opendaylight/md-sal/sal-akka-raft/src/main/java/org/opendaylight/controller/cluster/raft/behaviors/AbstractLeader.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.behaviors;
  10
  11 import akka.actor.ActorRef;
  12 import akka.actor.ActorSelection;
  13 import akka.actor.Cancellable;
  14 import com.google.common.annotations.VisibleForTesting;
  15 import com.google.common.base.Optional;
  16 import com.google.common.base.Preconditions;
  17 import com.google.common.collect.ImmutableMap;
  18 import com.google.common.collect.ImmutableMap.Builder;
  19 import com.google.protobuf.ByteString;
  20 import java.io.IOException;
  21 import java.util.ArrayList;
  22 import java.util.Collection;
  23 import java.util.Collections;
  24 import java.util.HashMap;
  25 import java.util.List;
  26 import java.util.Map;
  27 import java.util.Map.Entry;
  28 import java.util.concurrent.TimeUnit;
  29 import org.opendaylight.controller.cluster.raft.ClientRequestTracker;
  30 import org.opendaylight.controller.cluster.raft.ClientRequestTrackerImpl;
  31 import org.opendaylight.controller.cluster.raft.FollowerLogInformation;
  32 import org.opendaylight.controller.cluster.raft.FollowerLogInformationImpl;
  33 import org.opendaylight.controller.cluster.raft.RaftActorContext;
  34 import org.opendaylight.controller.cluster.raft.RaftState;
  35 import org.opendaylight.controller.cluster.raft.ReplicatedLogEntry;
  36 import org.opendaylight.controller.cluster.raft.base.messages.CaptureSnapshot;
  37 import org.opendaylight.controller.cluster.raft.base.messages.InitiateInstallSnapshot;
  38 import org.opendaylight.controller.cluster.raft.base.messages.Replicate;
  39 import org.opendaylight.controller.cluster.raft.base.messages.SendHeartBeat;
  40 import org.opendaylight.controller.cluster.raft.base.messages.SendInstallSnapshot;
  41 import org.opendaylight.controller.cluster.raft.messages.AppendEntries;
  42 import org.opendaylight.controller.cluster.raft.messages.AppendEntriesReply;
  43 import org.opendaylight.controller.cluster.raft.messages.InstallSnapshot;
  44 import org.opendaylight.controller.cluster.raft.messages.InstallSnapshotReply;
  45 import org.opendaylight.controller.cluster.raft.messages.RaftRPC;
  46 import org.opendaylight.controller.cluster.raft.messages.RequestVoteReply;
  47 import scala.concurrent.duration.FiniteDuration;
  48
  49 /**
  50  * The behavior of a RaftActor when it is in the Leader state
  51  * <p/>
  52  * Leaders:
  53  * <ul>
  54  * <li> Upon election: send initial empty AppendEntries RPCs
  55  * (heartbeat) to each server; repeat during idle periods to
  56  * prevent election timeouts (§5.2)
  57  * <li> If command received from client: append entry to local log,
  58  * respond after entry applied to state machine (§5.3)
  59  * <li> If last log index ≥ nextIndex for a follower: send
  60  * AppendEntries RPC with log entries starting at nextIndex
  61  * <ul>
  62  * <li> If successful: update nextIndex and matchIndex for
  63  * follower (§5.3)
  64  * <li> If AppendEntries fails because of log inconsistency:
  65  * decrement nextIndex and retry (§5.3)
  66  * </ul>
  67  * <li> If there exists an N such that N > commitIndex, a majority
  68  * of matchIndex[i] ≥ N, and log[N].term == currentTerm:
  69  * set commitIndex = N (§5.3, §5.4).
