());
 229         context.setCommitIndex(ale.getToIndex());
 230     }
 231
 232     private void batchRecoveredLogEntry(ReplicatedLogEntry logEntry) {
 233         initRecoveryTimer();
 234
 235         int batchSize = context.getConfigParams().getJournalRecoveryLogBatchSize();
 236         if(currentRecoveryBatchCount == 0) {
 237             startLogRecoveryBatch(batchSize);
 238         }
 239
 240         appendRecoveredLogEntry(logEntry.getData());
 241
 242         if(++currentRecoveryBatchCount >= batchSize) {
 243             endCurrentLogRecoveryBatch();
 244         }
 245     }
 246
 247     private void endCurrentLogRecoveryBatch() {
 248         applyCurrentLogRecoveryBatch();
 249         currentRecoveryBatchCount = 0;
 250     }
 251
 252     private void onRecoveryCompletedMessage() {
 253         if(currentRecoveryBatchCount > 0) {
 254             endCurrentLogRecoveryBatch();
 255         }
 256
 257         onRecoveryComplete();
 258
 259         String recoveryTime = "";
 260         if(recoveryTimer != null) {
 261             recoveryTimer.stop();
 262             recoveryTime = " in " + recoveryTimer.toString();
 263             recoveryTimer = null;
 264         }
 265
 266         LOG.info(
 267             "Recovery completed" + recoveryTime + " - Switching actor to Follower - " +
 268                 "Persistence Id =  " + persistenceId() +
 269                 " Last index in log={}, snapshotIndex={}, snapshotTerm={}, " +
 270                 "journal-size={}",
 271             replicatedLog.lastIndex(), replicatedLog.snapshotIndex,
 272             replicatedLog.snapshotTerm, replicatedLog.size());
 273
 274         currentBehavior = new Follower(context);
 275         onStateChanged();
 276     }
 277
 278     @Override public void handleCommand(Object message) {
 279         if (message instanceof ApplyState){
 280             ApplyState applyState = (ApplyState) message;
 281
 282             if(LOG.isDebugEnabled()) {
 283                 LOG.debug("Applying state for log index {} data {}",
 284                     applyState.getReplicatedLogEntry().getIndex(),
 285                     applyState.getReplicatedLogEntry().getData());
 286             }
 287
 288             applyState(applyState.getClientActor(), applyState.getIdentifier(),
 289                 applyState.getReplicatedLogEntry().getData());
 290
 291         } else if (message instanceof ApplyLogEntries){
 292             ApplyLogEntries ale = (ApplyLogEntries) message;
 293             if(LOG.isDebugEnabled()) {
 294                 LOG.debug("Persisting ApplyLogEntries with index={}", ale.getToIndex());
 295             }
 296             persistence().persist(new ApplyLogEntries(ale.getToIndex()), new Procedure<ApplyLogEntries>() {
 297                 @Override
 298                 public void apply(ApplyLogEntries param) throws Exception {
 299                 }
 300             });
 301
 302         } else if(message instanceof ApplySnapshot ) {
 303             Snapshot snapshot = ((ApplySnapshot) message).getSnapshot();
 304
 305             if(LOG.isDebugEnabled()) {
 306                 LOG.debug("ApplySnapshot called on Follower Actor " +
 307                         "snapshotIndex:{}, snapshotTerm:{}", snapshot.getLastAppliedIndex(),
 308                     snapshot.getLastAppliedTerm()
 309                 );
 310             }
 311             applySnapshot(ByteString.copyFrom(snapshot.getState()));
 312
 313             //clears the followers log, sets the snapshot index to ensure adjusted-index works
 314             replicatedLog = new ReplicatedLogImpl(snapshot);
 315             context.setReplicatedLog(replicatedLog);
 316             context.setLastApplied(snapshot.getLastAppliedIndex());
 317
 318         } else if (message instanceof FindLeader) {
 319             getSender().tell(
 320                 new FindLeaderReply(getLeaderAddress()),
 321                 getSelf()
 322             );
 323
 324         } else if (message instanceof SaveSnapshotSuccess) {
 325             SaveSnapshotSuccess success = (SaveSnapshotSuccess) message;
 326             LOG.info("SaveSnapshotSuccess received for snapshot");
 327
 328             long sequenceNumber = success.metadata().sequenceNr();
 329
 330             commitSnapshot(sequenceNumber);
 331
 332         } else if (message instanceof SaveSnapshotFailure) {
 333             SaveSnapshotFailure saveSnapshotFailure = (SaveSnapshotFailure) message;
 334
 335             LOG.info("saveSnapshotFailure.metadata():{}", saveSnapshotFailure.metadata().toString());
 336             LOG.error(saveSnapshotFailure.cause(), "SaveSnapshotFailure received for snapshot Cause:");
 337
 338             context.getReplicatedLog().snapshotRollback();
 339
 340             LOG.info("Replicated Log rollbacked. Snapshot will be attempted in the next cycle." +
 341                 "snapshotIndex:{}, snapshotTerm:{}, log-size:{}",
 342                 context.getReplicatedLog().getSnapshotIndex(),
 343                 context.getReplicatedLog().getSnapshotTerm(),
 344                 context.getReplicatedLog().size());
 345
 346         } else if (message instanceof AddRaftPeer){
 347
 348             // FIXME : Do not add raft peers like this.
