opendaylight/md-sal/sal-akka-raft/src/main/java/org/opendaylight/controller/cluster/raft/behaviors/Follower.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 package org.opendaylight.controller.cluster.raft.behaviors;
   9
  10 import akka.actor.ActorRef;
  11 import akka.actor.ActorSelection;
  12 import akka.actor.Address;
  13 import akka.cluster.Cluster;
  14 import akka.cluster.ClusterEvent.CurrentClusterState;
  15 import akka.cluster.Member;
  16 import akka.cluster.MemberStatus;
  17 import com.google.common.annotations.VisibleForTesting;
  18 import com.google.common.base.Stopwatch;
  19 import edu.umd.cs.findbugs.annotations.SuppressFBWarnings;
  20 import java.io.IOException;
  21 import java.util.HashSet;
  22 import java.util.List;
  23 import java.util.Optional;
  24 import java.util.Set;
  25 import java.util.concurrent.TimeUnit;
  26 import java.util.concurrent.atomic.AtomicBoolean;
  27 import java.util.function.Consumer;
  28 import org.eclipse.jdt.annotation.Nullable;
  29 import org.opendaylight.controller.cluster.messaging.MessageAssembler;
  30 import org.opendaylight.controller.cluster.raft.RaftActorContext;
  31 import org.opendaylight.controller.cluster.raft.RaftState;
  32 import org.opendaylight.controller.cluster.raft.ReplicatedLogEntry;
  33 import org.opendaylight.controller.cluster.raft.base.messages.ApplySnapshot;
  34 import org.opendaylight.controller.cluster.raft.base.messages.ApplyState;
  35 import org.opendaylight.controller.cluster.raft.base.messages.ElectionTimeout;
  36 import org.opendaylight.controller.cluster.raft.base.messages.TimeoutNow;
  37 import org.opendaylight.controller.cluster.raft.messages.AppendEntries;
  38 import org.opendaylight.controller.cluster.raft.messages.AppendEntriesReply;
  39 import org.opendaylight.controller.cluster.raft.messages.InstallSnapshot;
  40 import org.opendaylight.controller.cluster.raft.messages.InstallSnapshotReply;
  41 import org.opendaylight.controller.cluster.raft.messages.RaftRPC;
  42 import org.opendaylight.controller.cluster.raft.messages.RequestVote;
  43 import org.opendaylight.controller.cluster.raft.messages.RequestVoteReply;
  44 import org.opendaylight.controller.cluster.raft.persisted.ServerConfigurationPayload;
  45 import org.opendaylight.controller.cluster.raft.persisted.Snapshot;
  46
  47 /**
  48  * The behavior of a RaftActor in the Follower raft state.
  49  * <ul>
  50  * <li> Respond to RPCs from candidates and leaders
  51  * <li> If election timeout elapses without receiving AppendEntries
  52  * RPC from current leader or granting vote to candidate:
  53  * convert to candidate
  54  * </ul>
  55  */
  56 // Non-final for testing
  57 public class Follower extends AbstractRaftActorBehavior {
  58     private static final long MAX_ELECTION_TIMEOUT_FACTOR = 18;
  59
  60     private final SyncStatusTracker initialSyncStatusTracker;
  61
  62     private final MessageAssembler appendEntriesMessageAssembler;
  63
  64     private final Stopwatch lastLeaderMessageTimer = Stopwatch.createStarted();
  65     private SnapshotTracker snapshotTracker = null;
  66     private String leaderId;
  67     private short leaderPayloadVersion;
  68
  69     public Follower(final RaftActorContext context) {
  70         this(context, null, (short)-1);
  71     }
  72
  73     @SuppressFBWarnings(value = "MC_OVERRIDABLE_METHOD_CALL_IN_CONSTRUCTOR",
  74         justification = "electionDuration() is not final for Candidate override")
  75     public Follower(final RaftActorContext context, final String initialLeaderId,
  76             final short initialLeaderPayloadVersion) {
  77         super(context, RaftState.Follower);
  78         leaderId = initialLeaderId;
  79         leaderPayloadVersion = initialLeaderPayloadVersion;
  80
  81         initialSyncStatusTracker = new SyncStatusTracker(context.getActor(), getId(), context.getConfigParams()
  82             .getSyncIndexThreshold());
  83
  84         appendEntriesMessageAssembler = MessageAssembler.builder().logContext(logName())
  85                 .fileBackedStreamFactory(context.getFileBackedOutputStreamFactory())
  86                 .assembledMessageCallback((message, sender) -> handleMessage(sender, message)).build();
  87
  88         if (context.getPeerIds().isEmpty() && getLeaderId() == null) {
  89             actor().tell(TimeoutNow.INSTANCE, actor());
  90         } else {
  91             scheduleElection(electionDuration());
  92         }
  93     }
  94
  95     @Override
  96     public final String getLeaderId() {
  97         return leaderId;
  98     }
  99
 100     @VisibleForTesting
 101     protected final void setLeaderId(final @Nullable String leaderId) {
 102         this.leaderId = leaderId;
 103     }
 104
 105     @Override
 106     public short getLeaderPayloadVersion() {
 107         return leaderPayloadVersion;
 108     }
 109
 110     @VisibleForTesting