  70  */
  71 public abstract class AbstractLeader extends AbstractRaftActorBehavior {
  72
  73     // The index of the first chunk that is sent when installing a snapshot
  74     public static final int FIRST_CHUNK_INDEX = 1;
  75
  76     // The index that the follower should respond with if it needs the install snapshot to be reset
  77     public static final int INVALID_CHUNK_INDEX = -1;
  78
  79     // This would be passed as the hash code of the last chunk when sending the first chunk
  80     public static final int INITIAL_LAST_CHUNK_HASH_CODE = -1;
  81
  82     private final Map<String, FollowerLogInformation> followerToLog;
  83     private final Map<String, FollowerToSnapshot> mapFollowerToSnapshot = new HashMap<>();
  84
  85     private Cancellable heartbeatSchedule = null;
  86
  87     private List<ClientRequestTracker> trackerList = new ArrayList<>();
  88
  89     protected final int minReplicationCount;
  90
  91     protected final int minIsolatedLeaderPeerCount;
  92
  93     private Optional<ByteString> snapshot;
  94
  95     public AbstractLeader(RaftActorContext context) {
  96         super(context);
  97
  98         final Builder<String, FollowerLogInformation> ftlBuilder = ImmutableMap.builder();
  99         for (String followerId : context.getPeerAddresses().keySet()) {
 100             FollowerLogInformation followerLogInformation =
 101                 new FollowerLogInformationImpl(followerId,
 102                     context.getCommitIndex(), -1,
 103                     context.getConfigParams().getElectionTimeOutInterval());
 104
 105             ftlBuilder.put(followerId, followerLogInformation);
 106         }
 107         followerToLog = ftlBuilder.build();
 108
 109         leaderId = context.getId();
 110
 111         LOG.debug("Election:Leader has following peers: {}", getFollowerIds());
 112
 113         minReplicationCount = getMajorityVoteCount(getFollowerIds().size());
 114
 115         // the isolated Leader peer count will be 1 less than the majority vote count.
 116         // this is because the vote count has the self vote counted in it
 117         // for e.g
 118         // 0 peers = 1 votesRequired , minIsolatedLeaderPeerCount = 0
 119         // 2 peers = 2 votesRequired , minIsolatedLeaderPeerCount = 1
 120         // 4 peers = 3 votesRequired, minIsolatedLeaderPeerCount = 2
 121         minIsolatedLeaderPeerCount = minReplicationCount > 0 ? (minReplicationCount - 1) : 0;
 122
 123         snapshot = Optional.absent();
 124
 125         // Immediately schedule a heartbeat
 126         // Upon election: send initial empty AppendEntries RPCs
 127         // (heartbeat) to each server; repeat during idle periods to
 128         // prevent election timeouts (§5.2)
 129         scheduleHeartBeat(new FiniteDuration(0, TimeUnit.SECONDS));
 130     }
 131
 132     /**
 133      * Return an immutable collection of follower identifiers.
 134      *
 135      * @return Collection of follower IDs
 136      */
 137     protected final Collection<String> getFollowerIds() {
 138         return followerToLog.keySet();
 139     }
 140
 141     private Optional<ByteString> getSnapshot() {
 142         return snapshot;
 143     }
 144
 145     @VisibleForTesting
 146     void setSnapshot(Optional<ByteString> snapshot) {
 147         this.snapshot = snapshot;
 148     }
 149
 150     @Override
 151     protected RaftActorBehavior handleAppendEntries(ActorRef sender,
 152         AppendEntries appendEntries) {
 153
 154         if(LOG.isDebugEnabled()) {
 155             LOG.debug(appendEntries.toString());
 156         }
 157
 158         return this;
 159     }
 160
 161     @Override
 162     protected RaftActorBehavior handleAppendEntriesReply(ActorRef sender,
 163         AppendEntriesReply appendEntriesReply) {
 164
 165         if(! appendEntriesReply.isSuccess()) {
 166             if(LOG.isDebugEnabled()) {
 167                 LOG.debug(appendEntriesReply.toString());
 168             }
 169         }
 170
 171         // Update the FollowerLogInformation
 172         String followerId = appendEntriesReply.getFollowerId();
 173         FollowerLogInformation followerLogInformation =
 174             followerToLog.get(followerId);
 175
 176         if(followerLogInformation == null){
 177             LOG.error("Unknown follower {}", followerId);
 178             return this;
 179         }
 180
 181         followerLogInformation.markFollowerActive();
 182
 183         if (appendEntriesReply.isSuccess()) {
 184             followerLogInformation
 185                 .setMatchIndex(appendEntriesReply.getLogLastIndex());
 186             followerLogInformation
 187                 .setNextIndex(appendEntriesReply.getLogLastIndex() + 1);
 188         } else {
 189
 190             // TODO: When we find that the follower is out of sync with the
 191             // Leader we simply decrement that followers next index by 1.