 349             // When adding a new Peer we have to ensure that the a majority of
 350             // the peers know about the new Peer. Doing it this way may cause
 351             // a situation where multiple Leaders may emerge
 352             AddRaftPeer arp = (AddRaftPeer)message;
 353            context.addToPeers(arp.getName(), arp.getAddress());
 354
 355         } else if (message instanceof RemoveRaftPeer){
 356
 357             RemoveRaftPeer rrp = (RemoveRaftPeer)message;
 358             context.removePeer(rrp.getName());
 359
 360         } else if (message instanceof CaptureSnapshot) {
 361             LOG.info("CaptureSnapshot received by actor");
 362             CaptureSnapshot cs = (CaptureSnapshot)message;
 363             captureSnapshot = cs;
 364             createSnapshot();
 365
 366         } else if (message instanceof CaptureSnapshotReply){
 367             LOG.info("CaptureSnapshotReply received by actor");
 368             CaptureSnapshotReply csr = (CaptureSnapshotReply) message;
 369
 370             ByteString stateInBytes = csr.getSnapshot();
 371             LOG.info("CaptureSnapshotReply stateInBytes size:{}", stateInBytes.size());
 372             handleCaptureSnapshotReply(stateInBytes);
 373
 374         } else {
 375             if (!(message instanceof AppendEntriesMessages.AppendEntries)
 376                 && !(message instanceof AppendEntriesReply) && !(message instanceof SendHeartBeat)) {
 377                 if(LOG.isDebugEnabled()) {
 378                     LOG.debug("onReceiveCommand: message: {}", message.getClass());
 379                 }
 380             }
 381
 382             RaftActorBehavior oldBehavior = currentBehavior;
 383             currentBehavior = currentBehavior.handleMessage(getSender(), message);
 384
 385             if(oldBehavior != currentBehavior){
 386                 onStateChanged();
 387             }
 388
 389             onLeaderChanged(oldBehavior.getLeaderId(), currentBehavior.getLeaderId());
 390         }
 391     }
 392
 393     public java.util.Set<String> getPeers() {
 394
 395         return context.getPeerAddresses().keySet();
 396     }
 397
 398     protected String getReplicatedLogState() {
 399         return "snapshotIndex=" + context.getReplicatedLog().getSnapshotIndex()
 400             + ", snapshotTerm=" + context.getReplicatedLog().getSnapshotTerm()
 401             + ", im-mem journal size=" + context.getReplicatedLog().size();
 402     }
 403
 404
 405     /**
 406      * When a derived RaftActor needs to persist something it must call
 407      * persistData.