 111     protected final void setLeaderPayloadVersion(final short leaderPayloadVersion) {
 112         this.leaderPayloadVersion = leaderPayloadVersion;
 113     }
 114
 115     private void restartLastLeaderMessageTimer() {
 116         if (lastLeaderMessageTimer.isRunning()) {
 117             lastLeaderMessageTimer.reset();
 118         }
 119
 120         lastLeaderMessageTimer.start();
 121     }
 122
 123     private boolean isLogEntryPresent(final long index) {
 124         if (context.getReplicatedLog().isInSnapshot(index)) {
 125             return true;
 126         }
 127
 128         ReplicatedLogEntry entry = context.getReplicatedLog().get(index);
 129         return entry != null;
 130
 131     }
 132
 133     private void updateInitialSyncStatus(final long currentLeaderCommit, final String newLeaderId) {
 134         initialSyncStatusTracker.update(newLeaderId, currentLeaderCommit, context.getCommitIndex());
 135     }
 136
 137     @Override
 138     protected RaftActorBehavior handleAppendEntries(final ActorRef sender, final AppendEntries appendEntries) {
 139         int numLogEntries = appendEntries.getEntries().size();
 140         if (log.isTraceEnabled()) {
 141             log.trace("{}: handleAppendEntries: {}", logName(), appendEntries);
 142         } else if (log.isDebugEnabled() && numLogEntries > 0) {
 143             log.debug("{}: handleAppendEntries: {}", logName(), appendEntries);
 144         }
 145
 146         if (snapshotTracker != null && !snapshotTracker.getLeaderId().equals(appendEntries.getLeaderId())) {
 147             log.debug("{}: snapshot install is in progress but the prior snapshot leaderId {} does not match the "
 148                 + "AppendEntries leaderId {}", logName(), snapshotTracker.getLeaderId(), appendEntries.getLeaderId());
 149             closeSnapshotTracker();
 150         }
 151
 152         if (snapshotTracker != null || context.getSnapshotManager().isApplying()) {
 153             // if snapshot install is in progress, follower should just acknowledge append entries with a reply.
 154             AppendEntriesReply reply = new AppendEntriesReply(context.getId(), currentTerm(), true,
 155                     lastIndex(), lastTerm(), context.getPayloadVersion(), false, needsLeaderAddress(),
 156                     appendEntries.getLeaderRaftVersion());
 157
 158             log.debug("{}: snapshot install is in progress, replying immediately with {}", logName(), reply);
 159             sender.tell(reply, actor());
 160
 161             return this;
 162         }
 163
 164         // If we got here then we do appear to be talking to the leader
 165         leaderId = appendEntries.getLeaderId();
 166         leaderPayloadVersion = appendEntries.getPayloadVersion();
 167
 168         final var leaderAddress = appendEntries.leaderAddress();
 169         if (leaderAddress != null) {
 170             log.debug("New leader address: {}", leaderAddress);
 171             context.setPeerAddress(leaderId, leaderAddress);
 172             context.getConfigParams().getPeerAddressResolver().setResolved(leaderId, leaderAddress);
 173         }
 174
 175         // First check if the logs are in sync or not
 176         if (isOutOfSync(appendEntries, sender)) {
 177             updateInitialSyncStatus(appendEntries.getLeaderCommit(), appendEntries.getLeaderId());
 178             return this;
 179         }
 180
 181         if (!processNewEntries(appendEntries, sender)) {
 182             updateInitialSyncStatus(appendEntries.getLeaderCommit(), appendEntries.getLeaderId());
 183             return this;
 184         }
 185
 186         long lastIndex = lastIndex();
 187         long prevCommitIndex = context.getCommitIndex();
 188
 189         // If leaderCommit > commitIndex, set commitIndex = min(leaderCommit, index of last new entry)
 190         if (appendEntries.getLeaderCommit() > prevCommitIndex) {
 191             context.setCommitIndex(Math.min(appendEntries.getLeaderCommit(), lastIndex));
 192         }
 193
 194         if (prevCommitIndex != context.getCommitIndex()) {
 195             log.debug("{}: Commit index set to {}", logName(), context.getCommitIndex());
 196         }
 197
 198         AppendEntriesReply reply = new AppendEntriesReply(context.getId(), currentTerm(), true,
 199                 lastIndex, lastTerm(), context.getPayloadVersion(), false, needsLeaderAddress(),
 200                 appendEntries.getLeaderRaftVersion());
 201
 202         if (log.isTraceEnabled()) {
 203             log.trace("{}: handleAppendEntries returning : {}", logName(), reply);
 204         } else if (log.isDebugEnabled() && numLogEntries > 0) {
 205             log.debug("{}: handleAppendEntries returning : {}", logName(), reply);
 206         }
 207
 208         // Reply to the leader before applying any previous state so as not to hold up leader consensus.
