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.Collection;
  22 import java.util.Collections;
  23 import java.util.HashMap;
  24 import java.util.Iterator;
  25 import java.util.LinkedList;
  26 import java.util.List;
  27 import java.util.Map;
  28 import java.util.Map.Entry;
  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.Replicate;
  38 import org.opendaylight.controller.cluster.raft.base.messages.SendHeartBeat;
  39 import org.opendaylight.controller.cluster.raft.base.messages.SendInstallSnapshot;
  40 import org.opendaylight.controller.cluster.raft.messages.AppendEntries;
  41 import org.opendaylight.controller.cluster.raft.messages.AppendEntriesReply;
  42 import org.opendaylight.controller.cluster.raft.messages.InstallSnapshot;
  43 import org.opendaylight.controller.cluster.raft.messages.InstallSnapshotReply;
  44 import org.opendaylight.controller.cluster.raft.messages.RaftRPC;
  45 import org.opendaylight.controller.cluster.raft.messages.RequestVoteReply;
  46 import scala.concurrent.duration.FiniteDuration;
  47
  48 /**
  49  * The behavior of a RaftActor when it is in the Leader state
  50  * <p/>
  51  * Leaders:
  52  * <ul>
  53  * <li> Upon election: send initial empty AppendEntries RPCs
  54  * (heartbeat) to each server; repeat during idle periods to
  55  * prevent election timeouts (§5.2)
  56  * <li> If command received from client: append entry to local log,
  57  * respond after entry applied to state machine (§5.3)
  58  * <li> If last log index ≥ nextIndex for a follower: send
  59  * AppendEntries RPC with log entries starting at nextIndex
  60  * <ul>
  61  * <li> If successful: update nextIndex and matchIndex for
  62  * follower (§5.3)
  63  * <li> If AppendEntries fails because of log inconsistency:
  64  * decrement nextIndex and retry (§5.3)
  65  * </ul>
  66  * <li> If there exists an N such that N > commitIndex, a majority
  67  * of matchIndex[i] ≥ N, and log[N].term == currentTerm:
  68  * set commitIndex = N (§5.3, §5.4).
  69  */
  70 public abstract class AbstractLeader extends AbstractRaftActorBehavior {
  71
  72     // The index of the first chunk that is sent when installing a snapshot
  73     public static final int FIRST_CHUNK_INDEX = 1;
  74
  75     // The index that the follower should respond with if it needs the install snapshot to be reset
  76     public static final int INVALID_CHUNK_INDEX = -1;
  77
  78     // This would be passed as the hash code of the last chunk when sending the first chunk
  79     public static final int INITIAL_LAST_CHUNK_HASH_CODE = -1;
  80
  81     private final Map<String, FollowerLogInformation> followerToLog;
  82     private final Map<String, FollowerToSnapshot> mapFollowerToSnapshot = new HashMap<>();
  83
  84     private Cancellable heartbeatSchedule = null;
  85
  86     private final Collection<ClientRequestTracker> trackerList = new LinkedList<>();
  87
  88     protected final int minReplicationCount;
  89
  90     protected final int minIsolatedLeaderPeerCount;
  91
  92     private Optional<ByteString> snapshot;
  93
  94     public AbstractLeader(RaftActorContext context) {
  95         super(context, RaftState.Leader);
  96
  97         final Builder<String, FollowerLogInformation> ftlBuilder = ImmutableMap.builder();
  98         for (String followerId : context.getPeerAddresses().keySet()) {
  99             FollowerLogInformation followerLogInformation =
 100                 new FollowerLogInformationImpl(followerId,
 101                     context.getCommitIndex(), -1,
 102                     context.getConfigParams().getElectionTimeOutInterval());
 103
 104             ftlBuilder.put(followerId, followerLogInformation);
 105         }
 106         followerToLog = ftlBuilder.build();
 107
 108         leaderId = context.getId();
 109
 110         LOG.debug("{}: Election: Leader has following peers: {}", logName(), getFollowerIds());
 111
 112         minReplicationCount = getMajorityVoteCount(getFollowerIds().size());
 113
 114         // the isolated Leader peer count will be 1 less than the majority vote count.