 192             // Would it be possible to do better than this? The RAFT spec
 193             // does not explicitly deal with it but may be something for us to
 194             // think about
 195
 196             followerLogInformation.decrNextIndex();
 197         }
 198
 199         // Now figure out if this reply warrants a change in the commitIndex
 200         // If there exists an N such that N > commitIndex, a majority
 201         // of matchIndex[i] ≥ N, and log[N].term == currentTerm:
 202         // set commitIndex = N (§5.3, §5.4).
 203         for (long N = context.getCommitIndex() + 1; ; N++) {
 204             int replicatedCount = 1;
 205
 206             for (FollowerLogInformation info : followerToLog.values()) {
 207                 if (info.getMatchIndex() >= N) {
 208                     replicatedCount++;
 209                 }
 210             }
 211
 212             if (replicatedCount >= minReplicationCount) {
 213                 ReplicatedLogEntry replicatedLogEntry = context.getReplicatedLog().get(N);
 214                 if (replicatedLogEntry != null &&
 215                     replicatedLogEntry.getTerm() == currentTerm()) {
 216                     context.setCommitIndex(N);
 217                 }
 218             } else {
 219                 break;
 220             }
 221         }
 222
 223         // Apply the change to the state machine
 224         if (context.getCommitIndex() > context.getLastApplied()) {
 225             applyLogToStateMachine(context.getCommitIndex());
 226         }
 227
 228         return this;
 229     }
 230
 231     @Override
 232     protected ClientRequestTracker removeClientRequestTracker(long logIndex) {
 233
 234         ClientRequestTracker toRemove = findClientRequestTracker(logIndex);
 235         if(toRemove != null) {
 236             trackerList.remove(toRemove);
 237         }
 238
 239         return toRemove;
 240     }
 241
 242     @Override
 243     protected ClientRequestTracker findClientRequestTracker(long logIndex) {
 244         for (ClientRequestTracker tracker : trackerList) {
 245             if (tracker.getIndex() == logIndex) {
 246                 return tracker;
 247             }
 248         }
 249         return null;
 250     }
 251
 252     @Override
 253     protected RaftActorBehavior handleRequestVoteReply(ActorRef sender,
 254         RequestVoteReply requestVoteReply) {
 255         return this;
 256     }
 257
 258     @Override
 259     public RaftState state() {
 260         return RaftState.Leader;
 261     }
 262
 263     @Override
 264     public RaftActorBehavior handleMessage(ActorRef sender, Object originalMessage) {
 265         Preconditions.checkNotNull(sender, "sender should not be null");
 266
 267         Object message = fromSerializableMessage(originalMessage);
 268
 269         if (message instanceof RaftRPC) {
 270             RaftRPC rpc = (RaftRPC) message;
 271             // If RPC request or response contains term T > currentTerm:
 272             // set currentTerm = T, convert to follower (§5.1)
 273             // This applies to all RPC messages and responses
 274             if (rpc.getTerm() > context.getTermInformation().getCurrentTerm()) {
 275                 context.getTermInformation().updateAndPersist(rpc.getTerm(), null);
 276
 277                 return switchBehavior(new Follower(context));
 278             }
 279         }
 280
 281         try {
 282             if (message instanceof SendHeartBeat) {
 283                 sendHeartBeat();
 284                 return this;
 285
 286             } else if(message instanceof InitiateInstallSnapshot) {
 287                 installSnapshotIfNeeded();
 288
 289             } else if(message instanceof SendInstallSnapshot) {
 290                 // received from RaftActor
 291                 setSnapshot(Optional.of(((SendInstallSnapshot) message).getSnapshot()));
 292                 sendInstallSnapshot();
 293
 294             } else if (message instanceof Replicate) {
 295                 replicate((Replicate) message);
 296
 297             } else if (message instanceof InstallSnapshotReply){
 298                 handleInstallSnapshotReply((InstallSnapshotReply) message);
 299
 300             }
 301         } finally {
 302             scheduleHeartBeat(context.getConfigParams().getHeartBeatInterval());
 303         }
 304
 305         return super.handleMessage(sender, message);
 306     }
 307
 308     private void handleInstallSnapshotReply(InstallSnapshotReply reply) {