 209         sender.tell(reply, actor());
 210
 211         updateInitialSyncStatus(appendEntries.getLeaderCommit(), appendEntries.getLeaderId());
 212
 213         // If leaderCommit > lastApplied, increment lastApplied and apply log[lastApplied] to state machine (§5.3).
 214         // lastApplied can be equal to lastIndex.
 215         if (appendEntries.getLeaderCommit() > context.getLastApplied() && context.getLastApplied() < lastIndex) {
 216             if (log.isDebugEnabled()) {
 217                 log.debug("{}: applyLogToStateMachine, appendEntries.getLeaderCommit(): {}, "
 218                         + "context.getLastApplied(): {}, lastIndex(): {}", logName(),
 219                     appendEntries.getLeaderCommit(), context.getLastApplied(), lastIndex);
 220             }
 221
 222             applyLogToStateMachine(appendEntries.getLeaderCommit());
 223         }
 224
 225         if (!context.getSnapshotManager().isCapturing()) {
 226             super.performSnapshotWithoutCapture(appendEntries.getReplicatedToAllIndex());
 227         }
 228
 229         appendEntriesMessageAssembler.checkExpiredAssembledMessageState();
 230
 231         return this;
 232     }
 233
 234     private boolean processNewEntries(final AppendEntries appendEntries, final ActorRef sender) {
 235         int numLogEntries = appendEntries.getEntries().size();
 236         if (numLogEntries == 0) {
 237             return true;
 238         }
 239
 240         log.debug("{}: Number of entries to be appended = {}", logName(), numLogEntries);
 241
 242         long lastIndex = lastIndex();
 243         int addEntriesFrom = 0;
 244
 245         // First check for conflicting entries. If an existing entry conflicts with a new one (same index but different
 246         // term), delete the existing entry and all that follow it (§5.3)
 247         if (context.getReplicatedLog().size() > 0) {
 248             // Find the entry up until the one that is not in the follower's log
 249             for (int i = 0;i < numLogEntries; i++, addEntriesFrom++) {
 250                 ReplicatedLogEntry matchEntry = appendEntries.getEntries().get(i);
 251
 252                 if (!isLogEntryPresent(matchEntry.getIndex())) {
 253                     // newEntry not found in the log
 254                     break;
 255                 }
 256
 257                 long existingEntryTerm = getLogEntryTerm(matchEntry.getIndex());
 258
 259                 log.debug("{}: matchEntry {} is present: existingEntryTerm: {}", logName(), matchEntry,
 260                         existingEntryTerm);
 261
 262                 // existingEntryTerm == -1 means it's in the snapshot and not in the log. We don't know
 263                 // what the term was so we'll assume it matches.
 264                 if (existingEntryTerm == -1 || existingEntryTerm == matchEntry.getTerm()) {
 265                     continue;
 266                 }
 267
 268                 if (!context.getRaftPolicy().applyModificationToStateBeforeConsensus()) {
 269                     log.info("{}: Removing entries from log starting at {}, commitIndex: {}, lastApplied: {}",
 270                             logName(), matchEntry.getIndex(), context.getCommitIndex(), context.getLastApplied());
 271
 272                     // Entries do not match so remove all subsequent entries but only if the existing entries haven't
 273                     // been applied to the state yet.