 115         // this is because the vote count has the self vote counted in it
 116         // for e.g
 117         // 0 peers = 1 votesRequired , minIsolatedLeaderPeerCount = 0
 118         // 2 peers = 2 votesRequired , minIsolatedLeaderPeerCount = 1
 119         // 4 peers = 3 votesRequired, minIsolatedLeaderPeerCount = 2
 120         minIsolatedLeaderPeerCount = minReplicationCount > 0 ? (minReplicationCount - 1) : 0;
 121
 122         snapshot = Optional.absent();
 123
 124         // Immediately schedule a heartbeat
 125         // Upon election: send initial empty AppendEntries RPCs
 126         // (heartbeat) to each server; repeat during idle periods to
 127         // prevent election timeouts (§5.2)
 128         sendAppendEntries(0, false);
 129
 130         // It is important to schedule this heartbeat here
 131         scheduleHeartBeat(context.getConfigParams().getHeartBeatInterval());
 132     }
 133
 134     /**
 135      * Return an immutable collection of follower identifiers.
 136      *
 137      * @return Collection of follower IDs
 138      */
 139     protected final Collection<String> getFollowerIds() {
 140         return followerToLog.keySet();
 141     }
 142
 143     @VisibleForTesting
 144     void setSnapshot(Optional<ByteString> snapshot) {
 145         this.snapshot = snapshot;
 146     }
 147
 148     @Override
 149     protected RaftActorBehavior handleAppendEntries(ActorRef sender,
 150         AppendEntries appendEntries) {
 151
 152         LOG.debug("{}: handleAppendEntries: {}", logName(), appendEntries);
 153
 154         return this;
 155     }
 156
 157     @Override
 158     protected RaftActorBehavior handleAppendEntriesReply(ActorRef sender,
 159         AppendEntriesReply appendEntriesReply) {
 160
 161         if(LOG.isTraceEnabled()) {
 162             LOG.trace("{}: handleAppendEntriesReply: {}", logName(), appendEntriesReply);
 163         } else if(LOG.isDebugEnabled() && !appendEntriesReply.isSuccess()) {
 164             LOG.debug("{}: handleAppendEntriesReply: {}", logName(), appendEntriesReply);
 165         }
 166
 167         // Update the FollowerLogInformation
 168         String followerId = appendEntriesReply.getFollowerId();
 169         FollowerLogInformation followerLogInformation =
 170             followerToLog.get(followerId);
 171
 172         if(followerLogInformation == null){
 173             LOG.error("{}: handleAppendEntriesReply - unknown follower {}", logName(), followerId);
 174             return this;
 175         }
 176
 177         if(followerLogInformation.timeSinceLastActivity() >
 178                 context.getConfigParams().getElectionTimeOutInterval().toMillis()) {
 179             LOG.error("{} : handleAppendEntriesReply delayed beyond election timeout, " +
 180                             "appendEntriesReply : {}, timeSinceLastActivity : {}, lastApplied : {}, commitIndex : {}",
 181                     logName(), appendEntriesReply, followerLogInformation.timeSinceLastActivity(),
 182                     context.getLastApplied(), context.getCommitIndex());
 183         }
 184
 185         followerLogInformation.markFollowerActive();
 186
 187         if (appendEntriesReply.isSuccess()) {
 188             followerLogInformation
 189                 .setMatchIndex(appendEntriesReply.getLogLastIndex());
 190             followerLogInformation
 191                 .setNextIndex(appendEntriesReply.getLogLastIndex() + 1);
 192         } else {
 193
 194             // TODO: When we find that the follower is out of sync with the
 195             // Leader we simply decrement that followers next index by 1.
 196             // Would it be possible to do better than this? The RAFT spec
 197             // does not explicitly deal with it but may be something for us to
 198             // think about
 199
 200             followerLogInformation.decrNextIndex();
 201         }
 202
 203         // Now figure out if this reply warrants a change in the commitIndex
 204         // If there exists an N such that N > commitIndex, a majority
 205         // of matchIndex[i] ≥ N, and log[N].term == currentTerm:
 206         // set commitIndex = N (§5.3, §5.4).
 207         for (long N = context.getCommitIndex() + 1; ; N++) {
 208             int replicatedCount = 1;
 209
 210             for (FollowerLogInformation info : followerToLog.values()) {
 211                 if (info.getMatchIndex() >= N) {
 212                     replicatedCount++;
 213                 }
 214             }
 215
 216             if (replicatedCount >= minReplicationCount) {
 217                 ReplicatedLogEntry replicatedLogEntry = context.getReplicatedLog().get(N);
 218                 if (replicatedLogEntry != null &&
 219                     replicatedLogEntry.getTerm() == currentTerm()) {
 220                     context.setCommitIndex(N);
 221                 }
 222             } else {
 223                 break;
 224             }
 225         }
 226
 227         // Apply the change to the state machine
 228         if (context.getCommitIndex() > context.getLastApplied()) {
 229             LOG.debug("{}: handleAppendEntriesReply: applying to log - commitIndex: {}, lastAppliedIndex: {}",
 230                     logName(), context.getCommitIndex(), context.getLastApplied());
 231
 232             applyLogToStateMachine(context.getCommitIndex());
 233         }
 234
 235         if (!context.isSnapshotCaptureInitiated()) {
 236             purgeInMemoryLog();
 237         }
 238
 239         //Send the next log entry immediately, if possible, no need to wait for heartbeat to trigger that event
 240         sendUpdatesToFollower(followerId, followerLogInformation, false, false);
 241         return this;
 242     }
 243
 244     private void purgeInMemoryLog() {
 245         //find the lowest index across followers which has been replicated to all.