 309         String followerId = reply.getFollowerId();
 310         FollowerToSnapshot followerToSnapshot = mapFollowerToSnapshot.get(followerId);
 311         FollowerLogInformation followerLogInformation = followerToLog.get(followerId);
 312         followerLogInformation.markFollowerActive();
 313
 314         if (followerToSnapshot != null &&
 315             followerToSnapshot.getChunkIndex() == reply.getChunkIndex()) {
 316
 317             if (reply.isSuccess()) {
 318                 if(followerToSnapshot.isLastChunk(reply.getChunkIndex())) {
 319                     //this was the last chunk reply
 320                     if(LOG.isDebugEnabled()) {
 321                         LOG.debug("InstallSnapshotReply received, " +
 322                                 "last chunk received, Chunk:{}. Follower:{} Setting nextIndex:{}",
 323                             reply.getChunkIndex(), followerId,
 324                             context.getReplicatedLog().getSnapshotIndex() + 1
 325                         );
 326                     }
 327
 328                     followerLogInformation.setMatchIndex(
 329                         context.getReplicatedLog().getSnapshotIndex());
 330                     followerLogInformation.setNextIndex(
 331                         context.getReplicatedLog().getSnapshotIndex() + 1);
 332                     mapFollowerToSnapshot.remove(followerId);
 333
 334                     if(LOG.isDebugEnabled()) {
 335                         LOG.debug("followerToLog.get(followerId).getNextIndex()=" +
 336                             followerToLog.get(followerId).getNextIndex());
 337                     }
 338
 339                     if (mapFollowerToSnapshot.isEmpty()) {
 340                         // once there are no pending followers receiving snapshots
 341                         // we can remove snapshot from the memory
 342                         setSnapshot(Optional.<ByteString>absent());
 343                     }
 344
 345                 } else {
 346                     followerToSnapshot.markSendStatus(true);
 347                 }
 348             } else {
 349                 LOG.info("InstallSnapshotReply received, " +
 350                         "sending snapshot chunk failed, Will retry, Chunk:{}",
 351                     reply.getChunkIndex()
 352                 );
 353
 354                 followerToSnapshot.markSendStatus(false);
 355             }
 356
 357         } else {
 358             LOG.error("ERROR!!" +
 359                     "FollowerId in InstallSnapshotReply not known to Leader" +
 360                     " or Chunk Index in InstallSnapshotReply not matching {} != {}",
 361                 followerToSnapshot.getChunkIndex(), reply.getChunkIndex()
 362             );
 363
 364             if(reply.getChunkIndex() == INVALID_CHUNK_INDEX){
 365                 // Since the Follower did not find this index to be valid we should reset the follower snapshot
 366                 // so that Installing the snapshot can resume from the beginning
 367                 followerToSnapshot.reset();
 368             }
 369         }
 370     }
 371
 372     private void replicate(Replicate replicate) {
 373         long logIndex = replicate.getReplicatedLogEntry().getIndex();
 374
 375         if(LOG.isDebugEnabled()) {
 376             LOG.debug("Replicate message {}", logIndex);
 377         }
 378
 379         // Create a tracker entry we will use this later to notify the
 380         // client actor
 381         trackerList.add(
 382             new ClientRequestTrackerImpl(replicate.getClientActor(),
 383                 replicate.getIdentifier(),
 384                 logIndex)
 385         );
 386
 387         if (followerToLog.isEmpty()) {
 388             context.setCommitIndex(logIndex);
 389             applyLogToStateMachine(logIndex);
 390         } else {
 391             sendAppendEntries();
 392         }
 393     }
 394
 395     private void sendAppendEntries() {
 396         // Send an AppendEntries to all followers
 397         for (Entry<String, FollowerLogInformation> e : followerToLog.entrySet()) {
 398             final String followerId = e.getKey();
 399             ActorSelection followerActor = context.getPeerActorSelection(followerId);
 400
 401             if (followerActor != null) {
 402                 FollowerLogInformation followerLogInformation = followerToLog.get(followerId);
 403                 long followerNextIndex = followerLogInformation.getNextIndex();
 404                 boolean isFollowerActive = followerLogInformation.isFollowerActive();