 274                     if (matchEntry.getIndex() <= context.getLastApplied()
 275                             || !context.getReplicatedLog().removeFromAndPersist(matchEntry.getIndex())) {
 276                         // Could not remove the entries - this means the matchEntry index must be in the
 277                         // snapshot and not the log. In this case the prior entries are part of the state
 278                         // so we must send back a reply to force a snapshot to completely re-sync the
 279                         // follower's log and state.
 280
 281                         log.info("{}: Could not remove entries - sending reply to force snapshot", logName());
 282                         sender.tell(new AppendEntriesReply(context.getId(), currentTerm(), false, lastIndex,
 283                                 lastTerm(), context.getPayloadVersion(), true, needsLeaderAddress(),
 284                                 appendEntries.getLeaderRaftVersion()), actor());
 285                         return false;
 286                     }
 287
 288                     break;
 289                 } else {
 290                     sender.tell(new AppendEntriesReply(context.getId(), currentTerm(), false, lastIndex,
 291                             lastTerm(), context.getPayloadVersion(), true, needsLeaderAddress(),
 292                             appendEntries.getLeaderRaftVersion()), actor());
 293                     return false;
 294                 }
 295             }
 296         }
 297
 298         lastIndex = lastIndex();
 299         log.debug("{}: After cleanup, lastIndex: {}, entries to be added from: {}", logName(), lastIndex,
 300                 addEntriesFrom);
 301
 302         // When persistence successfully completes for each new log entry appended, we need to determine if we
 303         // should capture a snapshot to compact the persisted log. shouldCaptureSnapshot tracks whether or not
 304         // one of the log entries has exceeded the log size threshold whereby a snapshot should be taken. However
 305         // we don't initiate the snapshot at that log entry but rather after the last log entry has been persisted.
 306         // This is done because subsequent log entries after the one that tripped the threshold may have been
 307         // applied to the state already, as the persistence callback occurs async, and we want those entries
 308         // purged from the persisted log as well.
 309         final AtomicBoolean shouldCaptureSnapshot = new AtomicBoolean(false);
 310         final Consumer<ReplicatedLogEntry> appendAndPersistCallback = logEntry -> {
 311             final List<ReplicatedLogEntry> entries = appendEntries.getEntries();
 312             final ReplicatedLogEntry lastEntryToAppend = entries.get(entries.size() - 1);
 313             if (shouldCaptureSnapshot.get() && logEntry == lastEntryToAppend) {
 314                 context.getSnapshotManager().capture(context.getReplicatedLog().last(), getReplicatedToAllIndex());
 315             }
 316         };
 317
 318         // Append any new entries not already in the log
 319         for (int i = addEntriesFrom; i < numLogEntries; i++) {
 320             ReplicatedLogEntry entry = appendEntries.getEntries().get(i);
 321
 322             log.debug("{}: Append entry to log {}", logName(), entry.getData());
 323
 324             context.getReplicatedLog().appendAndPersist(entry, appendAndPersistCallback, false);
 325
 326             shouldCaptureSnapshot.compareAndSet(false,
 327                     context.getReplicatedLog().shouldCaptureSnapshot(entry.getIndex()));
 328
 329             if (entry.getData() instanceof ServerConfigurationPayload serverConfiguration) {
 330                 context.updatePeerIds(serverConfiguration);
 331             }
 332         }
 333
 334         log.debug("{}: Log size is now {}", logName(), context.getReplicatedLog().size());
 335
 336         return true;
 337     }
 338
 339     private boolean isOutOfSync(final AppendEntries appendEntries, final ActorRef sender) {
 340
 341         final long lastIndex = lastIndex();
 342         if (lastIndex == -1 && appendEntries.getPrevLogIndex() != -1) {
 343
 344             // The follower's log is out of sync because the leader does have an entry at prevLogIndex and this
 345             // follower has no entries in it's log.
 346
 347             log.info("{}: The followers log is empty and the senders prevLogIndex is {}", logName(),
 348                 appendEntries.getPrevLogIndex());
 349
 350             sendOutOfSyncAppendEntriesReply(sender, false, appendEntries.getLeaderRaftVersion());
 351             return true;
 352         }
 353
 354         if (lastIndex > -1) {
 355             if (isLogEntryPresent(appendEntries.getPrevLogIndex())) {
 356                 final long leadersPrevLogTermInFollowersLogOrSnapshot =
 357                         getLogEntryOrSnapshotTerm(appendEntries.getPrevLogIndex());
 358                 if (leadersPrevLogTermInFollowersLogOrSnapshot != appendEntries.getPrevLogTerm()) {
 359
 360                     // The follower's log is out of sync because the Leader's prevLogIndex entry does exist
 361                     // in the follower's log or snapshot but it has a different term.