 246         // lastApplied if there are no followers, so that we keep clearing the log for single-node
 247         // we would delete the in-mem log from that index on, in-order to minimize mem usage
 248         // we would also share this info thru AE with the followers so that they can delete their log entries as well.
 249         long minReplicatedToAllIndex = followerToLog.isEmpty() ? context.getLastApplied() : Long.MAX_VALUE;
 250         for (FollowerLogInformation info : followerToLog.values()) {
 251             minReplicatedToAllIndex = Math.min(minReplicatedToAllIndex, info.getMatchIndex());
 252         }
 253
 254         super.performSnapshotWithoutCapture(minReplicatedToAllIndex);
 255     }
 256
 257     @Override
 258     protected ClientRequestTracker removeClientRequestTracker(long logIndex) {
 259         final Iterator<ClientRequestTracker> it = trackerList.iterator();
 260         while (it.hasNext()) {
 261             final ClientRequestTracker t = it.next();
 262             if (t.getIndex() == logIndex) {
 263                 it.remove();
 264                 return t;
 265             }
 266         }
 267
 268         return null;
 269     }
 270
 271     @Override
 272     protected ClientRequestTracker findClientRequestTracker(long logIndex) {
 273         for (ClientRequestTracker tracker : trackerList) {
 274             if (tracker.getIndex() == logIndex) {
 275                 return tracker;
 276             }
 277         }
 278         return null;
 279     }
 280
 281     @Override
 282     protected RaftActorBehavior handleRequestVoteReply(ActorRef sender,
 283         RequestVoteReply requestVoteReply) {
 284         return this;
 285     }
 286
 287     protected void beforeSendHeartbeat(){}
 288
 289     @Override
 290     public RaftActorBehavior handleMessage(ActorRef sender, Object originalMessage) {
 291         Preconditions.checkNotNull(sender, "sender should not be null");
 292
 293         Object message = fromSerializableMessage(originalMessage);
 294
 295         if (message instanceof RaftRPC) {
 296             RaftRPC rpc = (RaftRPC) message;
 297             // If RPC request or response contains term T > currentTerm:
 298             // set currentTerm = T, convert to follower (§5.1)
 299             // This applies to all RPC messages and responses
 300             if (rpc.getTerm() > context.getTermInformation().getCurrentTerm()) {
 301                 LOG.debug("{}: Term {} in \"{}\" message is greater than leader's term {} - switching to Follower",
 302                         logName(), rpc.getTerm(), rpc, context.getTermInformation().getCurrentTerm());
 303
 304                 context.getTermInformation().updateAndPersist(rpc.getTerm(), null);
 305
 306                 return switchBehavior(new Follower(context));
 307             }
 308         }
 309
 310         if (message instanceof SendHeartBeat) {
 311             beforeSendHeartbeat();
 312             sendHeartBeat();
 313             scheduleHeartBeat(context.getConfigParams().getHeartBeatInterval());
 314             return this;
 315
 316         } else if(message instanceof SendInstallSnapshot) {
 317             // received from RaftActor
 318             setSnapshot(Optional.of(((SendInstallSnapshot) message).getSnapshot()));
 319             sendInstallSnapshot();
 320
 321         } else if (message instanceof Replicate) {
 322             replicate((Replicate) message);
 323
 324         } else if (message instanceof InstallSnapshotReply){
 325             handleInstallSnapshotReply((InstallSnapshotReply) message);
 326
 327         }
 328
 329
 330         return super.handleMessage(sender, message);
 331     }
 332
 333     private void handleInstallSnapshotReply(InstallSnapshotReply reply) {
 334         LOG.debug("{}: handleInstallSnapshotReply: {}", logName(), reply);
 335
 336         String followerId = reply.getFollowerId();
 337         FollowerToSnapshot followerToSnapshot = mapFollowerToSnapshot.get(followerId);
 338
 339         if (followerToSnapshot == null) {
 340             LOG.error("{}: FollowerId {} in InstallSnapshotReply not known to Leader",
 341                     logName(), followerId);
 342             return;
 343         }
 344
 345         FollowerLogInformation followerLogInformation = followerToLog.get(followerId);
 346         followerLogInformation.markFollowerActive();
 347
 348         if (followerToSnapshot.getChunkIndex() == reply.getChunkIndex()) {
 349             boolean wasLastChunk = false;
 350             if (reply.isSuccess()) {
 351                 if(followerToSnapshot.isLastChunk(reply.getChunkIndex())) {
 352                     //this was the last chunk reply