 405                 List<ReplicatedLogEntry> entries = null;
 406
 407                 if (mapFollowerToSnapshot.get(followerId) != null) {
 408                     // if install snapshot is in process , then sent next chunk if possible
 409                     if (isFollowerActive && mapFollowerToSnapshot.get(followerId).canSendNextChunk()) {
 410                         sendSnapshotChunk(followerActor, followerId);
 411                     } else {
 412                         // we send a heartbeat even if we have not received a reply for the last chunk
 413                         sendAppendEntriesToFollower(followerActor, followerNextIndex,
 414                             Collections.<ReplicatedLogEntry>emptyList());
 415                     }
 416
 417                 } else {
 418                     long leaderLastIndex = context.getReplicatedLog().lastIndex();
 419                     long leaderSnapShotIndex = context.getReplicatedLog().getSnapshotIndex();
 420
 421                     if (isFollowerActive &&
 422                         context.getReplicatedLog().isPresent(followerNextIndex)) {
 423                         // FIXME : Sending one entry at a time
 424                         entries = context.getReplicatedLog().getFrom(followerNextIndex, 1);
 425
 426                     } else if (isFollowerActive && followerNextIndex >= 0 &&
 427                         leaderLastIndex >= followerNextIndex ) {
 428                         // if the followers next index is not present in the leaders log, and
 429                         // if the follower is just not starting and if leader's index is more than followers index
 430                         // then snapshot should be sent
 431
 432                         if(LOG.isDebugEnabled()) {
 433                             LOG.debug("InitiateInstallSnapshot to follower:{}," +
 434                                     "follower-nextIndex:{}, leader-snapshot-index:{},  " +
 435                                     "leader-last-index:{}", followerId,
 436                                 followerNextIndex, leaderSnapShotIndex, leaderLastIndex
 437                             );
 438                         }
 439                         actor().tell(new InitiateInstallSnapshot(), actor());
 440
 441                         // we would want to sent AE as the capture snapshot might take time
 442                         entries =  Collections.<ReplicatedLogEntry>emptyList();
 443
 444                     } else {
 445                         //we send an AppendEntries, even if the follower is inactive
 446                         // in-order to update the followers timestamp, in case it becomes active again
 447                         entries =  Collections.<ReplicatedLogEntry>emptyList();
 448                     }
 449
 450                     sendAppendEntriesToFollower(followerActor, followerNextIndex, entries);
 451
 452                 }
 453             }
 454         }
 455     }
 456
 457     private void sendAppendEntriesToFollower(ActorSelection followerActor, long followerNextIndex,
 458         List<ReplicatedLogEntry> entries) {
 459         followerActor.tell(
 460             new AppendEntries(currentTerm(), context.getId(),
 461                 prevLogIndex(followerNextIndex),
 462                 prevLogTerm(followerNextIndex), entries,
 463                 context.getCommitIndex()).toSerializable(),
 464             actor()
 465         );
 466     }
 467
 468     /**
 469      * An installSnapshot is scheduled at a interval that is a multiple of
 470      * a HEARTBEAT_INTERVAL. This is to avoid the need to check for installing
 471      * snapshots at every heartbeat.
 472      *
 473      * Install Snapshot works as follows
 474      * 1. Leader sends a InitiateInstallSnapshot message to self
 475      * 2. Leader then initiates the capture snapshot by sending a CaptureSnapshot message to actor
 476      * 3. RaftActor on receipt of the CaptureSnapshotReply (from Shard), stores the received snapshot in the replicated log
 477      * and makes a call to Leader's handleMessage , with SendInstallSnapshot message.
 478      * 4. Leader , picks the snapshot from im-mem ReplicatedLog and sends it in chunks to the Follower
 479      * 5. On complete, Follower sends back a InstallSnapshotReply.
 480      * 6. On receipt of the InstallSnapshotReply for the last chunk, Leader marks the install complete for that follower
 481      * and replenishes the memory by deleting the snapshot in Replicated log.