 408      *
 409      * @param clientActor
 410      * @param identifier
 411      * @param data
 412      */
 413     protected void persistData(ActorRef clientActor, String identifier,
 414         Payload data) {
 415
 416         ReplicatedLogEntry replicatedLogEntry = new ReplicatedLogImplEntry(
 417             context.getReplicatedLog().lastIndex() + 1,
 418             context.getTermInformation().getCurrentTerm(), data);
 419
 420         if(LOG.isDebugEnabled()) {
 421             LOG.debug("Persist data {}", replicatedLogEntry);
 422         }
 423
 424         replicatedLog
 425             .appendAndPersist(clientActor, identifier, replicatedLogEntry);
 426     }
 427
 428     protected String getId() {
 429         return context.getId();
 430     }
 431
 432     /**
 433      * Derived actors can call the isLeader method to check if the current
 434      * RaftActor is the Leader or not
 435      *
 436      * @return true it this RaftActor is a Leader false otherwise
 437      */
 438     protected boolean isLeader() {
 439         return context.getId().equals(currentBehavior.getLeaderId());
 440     }
 441
 442     /**
 443      * Derived actor can call getLeader if they need a reference to the Leader.
 444      * This would be useful for example in forwarding a request to an actor
 445      * which is the leader
 446      *
 447      * @return A reference to the leader if known, null otherwise
 448      */
 449     protected ActorSelection getLeader(){
 450         String leaderAddress = getLeaderAddress();
 451
 452         if(leaderAddress == null){
 453             return null;
 454         }
 455
 456         return context.actorSelection(leaderAddress);
 457     }
 458
 459     /**
 460      *
 461      * @return the current leader's id
 462      */
 463     protected String getLeaderId(){
 464         return currentBehavior.getLeaderId();
 465     }
 466
 467     protected RaftState getRaftState() {
 468         return currentBehavior.state();
 469     }
 470
 471     protected ReplicatedLogEntry getLastLogEntry() {
 472         return replicatedLog.last();
 473     }
 474
 475     protected Long getCurrentTerm(){
 476         return context.getTermInformation().getCurrentTerm();
 477     }
 478
 479     protected Long getCommitIndex(){
 480         return context.getCommitIndex();
 481     }
 482
 483     protected Long getLastApplied(){
 484         return context.getLastApplied();
 485     }
 486
 487     protected RaftActorContext getRaftActorContext() {
 488         return context;
 489     }
 490
 491     /**
 492      * setPeerAddress sets the address of a known peer at a later time.
 493      * <p>
 494      * This is to account for situations where a we know that a peer
 495      * exists but we do not know an address up-front. This may also be used in
 496      * situations where a known peer starts off in a different location and we
 497      * need to change it's address
 498      * <p>
 499      * Note that if the peerId does not match the list of peers passed to
 500      * this actor during construction an IllegalStateException will be thrown.
 501      *
 502      * @param peerId
 503      * @param peerAddress
 504      */
 505     protected void setPeerAddress(String peerId, String peerAddress){
 506         context.setPeerAddress(peerId, peerAddress);
 507     }
 508
 509     protected void commitSnapshot(long sequenceNumber) {
 510         context.getReplicatedLog().snapshotCommit();
 511
 512         // TODO: Not sure if we want to be this aggressive with trimming stuff
 513         trimPersistentData(sequenceNumber);
 514     }
 515
 516     /**
 517      * The applyState method will be called by the RaftActor when some data
 518      * needs to be applied to the actor's state
 519      *
 520      * @param clientActor A reference to the client who sent this message. This
 521      *                    is the same reference that was passed to persistData
 522      *                    by the derived actor. clientActor may be null when
 523      *                    the RaftActor is behaving as a follower or during
 524      *                    recovery.
 525      * @param identifier  The identifier of the persisted data. This is also
 526      *                    the same identifier that was passed to persistData by
 527      *                    the derived actor. identifier may be null when
 528      *                    the RaftActor is behaving as a follower or during
 529      *                    recovery
 530      * @param data        A piece of data that was persisted by the persistData call.
 531      *                    This should NEVER be null.
 532      */
 533     protected abstract void applyState(ActorRef clientActor, String identifier,
 534         Object data);
 535
 536     /**
 537      * This method is called during recovery at the start of a batch of state entries. Derived
 538      * classes should perform any initialization needed to start a batch.