 353                     if(LOG.isDebugEnabled()) {
 354                         LOG.debug("{}: InstallSnapshotReply received, " +
 355                                 "last chunk received, Chunk: {}. Follower: {} Setting nextIndex: {}",
 356                                 logName(), reply.getChunkIndex(), followerId,
 357                             context.getReplicatedLog().getSnapshotIndex() + 1
 358                         );
 359                     }
 360
 361                     followerLogInformation.setMatchIndex(
 362                         context.getReplicatedLog().getSnapshotIndex());
 363                     followerLogInformation.setNextIndex(
 364                         context.getReplicatedLog().getSnapshotIndex() + 1);
 365                     mapFollowerToSnapshot.remove(followerId);
 366
 367                     LOG.debug("{}: follower: {}, matchIndex set to {}, nextIndex set to {}",
 368                                 logName(), followerId, followerLogInformation.getMatchIndex(),
 369                                 followerLogInformation.getNextIndex());
 370
 371                     if (mapFollowerToSnapshot.isEmpty()) {
 372                         // once there are no pending followers receiving snapshots
 373                         // we can remove snapshot from the memory
 374                         setSnapshot(Optional.<ByteString>absent());
 375                     }
 376                     wasLastChunk = true;
 377
 378                 } else {
 379                     followerToSnapshot.markSendStatus(true);
 380                 }
 381             } else {
 382                 LOG.info("{}: InstallSnapshotReply received sending snapshot chunk failed, Will retry, Chunk: {}",
 383                         logName(), reply.getChunkIndex());
 384
 385                 followerToSnapshot.markSendStatus(false);
 386             }
 387
 388             if (!wasLastChunk && followerToSnapshot.canSendNextChunk()) {
 389                 ActorSelection followerActor = context.getPeerActorSelection(followerId);
 390                 if(followerActor != null) {
 391                     sendSnapshotChunk(followerActor, followerId);
 392                 }
 393             }
 394
 395         } else {
 396             LOG.error("{}: Chunk index {} in InstallSnapshotReply from follower {} does not match expected index {}",
 397                     logName(), reply.getChunkIndex(), followerId,
 398                     followerToSnapshot.getChunkIndex());
 399
 400             if(reply.getChunkIndex() == INVALID_CHUNK_INDEX){
 401                 // Since the Follower did not find this index to be valid we should reset the follower snapshot
 402                 // so that Installing the snapshot can resume from the beginning
 403                 followerToSnapshot.reset();
 404             }
 405         }
 406     }
 407
 408     private void replicate(Replicate replicate) {
 409         long logIndex = replicate.getReplicatedLogEntry().getIndex();
 410
 411         LOG.debug("{}: Replicate message: identifier: {}, logIndex: {}", logName(),
 412                 replicate.getIdentifier(), logIndex);
 413
 414         // Create a tracker entry we will use this later to notify the
 415         // client actor
 416         trackerList.add(
 417             new ClientRequestTrackerImpl(replicate.getClientActor(),
 418                 replicate.getIdentifier(),
 419                 logIndex)
 420         );
 421
 422         if (followerToLog.isEmpty()) {
 423             context.setCommitIndex(logIndex);
 424             applyLogToStateMachine(logIndex);
 425         } else {
 426             sendAppendEntries(0, false);
 427         }
 428     }
 429
 430     private void sendAppendEntries(long timeSinceLastActivityInterval, boolean isHeartbeat) {
 431         // Send an AppendEntries to all followers
 432         for (Entry<String, FollowerLogInformation> e : followerToLog.entrySet()) {
 433             final String followerId = e.getKey();
 434             final FollowerLogInformation followerLogInformation = e.getValue();
 435             // This checks helps not to send a repeat message to the follower
 436             if(!followerLogInformation.isFollowerActive() ||
 437                     followerLogInformation.timeSinceLastActivity() >= timeSinceLastActivityInterval) {
 438                 sendUpdatesToFollower(followerId, followerLogInformation, true, isHeartbeat);
 439             }
 440         }
 441     }
 442
 443     /**
 444      *
 445      * This method checks if any update needs to be sent to the given follower. This includes append log entries,
 446      * sending next snapshot chunk, and initiating a snapshot.