 482      *
 483      */
 484     private void installSnapshotIfNeeded() {
 485         for (Entry<String, FollowerLogInformation> e : followerToLog.entrySet()) {
 486             final ActorSelection followerActor = context.getPeerActorSelection(e.getKey());
 487
 488             if (followerActor != null) {
 489                 long nextIndex = e.getValue().getNextIndex();
 490
 491                 if (!context.getReplicatedLog().isPresent(nextIndex) &&
 492                     context.getReplicatedLog().isInSnapshot(nextIndex)) {
 493                     LOG.info("{} follower needs a snapshot install", e.getKey());
 494                     if (snapshot.isPresent()) {
 495                         // if a snapshot is present in the memory, most likely another install is in progress
 496                         // no need to capture snapshot
 497                         sendSnapshotChunk(followerActor, e.getKey());
 498
 499                     } else {
 500                         initiateCaptureSnapshot();
 501                         //we just need 1 follower who would need snapshot to be installed.
 502                         // when we have the snapshot captured, we would again check (in SendInstallSnapshot)
 503                         // who needs an install and send to all who need
 504                         break;
 505                     }
 506
 507                 }
 508             }
 509         }
 510     }
 511
 512     // on every install snapshot, we try to capture the snapshot.
 513     // Once a capture is going on, another one issued will get ignored by RaftActor.
 514     private void initiateCaptureSnapshot() {
 515         LOG.info("Initiating Snapshot Capture to Install Snapshot, Leader:{}", getLeaderId());
 516         ReplicatedLogEntry lastAppliedEntry = context.getReplicatedLog().get(context.getLastApplied());
 517         long lastAppliedIndex = -1;
 518         long lastAppliedTerm = -1;
 519
 520         if (lastAppliedEntry != null) {
 521             lastAppliedIndex = lastAppliedEntry.getIndex();
 522             lastAppliedTerm = lastAppliedEntry.getTerm();
 523         } else if (context.getReplicatedLog().getSnapshotIndex() > -1)  {
 524             lastAppliedIndex = context.getReplicatedLog().getSnapshotIndex();
 525             lastAppliedTerm = context.getReplicatedLog().getSnapshotTerm();
 526         }
 527
 528         boolean isInstallSnapshotInitiated = true;
 529         actor().tell(new CaptureSnapshot(lastIndex(), lastTerm(),
 530                 lastAppliedIndex, lastAppliedTerm, isInstallSnapshotInitiated),
 531             actor());
 532     }
 533
 534
 535     private void sendInstallSnapshot() {
 536         for (Entry<String, FollowerLogInformation> e : followerToLog.entrySet()) {
 537             ActorSelection followerActor = context.getPeerActorSelection(e.getKey());
 538
 539             if (followerActor != null) {
 540                 long nextIndex = e.getValue().getNextIndex();
 541
 542                 if (!context.getReplicatedLog().isPresent(nextIndex) &&
 543                     context.getReplicatedLog().isInSnapshot(nextIndex)) {
 544                     sendSnapshotChunk(followerActor, e.getKey());
 545                 }
 546             }
 547         }
 548     }
 549
 550     /**
 551      *  Sends a snapshot chunk to a given follower
 552      *  InstallSnapshot should qualify as a heartbeat too.