 539      */
 540     protected abstract void startLogRecoveryBatch(int maxBatchSize);
 541
 542     /**
 543      * This method is called during recovery to append state data to the current batch. This method
 544      * is called 1 or more times after {@link #startLogRecoveryBatch}.
 545      *
 546      * @param data the state data
 547      */
 548     protected abstract void appendRecoveredLogEntry(Payload data);
 549
 550     /**
 551      * This method is called during recovery to reconstruct the state of the actor.
 552      *
 553      * @param snapshot A snapshot of the state of the actor
 554      */
 555     protected abstract void applyRecoverySnapshot(ByteString snapshot);
 556
 557     /**
 558      * This method is called during recovery at the end of a batch to apply the current batched
 559      * log entries. This method is called after {@link #appendRecoveredLogEntry}.
 560      */
 561     protected abstract void applyCurrentLogRecoveryBatch();
 562
 563     /**
 564      * This method is called when recovery is complete.
 565      */
 566     protected abstract void onRecoveryComplete();
 567
 568     /**
 569      * This method will be called by the RaftActor when a snapshot needs to be
 570      * created. The derived actor should respond with its current state.
 571      * <p/>
 572      * During recovery the state that is returned by the derived actor will
 573      * be passed back to it by calling the applySnapshot  method
 574      *
 575      * @return The current state of the actor
 576      */
 577     protected abstract void createSnapshot();
 578
 579     /**
 580      * This method can be called at any other point during normal
 581      * operations when the derived actor is out of sync with it's peers
 582      * and the only way to bring it in sync is by applying a snapshot
 583      *
 584      * @param snapshot A snapshot of the state of the actor
 585      */
 586     protected abstract void applySnapshot(ByteString snapshot);
 587
 588     /**
 589      * This method will be called by the RaftActor when the state of the
 590      * RaftActor changes. The derived actor can then use methods like
 591      * isLeader or getLeader to do something useful
 592      */
 593     protected abstract void onStateChanged();
 594
 595     protected abstract DataPersistenceProvider persistence();
 596
 597     protected void onLeaderChanged(String oldLeader, String newLeader){};
 598
 599     private void trimPersistentData(long sequenceNumber) {
 600         // Trim akka snapshots
 601         // FIXME : Not sure how exactly the SnapshotSelectionCriteria is applied
 602         // For now guessing that it is ANDed.
 603         persistence().deleteSnapshots(new SnapshotSelectionCriteria(
 604             sequenceNumber - context.getConfigParams().getSnapshotBatchCount(), 43200000));
 605
 606         // Trim akka journal
 607         persistence().deleteMessages(sequenceNumber);
 608     }
 609
 610     private String getLeaderAddress(){
 611         if(isLeader()){
 612             return getSelf().path().toString();
 613         }
 614         String leaderId = currentBehavior.getLeaderId();
 615         if (leaderId == null) {
 616             return null;
 617         }
 618         String peerAddress = context.getPeerAddress(leaderId);
 619         if(LOG.isDebugEnabled()) {
 620             LOG.debug("getLeaderAddress leaderId = {} peerAddress = {}",
 621                     leaderId, peerAddress);
 622         }
 623
 624         return peerAddress;
 625     }
 626
 627     private void handleCaptureSnapshotReply(ByteString stateInBytes) {
 628         // create a snapshot object from the state provided and save it
 629         // when snapshot is saved async, SaveSnapshotSuccess is raised.