 447      * @return true if any update is sent, false otherwise
 448      */
 449
 450     private void sendUpdatesToFollower(String followerId, FollowerLogInformation followerLogInformation,
 451                                        boolean sendHeartbeat, boolean isHeartbeat) {
 452
 453         ActorSelection followerActor = context.getPeerActorSelection(followerId);
 454         if (followerActor != null) {
 455             long followerNextIndex = followerLogInformation.getNextIndex();
 456             boolean isFollowerActive = followerLogInformation.isFollowerActive();
 457
 458             if (mapFollowerToSnapshot.get(followerId) != null) {
 459                 // if install snapshot is in process , then sent next chunk if possible
 460                 if (isFollowerActive && mapFollowerToSnapshot.get(followerId).canSendNextChunk()) {
 461                     sendSnapshotChunk(followerActor, followerId);
 462                 } else if(sendHeartbeat) {
 463                     // we send a heartbeat even if we have not received a reply for the last chunk
 464                     sendAppendEntriesToFollower(followerActor, followerLogInformation.getNextIndex(),
 465                         Collections.<ReplicatedLogEntry>emptyList(), followerId);
 466                 }
 467             } else {
 468                 long leaderLastIndex = context.getReplicatedLog().lastIndex();
 469                 long leaderSnapShotIndex = context.getReplicatedLog().getSnapshotIndex();
 470
 471                 if(!isHeartbeat || LOG.isTraceEnabled()) {
 472                     LOG.debug("{}: Checking sendAppendEntries for follower {}, leaderLastIndex: {}, leaderSnapShotIndex: {}",
 473                             logName(), followerId, leaderLastIndex, leaderSnapShotIndex);
 474                 }
 475
 476                 if (isFollowerActive && context.getReplicatedLog().isPresent(followerNextIndex)) {
 477
 478                     LOG.debug("{}: sendAppendEntries: {} is present for follower {}", logName(),
 479                             followerNextIndex, followerId);
 480
 481                     // FIXME : Sending one entry at a time
 482                     final List<ReplicatedLogEntry> entries = context.getReplicatedLog().getFrom(followerNextIndex, 1);
 483
 484                     sendAppendEntriesToFollower(followerActor, followerNextIndex, entries, followerId);
 485
 486                 } else if (isFollowerActive && followerNextIndex >= 0 &&
 487                     leaderLastIndex > followerNextIndex && !context.isSnapshotCaptureInitiated()) {
 488                     // if the followers next index is not present in the leaders log, and
 489                     // if the follower is just not starting and if leader's index is more than followers index
 490                     // then snapshot should be sent
 491
 492                     if (LOG.isDebugEnabled()) {
 493                         LOG.debug(String.format("%s: InitiateInstallSnapshot to follower: %s," +
 494                                     "follower-nextIndex: %d, leader-snapshot-index: %d,  " +
 495                                     "leader-last-index: %d", logName(), followerId,
 496                                     followerNextIndex, leaderSnapShotIndex, leaderLastIndex));
 497                     }
 498
 499                     // Send heartbeat to follower whenever install snapshot is initiated.
 500                     sendAppendEntriesToFollower(followerActor, followerLogInformation.getNextIndex(),
 501                             Collections.<ReplicatedLogEntry>emptyList(), followerId);
 502
 503                     initiateCaptureSnapshot(followerId, followerNextIndex);
 504
 505                 } else if(sendHeartbeat) {
 506                     //we send an AppendEntries, even if the follower is inactive
 507                     // in-order to update the followers timestamp, in case it becomes active again
 508                     sendAppendEntriesToFollower(followerActor, followerLogInformation.getNextIndex(),
 509                         Collections.<ReplicatedLogEntry>emptyList(), followerId);
 510                 }
 511
 512             }
 513         }
 514     }
 515
 516     private void sendAppendEntriesToFollower(ActorSelection followerActor, long followerNextIndex,
 517         List<ReplicatedLogEntry> entries, String followerId) {
 518         AppendEntries appendEntries = new AppendEntries(currentTerm(), context.getId(),
 519             prevLogIndex(followerNextIndex),
 520             prevLogTerm(followerNextIndex), entries,
 521             context.getCommitIndex(), super.getReplicatedToAllIndex());
 522
 523         if(!entries.isEmpty() || LOG.isTraceEnabled()) {
 524             LOG.debug("{}: Sending AppendEntries to follower {}: {}", logName(), followerId,
 525                     appendEntries);
 526         }
 527
 528         followerActor.tell(appendEntries.toSerializable(), actor());
 529     }
 530
 531     /**
 532      * Install Snapshot works as follows
 533      * 1. Leader initiates the capture snapshot by sending a CaptureSnapshot message to actor
 534      * 2. RaftActor on receipt of the CaptureSnapshotReply (from Shard), stores the received snapshot in the replicated log
 535      * and makes a call to Leader's handleMessage , with SendInstallSnapshot message.