 362
 363                     log.info("{}: The prevLogIndex {} was found in the log but the term {} is not equal to the append "
 364                         + "entries prevLogTerm {} - lastIndex: {}, snapshotIndex: {}, snapshotTerm: {}", logName(),
 365                         appendEntries.getPrevLogIndex(), leadersPrevLogTermInFollowersLogOrSnapshot,
 366                         appendEntries.getPrevLogTerm(), lastIndex, context.getReplicatedLog().getSnapshotIndex(),
 367                         context.getReplicatedLog().getSnapshotTerm());
 368
 369                     sendOutOfSyncAppendEntriesReply(sender, false, appendEntries.getLeaderRaftVersion());
 370                     return true;
 371                 }
 372             } else if (appendEntries.getPrevLogIndex() != -1) {
 373
 374                 // The follower's log is out of sync because the Leader's prevLogIndex entry was not found in it's log
 375
 376                 log.info("{}: The log is not empty but the prevLogIndex {} was not found in it - lastIndex: {}, "
 377                         + "snapshotIndex: {}, snapshotTerm: {}", logName(), appendEntries.getPrevLogIndex(), lastIndex,
 378                         context.getReplicatedLog().getSnapshotIndex(), context.getReplicatedLog().getSnapshotTerm());
 379
 380                 sendOutOfSyncAppendEntriesReply(sender, false, appendEntries.getLeaderRaftVersion());
 381                 return true;
 382             }
 383         }
 384
 385         if (appendEntries.getPrevLogIndex() == -1 && appendEntries.getPrevLogTerm() == -1
 386                 && appendEntries.getReplicatedToAllIndex() != -1) {
 387             if (!isLogEntryPresent(appendEntries.getReplicatedToAllIndex())) {
 388                 // This append entry comes from a leader who has it's log aggressively trimmed and so does not have
 389                 // the previous entry in it's in-memory journal
 390
 391                 log.info("{}: Cannot append entries because the replicatedToAllIndex {} does not appear to be in the "
 392                         + "in-memory journal - lastIndex: {}, snapshotIndex: {}, snapshotTerm: {}", logName(),
 393                         appendEntries.getReplicatedToAllIndex(), lastIndex,
 394                         context.getReplicatedLog().getSnapshotIndex(), context.getReplicatedLog().getSnapshotTerm());
 395
 396                 sendOutOfSyncAppendEntriesReply(sender, false, appendEntries.getLeaderRaftVersion());
 397                 return true;
 398             }
 399
 400             final List<ReplicatedLogEntry> entries = appendEntries.getEntries();
 401             if (entries.size() > 0 && !isLogEntryPresent(entries.get(0).getIndex() - 1)) {
 402                 log.info("{}: Cannot append entries because the calculated previousIndex {} was not found in the "
 403                         + "in-memory journal - lastIndex: {}, snapshotIndex: {}, snapshotTerm: {}", logName(),
 404                         entries.get(0).getIndex() - 1, lastIndex, context.getReplicatedLog().getSnapshotIndex(),
 405                         context.getReplicatedLog().getSnapshotTerm());
 406
 407                 sendOutOfSyncAppendEntriesReply(sender, false, appendEntries.getLeaderRaftVersion());
 408                 return true;
 409             }
 410         }
 411
 412         return false;
 413     }
 414
 415     private void sendOutOfSyncAppendEntriesReply(final ActorRef sender, final boolean forceInstallSnapshot,
 416             final short leaderRaftVersion) {
 417         // We found that the log was out of sync so just send a negative reply.