 553      */
 554     private void sendSnapshotChunk(ActorSelection followerActor, String followerId) {
 555         try {
 556             if (snapshot.isPresent()) {
 557                 followerActor.tell(
 558                     new InstallSnapshot(currentTerm(), context.getId(),
 559                         context.getReplicatedLog().getSnapshotIndex(),
 560                         context.getReplicatedLog().getSnapshotTerm(),
 561                         getNextSnapshotChunk(followerId,snapshot.get()),
 562                         mapFollowerToSnapshot.get(followerId).incrementChunkIndex(),
 563                         mapFollowerToSnapshot.get(followerId).getTotalChunks(),
 564                         Optional.of(mapFollowerToSnapshot.get(followerId).getLastChunkHashCode())
 565                     ).toSerializable(),
 566                     actor()
 567                 );
 568                 LOG.info("InstallSnapshot sent to follower {}, Chunk: {}/{}",
 569                     followerActor.path(), mapFollowerToSnapshot.get(followerId).getChunkIndex(),
 570                     mapFollowerToSnapshot.get(followerId).getTotalChunks());
 571             }
 572         } catch (IOException e) {
 573             LOG.error(e, "InstallSnapshot failed for Leader.");
 574         }
 575     }
 576
 577     /**
 578      * Acccepts snaphot as ByteString, enters into map for future chunks
 579      * creates and return a ByteString chunk
 580      */
 581     private ByteString getNextSnapshotChunk(String followerId, ByteString snapshotBytes) throws IOException {
 582         FollowerToSnapshot followerToSnapshot = mapFollowerToSnapshot.get(followerId);
 583         if (followerToSnapshot == null) {
 584             followerToSnapshot = new FollowerToSnapshot(snapshotBytes);
 585             mapFollowerToSnapshot.put(followerId, followerToSnapshot);
 586         }
 587         ByteString nextChunk = followerToSnapshot.getNextChunk();
 588         if (LOG.isDebugEnabled()) {
 589             LOG.debug("Leader's snapshot nextChunk size:{}", nextChunk.size());
 590         }
 591         return nextChunk;
 592     }
 593
 594     private void sendHeartBeat() {
 595         if (!followerToLog.isEmpty()) {
 596             sendAppendEntries();
 597         }
 598     }
 599
 600     private void stopHeartBeat() {
 601         if (heartbeatSchedule != null && !heartbeatSchedule.isCancelled()) {
 602             heartbeatSchedule.cancel();
 603         }
 604     }
 605
 606     private void scheduleHeartBeat(FiniteDuration interval) {
 607         if (followerToLog.isEmpty()) {
 608             // Optimization - do not bother scheduling a heartbeat as there are
 609             // no followers
 610             return;
 611         }
 612
 613         stopHeartBeat();
 614
 615         // Schedule a heartbeat. When the scheduler triggers a SendHeartbeat
 616         // message is sent to itself.
 617         // Scheduling the heartbeat only once here because heartbeats do not
 618         // need to be sent if there are other messages being sent to the remote
 619         // actor.
 620         heartbeatSchedule = context.getActorSystem().scheduler().scheduleOnce(
 621             interval, context.getActor(), new SendHeartBeat(),
 622             context.getActorSystem().dispatcher(), context.getActor());
 623     }
 624
 625     @Override
 626     public void close() throws Exception {
 627         stopHeartBeat();
 628     }
 629
 630     @Override
 631     public String getLeaderId() {
 632         return context.getId();
 633     }
 634
 635     protected boolean isLeaderIsolated() {
 636         int minPresent = minIsolatedLeaderPeerCount;
 637         for (FollowerLogInformation followerLogInformation : followerToLog.values()) {
 638             if (followerLogInformation.isFollowerActive()) {
 639                 --minPresent;
 640                 if (minPresent == 0) {
 641                     break;
 642                 }
 643             }
 644         }
 645         return (minPresent != 0);
 646     }
 647
 648     /**
 649      * Encapsulates the snapshot bytestring and handles the logic of sending
 650      * snapshot chunks
 651      */
 652     protected class FollowerToSnapshot {
 653         private ByteString snapshotBytes;
 654         private int offset = 0;
 655         // the next snapshot chunk is sent only if the replyReceivedForOffset matches offset
 656         private int replyReceivedForOffset;
 657         // if replyStatus is false, the previous chunk is attempted
 658         private boolean replyStatus = false;
 659         private int chunkIndex;
 660         private int totalChunks;
 661         private int lastChunkHashCode = AbstractLeader.INITIAL_LAST_CHUNK_HASH_CODE;
 662         private int nextChunkHashCode = AbstractLeader.INITIAL_LAST_CHUNK_HASH_CODE;
 663
 664         public FollowerToSnapshot(ByteString snapshotBytes) {
 665             this.snapshotBytes = snapshotBytes;
 666             int size = snapshotBytes.size();
 667             totalChunks = ( size / context.getConfigParams().getSnapshotChunkSize()) +