 536      * 3. Leader , picks the snapshot from im-mem ReplicatedLog and sends it in chunks to the Follower
 537      * 4. On complete, Follower sends back a InstallSnapshotReply.
 538      * 5. On receipt of the InstallSnapshotReply for the last chunk, Leader marks the install complete for that follower
 539      * and replenishes the memory by deleting the snapshot in Replicated log.
 540      * 6. If another follower requires a snapshot and a snapshot has been collected (via CaptureSnapshotReply)
 541      * then send the existing snapshot in chunks to the follower.
 542      * @param followerId
 543      * @param followerNextIndex
 544      */
 545     private void initiateCaptureSnapshot(String followerId, long followerNextIndex) {
 546         if (!context.getReplicatedLog().isPresent(followerNextIndex) &&
 547                 context.getReplicatedLog().isInSnapshot(followerNextIndex)) {
 548
 549             if (snapshot.isPresent()) {
 550                 // if a snapshot is present in the memory, most likely another install is in progress
 551                 // no need to capture snapshot.
 552                 // This could happen if another follower needs an install when one is going on.
 553                 final ActorSelection followerActor = context.getPeerActorSelection(followerId);
 554                 sendSnapshotChunk(followerActor, followerId);
 555
 556             } else if (!context.isSnapshotCaptureInitiated()) {
 557
 558                 LOG.info("{}: Initiating Snapshot Capture to Install Snapshot, Leader:{}", logName(), getLeaderId());
 559                 ReplicatedLogEntry lastAppliedEntry = context.getReplicatedLog().get(context.getLastApplied());
 560                 long lastAppliedIndex = -1;
 561                 long lastAppliedTerm = -1;
 562
 563                 if (lastAppliedEntry != null) {
 564                     lastAppliedIndex = lastAppliedEntry.getIndex();
 565                     lastAppliedTerm = lastAppliedEntry.getTerm();
 566                 } else if (context.getReplicatedLog().getSnapshotIndex() > -1) {
 567                     lastAppliedIndex = context.getReplicatedLog().getSnapshotIndex();
 568                     lastAppliedTerm = context.getReplicatedLog().getSnapshotTerm();
 569                 }
 570
 571                 boolean isInstallSnapshotInitiated = true;
 572                 long replicatedToAllIndex = super.getReplicatedToAllIndex();
 573                 ReplicatedLogEntry replicatedToAllEntry = context.getReplicatedLog().get(replicatedToAllIndex);
 574                 actor().tell(new CaptureSnapshot(lastIndex(), lastTerm(), lastAppliedIndex, lastAppliedTerm,
 575                     (replicatedToAllEntry != null ? replicatedToAllEntry.getIndex() : -1),
 576                     (replicatedToAllEntry != null ? replicatedToAllEntry.getTerm() : -1),
 577                     isInstallSnapshotInitiated), actor());
 578                 context.setSnapshotCaptureInitiated(true);
 579             }
 580         }
 581     }
 582
 583
 584     private void sendInstallSnapshot() {
 585         LOG.debug("{}: sendInstallSnapshot", logName());
 586         for (Entry<String, FollowerLogInformation> e : followerToLog.entrySet()) {
 587             ActorSelection followerActor = context.getPeerActorSelection(e.getKey());
 588
 589             if (followerActor != null) {
 590                 long nextIndex = e.getValue().getNextIndex();
 591
 592                 if (!context.getReplicatedLog().isPresent(nextIndex) &&
 593                     context.getReplicatedLog().isInSnapshot(nextIndex)) {
 594                     sendSnapshotChunk(followerActor, e.getKey());
 595                 }
 596             }
 597         }
 598     }
 599
 600     /**
 601      *  Sends a snapshot chunk to a given follower
 602      *  InstallSnapshot should qualify as a heartbeat too.
 603      */
 604     private void sendSnapshotChunk(ActorSelection followerActor, String followerId) {
 605         try {
 606             if (snapshot.isPresent()) {
 607                 ByteString nextSnapshotChunk = getNextSnapshotChunk(followerId,snapshot.get());
 608
 609                 // Note: the previous call to getNextSnapshotChunk has the side-effect of adding
 610                 // followerId to the followerToSnapshot map.