 630
 631         Snapshot sn = Snapshot.create(stateInBytes.toByteArray(),
 632             context.getReplicatedLog().getFrom(captureSnapshot.getLastAppliedIndex() + 1),
 633             captureSnapshot.getLastIndex(), captureSnapshot.getLastTerm(),
 634             captureSnapshot.getLastAppliedIndex(), captureSnapshot.getLastAppliedTerm());
 635
 636         persistence().saveSnapshot(sn);
 637
 638         LOG.info("Persisting of snapshot done:{}", sn.getLogMessage());
 639
 640         //be greedy and remove entries from in-mem journal which are in the snapshot
 641         // and update snapshotIndex and snapshotTerm without waiting for the success,
 642
 643         context.getReplicatedLog().snapshotPreCommit(
 644             captureSnapshot.getLastAppliedIndex(),
 645             captureSnapshot.getLastAppliedTerm());
 646
 647         LOG.info("Removed in-memory snapshotted entries, adjusted snaphsotIndex:{} " +
 648             "and term:{}", captureSnapshot.getLastAppliedIndex(),
 649             captureSnapshot.getLastAppliedTerm());
 650
 651         if (isLeader() && captureSnapshot.isInstallSnapshotInitiated()) {
 652             // this would be call straight to the leader and won't initiate in serialization
 653             currentBehavior.handleMessage(getSelf(), new SendInstallSnapshot(stateInBytes));
 654         }
 655
 656         captureSnapshot = null;
 657         hasSnapshotCaptureInitiated = false;
 658     }
 659
 660     private class ReplicatedLogImpl extends AbstractReplicatedLogImpl {
 661
 662         public ReplicatedLogImpl(Snapshot snapshot) {
 663             super(snapshot.getLastAppliedIndex(), snapshot.getLastAppliedTerm(),
 664                 snapshot.getUnAppliedEntries());
 665         }
 666
 667         public ReplicatedLogImpl() {
 668             super();
 669         }
 670
 671         @Override public void removeFromAndPersist(long logEntryIndex) {
 672             int adjustedIndex = adjustedIndex(logEntryIndex);
 673
 674             if (adjustedIndex < 0) {
 675                 return;
 676             }
 677
 678             // FIXME: Maybe this should be done after the command is saved
 679             journal.subList(adjustedIndex , journal.size()).clear();
 680
 681             persistence().persist(new DeleteEntries(adjustedIndex), new Procedure<DeleteEntries>(){
 682
 683                 @Override public void apply(DeleteEntries param)
 684                     throws Exception {
 685                     //FIXME : Doing nothing for now
 686                 }
 687             });
 688         }
 689
 690         @Override public void appendAndPersist(
 691             final ReplicatedLogEntry replicatedLogEntry) {
 692             appendAndPersist(null, null, replicatedLogEntry);
 693         }
 694
 695         public void appendAndPersist(final ActorRef clientActor,
 696             final String identifier,
 697             final ReplicatedLogEntry replicatedLogEntry) {
 698
 699             if(LOG.isDebugEnabled()) {
 700                 LOG.debug("Append log entry and persist {} ", replicatedLogEntry);
 701             }
 702
 703             // FIXME : By adding the replicated log entry to the in-memory journal we are not truly ensuring durability of the logs
 704             journal.add(replicatedLogEntry);
 705
 706             // When persisting events with persist it is guaranteed that the
 707             // persistent actor will not receive further commands between the
 708             // persist call and the execution(s) of the associated event
 709             // handler. This also holds for multiple persist calls in context
 710             // of a single command.
 711             persistence().persist(replicatedLogEntry,
 712                 new Procedure<ReplicatedLogEntry>() {
 713                     @Override
 714                     public void apply(ReplicatedLogEntry evt) throws Exception {
 715                         // when a snaphsot is being taken, captureSnapshot != null
 716                         if (hasSnapshotCaptureInitiated == false &&
 717                             journal.size() % context.getConfigParams().getSnapshotBatchCount() == 0) {
 718
 719                             LOG.info("Initiating Snapshot Capture..");
 720                             long lastAppliedIndex = -1;
 721                             long lastAppliedTerm = -1;
 722
 723                             ReplicatedLogEntry lastAppliedEntry = get(context.getLastApplied());
 724                             if (lastAppliedEntry != null) {
 725                                 lastAppliedIndex = lastAppliedEntry.getIndex();
 726                                 lastAppliedTerm = lastAppliedEntry.getTerm();
 727                             }
 728
 729                             if(LOG.isDebugEnabled()) {
 730                                 LOG.debug("Snapshot Capture logSize: {}", journal.size());
 731                                 LOG.debug("Snapshot Capture lastApplied:{} ",
 732                                     context.getLastApplied());
 733                                 LOG.debug("Snapshot Capture lastAppliedIndex:{}", lastAppliedIndex);
 734                                 LOG.debug("Snapshot Capture lastAppliedTerm:{}", lastAppliedTerm);
 735                             }
 736
 737                             // send a CaptureSnapshot to self to make the expensive operation async.