 668                 ((size % context.getConfigParams().getSnapshotChunkSize()) > 0 ? 1 : 0);
 669             if(LOG.isDebugEnabled()) {
 670                 LOG.debug("Snapshot {} bytes, total chunks to send:{}",
 671                     size, totalChunks);
 672             }
 673             replyReceivedForOffset = -1;
 674             chunkIndex = AbstractLeader.FIRST_CHUNK_INDEX;
 675         }
 676
 677         public ByteString getSnapshotBytes() {
 678             return snapshotBytes;
 679         }
 680
 681         public int incrementOffset() {
 682             if(replyStatus) {
 683                 // if prev chunk failed, we would want to sent the same chunk again
 684                 offset = offset + context.getConfigParams().getSnapshotChunkSize();
 685             }
 686             return offset;
 687         }
 688
 689         public int incrementChunkIndex() {
 690             if (replyStatus) {
 691                 // if prev chunk failed, we would want to sent the same chunk again
 692                 chunkIndex =  chunkIndex + 1;
 693             }
 694             return chunkIndex;
 695         }
 696
 697         public int getChunkIndex() {
 698             return chunkIndex;
 699         }
 700
 701         public int getTotalChunks() {
 702             return totalChunks;
 703         }
 704
 705         public boolean canSendNextChunk() {
 706             // we only send a false if a chunk is sent but we have not received a reply yet
 707             return replyReceivedForOffset == offset;
 708         }
 709
 710         public boolean isLastChunk(int chunkIndex) {
 711             return totalChunks == chunkIndex;
 712         }
 713
 714         public void markSendStatus(boolean success) {
 715             if (success) {
 716                 // if the chunk sent was successful
 717                 replyReceivedForOffset = offset;
 718                 replyStatus = true;
 719                 lastChunkHashCode = nextChunkHashCode;
 720             } else {
 721                 // if the chunk sent was failure
 722                 replyReceivedForOffset = offset;
 723                 replyStatus = false;
 724             }
 725         }
 726
 727         public ByteString getNextChunk() {
 728             int snapshotLength = getSnapshotBytes().size();
 729             int start = incrementOffset();
 730             int size = context.getConfigParams().getSnapshotChunkSize();
 731             if (context.getConfigParams().getSnapshotChunkSize() > snapshotLength) {
 732                 size = snapshotLength;
 733             } else {
 734                 if ((start + context.getConfigParams().getSnapshotChunkSize()) > snapshotLength) {
 735                     size = snapshotLength - start;
 736                 }
 737             }
 738
 739             if(LOG.isDebugEnabled()) {
 740                 LOG.debug("length={}, offset={},size={}",
 741                     snapshotLength, start, size);
 742             }
 743             ByteString substring = getSnapshotBytes().substring(start, start + size);
 744             nextChunkHashCode = substring.hashCode();
 745             return substring;
 746         }
 747
 748         /**
 749          * reset should be called when the Follower needs to be sent the snapshot from the beginning
 750          */
 751         public void reset(){
 752             offset = 0;
 753             replyStatus = false;
 754             replyReceivedForOffset = offset;
 755             chunkIndex = AbstractLeader.FIRST_CHUNK_INDEX;
 756             lastChunkHashCode = AbstractLeader.INITIAL_LAST_CHUNK_HASH_CODE;
 757         }
 758
 759         public int getLastChunkHashCode() {
 760             return lastChunkHashCode;
 761         }
 762     }
 763
 764     // called from example-actor for printing the follower-states
 765     public String printFollowerStates() {
 766         final StringBuilder sb = new StringBuilder();
 767
 768         sb.append('[');
 769         for (FollowerLogInformation followerLogInformation : followerToLog.values()) {
 770             sb.append('{');
 771             sb.append(followerLogInformation.getId());
 772             sb.append(" state:");
 773             sb.append(followerLogInformation.isFollowerActive());
 774             sb.append("},");
 775         }
 776         sb.append(']');
 777
 778         return sb.toString();
 779     }
 780
 781     @VisibleForTesting
 782     public FollowerLogInformation getFollower(String followerId) {
 783         return followerToLog.get(followerId);
 784     }
 785
 786     @VisibleForTesting
 787     protected void setFollowerSnapshot(String followerId, FollowerToSnapshot snapshot) {
 788         mapFollowerToSnapshot.put(followerId, snapshot);
 789     }
 790
 791     @VisibleForTesting
 792     public int followerSnapshotSize() {
 793         return mapFollowerToSnapshot.size();
 794     }
 795
 796     @VisibleForTesting
 797     public int followerLogSize() {
 798         return followerToLog.size();
 799     }
 800 }