 611                 FollowerToSnapshot followerToSnapshot = mapFollowerToSnapshot.get(followerId);
 612
 613                 followerActor.tell(
 614                     new InstallSnapshot(currentTerm(), context.getId(),
 615                         context.getReplicatedLog().getSnapshotIndex(),
 616                         context.getReplicatedLog().getSnapshotTerm(),
 617                         nextSnapshotChunk,
 618                         followerToSnapshot.incrementChunkIndex(),
 619                         followerToSnapshot.getTotalChunks(),
 620                         Optional.of(followerToSnapshot.getLastChunkHashCode())
 621                     ).toSerializable(),
 622                     actor()
 623                 );
 624                 LOG.info("{}: InstallSnapshot sent to follower {}, Chunk: {}/{}",
 625                         logName(), followerActor.path(),
 626                         followerToSnapshot.getChunkIndex(),
 627                         followerToSnapshot.getTotalChunks());
 628             }
 629         } catch (IOException e) {
 630             LOG.error("{}: InstallSnapshot failed for Leader.", logName(), e);
 631         }
 632     }
 633
 634     /**
 635      * Acccepts snaphot as ByteString, enters into map for future chunks
 636      * creates and return a ByteString chunk
 637      */
 638     private ByteString getNextSnapshotChunk(String followerId, ByteString snapshotBytes) throws IOException {
 639         FollowerToSnapshot followerToSnapshot = mapFollowerToSnapshot.get(followerId);
 640         if (followerToSnapshot == null) {
 641             followerToSnapshot = new FollowerToSnapshot(snapshotBytes);
 642             mapFollowerToSnapshot.put(followerId, followerToSnapshot);
 643         }
 644         ByteString nextChunk = followerToSnapshot.getNextChunk();
 645
 646         LOG.debug("{}: next snapshot chunk size for follower {}: {}", logName(), followerId, nextChunk.size());
 647
 648         return nextChunk;
 649     }
 650
 651     private void sendHeartBeat() {
 652         if (!followerToLog.isEmpty()) {
 653             LOG.trace("{}: Sending heartbeat", logName());
 654             sendAppendEntries(context.getConfigParams().getHeartBeatInterval().toMillis(), true);
 655         }
 656     }
 657
 658     private void stopHeartBeat() {
 659         if (heartbeatSchedule != null && !heartbeatSchedule.isCancelled()) {
 660             heartbeatSchedule.cancel();
 661         }
 662     }
 663
 664     private void scheduleHeartBeat(FiniteDuration interval) {
 665         if (followerToLog.isEmpty()) {
 666             // Optimization - do not bother scheduling a heartbeat as there are
 667             // no followers
 668             return;
 669         }
 670
 671         stopHeartBeat();
 672
 673         // Schedule a heartbeat. When the scheduler triggers a SendHeartbeat
 674         // message is sent to itself.
 675         // Scheduling the heartbeat only once here because heartbeats do not
 676         // need to be sent if there are other messages being sent to the remote
 677         // actor.
 678         heartbeatSchedule = context.getActorSystem().scheduler().scheduleOnce(
 679             interval, context.getActor(), new SendHeartBeat(),
 680             context.getActorSystem().dispatcher(), context.getActor());
 681     }
 682
 683     @Override
 684     public void close() throws Exception {
 685         stopHeartBeat();
 686     }
 687
 688     @Override
 689     public String getLeaderId() {
 690         return context.getId();
 691     }
 692
 693     protected boolean isLeaderIsolated() {
 694         int minPresent = minIsolatedLeaderPeerCount;
 695         for (FollowerLogInformation followerLogInformation : followerToLog.values()) {
 696             if (followerLogInformation.isFollowerActive()) {
 697                 --minPresent;
 698                 if (minPresent == 0) {
 699                     break;
 700                 }
 701             }
 702         }
 703         return (minPresent != 0);
 704     }
 705
 706     /**
 707      * Encapsulates the snapshot bytestring and handles the logic of sending
 708      * snapshot chunks
 709      */
 710     protected class FollowerToSnapshot {
 711         private final ByteString snapshotBytes;
 712         private int offset = 0;
 713         // the next snapshot chunk is sent only if the replyReceivedForOffset matches offset
 714         private int replyReceivedForOffset;
 715         // if replyStatus is false, the previous chunk is attempted
 716         private boolean replyStatus = false;
 717         private int chunkIndex;
 718         private final int totalChunks;
 719         private int lastChunkHashCode = AbstractLeader.INITIAL_LAST_CHUNK_HASH_CODE;
 720         private int nextChunkHashCode = AbstractLeader.INITIAL_LAST_CHUNK_HASH_CODE;
 721
 722         public FollowerToSnapshot(ByteString snapshotBytes) {
 723             this.snapshotBytes = snapshotBytes;
 724             int size = snapshotBytes.size();