 738                             getSelf().tell(new CaptureSnapshot(
 739                                 lastIndex(), lastTerm(), lastAppliedIndex, lastAppliedTerm),
 740                                 null);
 741                             hasSnapshotCaptureInitiated = true;
 742                         }
 743                         // Send message for replication
 744                         if (clientActor != null) {
 745                             currentBehavior.handleMessage(getSelf(),
 746                                 new Replicate(clientActor, identifier,
 747                                     replicatedLogEntry)
 748                             );
 749                         }
 750                     }
 751                 }
 752             );
 753         }
 754
 755     }
 756
 757     static class DeleteEntries implements Serializable {
 758         private final int fromIndex;
 759
 760
 761         public DeleteEntries(int fromIndex) {
 762             this.fromIndex = fromIndex;
 763         }
 764
 765         public int getFromIndex() {
 766             return fromIndex;
 767         }
 768     }
 769
 770
 771     private class ElectionTermImpl implements ElectionTerm {
 772         /**
 773          * Identifier of the actor whose election term information this is
 774          */
 775         private long currentTerm = 0;
 776         private String votedFor = null;
 777
 778         @Override
 779         public long getCurrentTerm() {
 780             return currentTerm;
 781         }
 782
 783         @Override
 784         public String getVotedFor() {
 785             return votedFor;
 786         }
 787
 788         @Override public void update(long currentTerm, String votedFor) {
 789             if(LOG.isDebugEnabled()) {
 790                 LOG.debug("Set currentTerm={}, votedFor={}", currentTerm, votedFor);
 791             }
 792             this.currentTerm = currentTerm;
 793             this.votedFor = votedFor;
 794         }
 795
 796         @Override
 797         public void updateAndPersist(long currentTerm, String votedFor){
 798             update(currentTerm, votedFor);
 799             // FIXME : Maybe first persist then update the state
 800             persistence().persist(new UpdateElectionTerm(this.currentTerm, this.votedFor), new Procedure<UpdateElectionTerm>(){
 801
 802                 @Override public void apply(UpdateElectionTerm param)
 803                     throws Exception {
 804
 805                 }
 806             });
 807         }
 808     }
 809
 810     static class UpdateElectionTerm implements Serializable {
 811         private final long currentTerm;
 812         private final String votedFor;
 813
 814         public UpdateElectionTerm(long currentTerm, String votedFor) {
 815             this.currentTerm = currentTerm;
 816             this.votedFor = votedFor;
 817         }
 818
 819         public long getCurrentTerm() {
 820             return currentTerm;
 821         }
 822
 823         public String getVotedFor() {
 824             return votedFor;
 825         }
 826     }
 827
 828     protected class NonPersistentRaftDataProvider extends NonPersistentDataProvider {
 829
 830         public NonPersistentRaftDataProvider(){
 831
 832         }
 833
 834         /**
 835          * The way snapshotting works is,
 836          * <ol>
 837          * <li> RaftActor calls createSnapshot on the Shard
 838          * <li> Shard sends a CaptureSnapshotReply and RaftActor then calls saveSnapshot
 839          * <li> When saveSnapshot is invoked on the akka-persistence API it uses the SnapshotStore to save the snapshot.
 840          * The SnapshotStore sends SaveSnapshotSuccess or SaveSnapshotFailure. When the RaftActor gets SaveSnapshot
 841          * success it commits the snapshot to the in-memory journal. This commitSnapshot is mimicking what is done
 842          * in SaveSnapshotSuccess.
 843          * </ol>
 844          * @param o
 845          */
 846         @Override
 847         public void saveSnapshot(Object o) {
 848             // Make saving Snapshot successful
 849             commitSnapshot(-1L);
 850         }
 851     }
 852
 853     @VisibleForTesting
 854     void setCurrentBehavior(AbstractRaftActorBehavior behavior) {
 855         currentBehavior = behavior;
 856     }
 857
 858     protected RaftActorBehavior getCurrentBehavior() {
 859         return currentBehavior;
 860     }
 861 }