 725             totalChunks = ( size / context.getConfigParams().getSnapshotChunkSize()) +
 726                 ((size % context.getConfigParams().getSnapshotChunkSize()) > 0 ? 1 : 0);
 727             if(LOG.isDebugEnabled()) {
 728                 LOG.debug("{}: Snapshot {} bytes, total chunks to send:{}",
 729                         logName(), size, totalChunks);
 730             }
 731             replyReceivedForOffset = -1;
 732             chunkIndex = AbstractLeader.FIRST_CHUNK_INDEX;
 733         }
 734
 735         public ByteString getSnapshotBytes() {
 736             return snapshotBytes;
 737         }
 738
 739         public int incrementOffset() {
 740             if(replyStatus) {
 741                 // if prev chunk failed, we would want to sent the same chunk again
 742                 offset = offset + context.getConfigParams().getSnapshotChunkSize();
 743             }
 744             return offset;
 745         }
 746
 747         public int incrementChunkIndex() {
 748             if (replyStatus) {
 749                 // if prev chunk failed, we would want to sent the same chunk again
 750                 chunkIndex =  chunkIndex + 1;
 751             }
 752             return chunkIndex;
 753         }
 754
 755         public int getChunkIndex() {
 756             return chunkIndex;
 757         }
 758
 759         public int getTotalChunks() {
 760             return totalChunks;
 761         }
 762
 763         public boolean canSendNextChunk() {
 764             // we only send a false if a chunk is sent but we have not received a reply yet
 765             return replyReceivedForOffset == offset;
 766         }
 767
 768         public boolean isLastChunk(int chunkIndex) {
 769             return totalChunks == chunkIndex;
 770         }
 771
 772         public void markSendStatus(boolean success) {
 773             if (success) {
 774                 // if the chunk sent was successful
 775                 replyReceivedForOffset = offset;
 776                 replyStatus = true;
 777                 lastChunkHashCode = nextChunkHashCode;
 778             } else {
 779                 // if the chunk sent was failure
 780                 replyReceivedForOffset = offset;
 781                 replyStatus = false;
 782             }
 783         }
 784
 785         public ByteString getNextChunk() {
 786             int snapshotLength = getSnapshotBytes().size();
 787             int start = incrementOffset();
 788             int size = context.getConfigParams().getSnapshotChunkSize();
 789             if (context.getConfigParams().getSnapshotChunkSize() > snapshotLength) {
 790                 size = snapshotLength;
 791             } else {
 792                 if ((start + context.getConfigParams().getSnapshotChunkSize()) > snapshotLength) {
 793                     size = snapshotLength - start;
 794                 }
 795             }
 796
 797
 798             LOG.debug("{}: Next chunk: length={}, offset={},size={}", logName(),
 799                     snapshotLength, start, size);
 800
 801             ByteString substring = getSnapshotBytes().substring(start, start + size);
 802             nextChunkHashCode = substring.hashCode();
 803             return substring;
 804         }
 805
 806         /**
 807          * reset should be called when the Follower needs to be sent the snapshot from the beginning
 808          */
 809         public void reset(){
 810             offset = 0;
 811             replyStatus = false;
 812             replyReceivedForOffset = offset;
 813             chunkIndex = AbstractLeader.FIRST_CHUNK_INDEX;
 814             lastChunkHashCode = AbstractLeader.INITIAL_LAST_CHUNK_HASH_CODE;
 815         }
 816
 817         public int getLastChunkHashCode() {
 818             return lastChunkHashCode;
 819         }
 820     }
 821
 822     // called from example-actor for printing the follower-states
 823     public String printFollowerStates() {
 824         final StringBuilder sb = new StringBuilder();
 825
 826         sb.append('[');
 827         for (FollowerLogInformation followerLogInformation : followerToLog.values()) {
 828             sb.append('{');
 829             sb.append(followerLogInformation.getId());
 830             sb.append(" state:");
 831             sb.append(followerLogInformation.isFollowerActive());
 832             sb.append("},");
 833         }
 834         sb.append(']');
 835
 836         return sb.toString();
 837     }
 838
 839     @VisibleForTesting
 840     public FollowerLogInformation getFollower(String followerId) {
 841         return followerToLog.get(followerId);
 842     }
 843
 844     @VisibleForTesting
 845     protected void setFollowerSnapshot(String followerId, FollowerToSnapshot snapshot) {
 846         mapFollowerToSnapshot.put(followerId, snapshot);
 847     }
 848
 849     @VisibleForTesting
 850     public int followerSnapshotSize() {
 851         return mapFollowerToSnapshot.size();
 852     }
 853
 854     @VisibleForTesting
 855     public int followerLogSize() {
 856         return followerToLog.size();
 857     }
 858 }