 418         final AppendEntriesReply reply = new AppendEntriesReply(context.getId(), currentTerm(), false, lastIndex(),
 419                 lastTerm(), context.getPayloadVersion(), forceInstallSnapshot, needsLeaderAddress(),
 420                 leaderRaftVersion);
 421
 422         log.info("{}: Follower is out-of-sync so sending negative reply: {}", logName(), reply);
 423         sender.tell(reply, actor());
 424     }
 425
 426     private boolean needsLeaderAddress() {
 427         return context.getPeerAddress(leaderId) == null;
 428     }
 429
 430     @Override
 431     protected RaftActorBehavior handleAppendEntriesReply(final ActorRef sender,
 432         final AppendEntriesReply appendEntriesReply) {
 433         return this;
 434     }
 435
 436     @Override
 437     protected RaftActorBehavior handleRequestVoteReply(final ActorRef sender,
 438         final RequestVoteReply requestVoteReply) {
 439         return this;
 440     }
 441
 442     @Override
 443     final ApplyState getApplyStateFor(final ReplicatedLogEntry entry) {
 444         return new ApplyState(null, null, entry);
 445     }
 446
 447     @Override
 448     public RaftActorBehavior handleMessage(final ActorRef sender, final Object message) {
 449         if (message instanceof ElectionTimeout || message instanceof TimeoutNow) {
 450             return handleElectionTimeout(message);
 451         }
 452
 453         if (appendEntriesMessageAssembler.handleMessage(message, actor())) {
 454             return this;
 455         }
 456
 457         if (!(message instanceof RaftRPC rpc)) {
 458             // The rest of the processing requires the message to be a RaftRPC
 459             return null;
 460         }
 461
 462         // If RPC request or response contains term T > currentTerm:
 463         // set currentTerm = T, convert to follower (§5.1)
 464         // This applies to all RPC messages and responses
 465         if (rpc.getTerm() > context.getTermInformation().getCurrentTerm() && shouldUpdateTerm(rpc)) {
 466             log.info("{}: Term {} in \"{}\" message is greater than follower's term {} - updating term",
 467                 logName(), rpc.getTerm(), rpc, context.getTermInformation().getCurrentTerm());
 468
 469             context.getTermInformation().updateAndPersist(rpc.getTerm(), null);
 470         }
 471
 472         if (rpc instanceof InstallSnapshot installSnapshot) {
 473             handleInstallSnapshot(sender, installSnapshot);
 474             restartLastLeaderMessageTimer();
 475             scheduleElection(electionDuration());
 476             return this;
 477         }
 478
 479         if (!(rpc instanceof RequestVote requestVote) || canGrantVote(requestVote)) {
 480             restartLastLeaderMessageTimer();
 481             scheduleElection(electionDuration());
 482         }
 483
 484         return super.handleMessage(sender, rpc);
 485     }
 486
 487     private RaftActorBehavior handleElectionTimeout(final Object message) {
 488         // If the message is ElectionTimeout, verify we haven't actually seen a message from the leader
 489         // during the election timeout interval. It may that the election timer expired b/c this actor
 490         // was busy and messages got delayed, in which case leader messages would be backed up in the
 491         // queue but would be processed before the ElectionTimeout message and thus would restart the
 492         // lastLeaderMessageTimer.
 493         long lastLeaderMessageInterval = lastLeaderMessageTimer.elapsed(TimeUnit.MILLISECONDS);
 494         long electionTimeoutInMillis = context.getConfigParams().getElectionTimeOutInterval().toMillis();
 495         boolean noLeaderMessageReceived = !lastLeaderMessageTimer.isRunning()
 496                 || lastLeaderMessageInterval >= electionTimeoutInMillis;
 497
 498         if (canStartElection()) {
 499             if (message instanceof TimeoutNow) {
 500                 log.debug("{}: Received TimeoutNow - switching to Candidate", logName());
 501                 return internalSwitchBehavior(RaftState.Candidate);
 502             } else if (noLeaderMessageReceived) {
 503                 // Check the cluster state to see if the leader is known to be up before we go to Candidate.
 504                 // However if we haven't heard from the leader in a long time even though the cluster state
 505                 // indicates it's up then something is wrong - leader might be stuck indefinitely - so switch
 506                 // to Candidate,
 507                 long maxElectionTimeout = electionTimeoutInMillis * MAX_ELECTION_TIMEOUT_FACTOR;
 508                 if (isLeaderAvailabilityKnown() && lastLeaderMessageInterval < maxElectionTimeout) {
 509                     log.debug("{}: Received ElectionTimeout but leader appears to be available", logName());
 510                     scheduleElection(electionDuration());
 511                 } else if (isThisFollowerIsolated()) {
 512                     log.debug("{}: this follower is isolated. Do not switch to Candidate for now.", logName());
 513                     setLeaderId(null);
 514                     scheduleElection(electionDuration());
 515                 } else {
 516                     log.debug("{}: Received ElectionTimeout - switching to Candidate", logName());
 517                     return internalSwitchBehavior(RaftState.Candidate);
 518                 }
 519             } else {
 520                 log.debug("{}: Received ElectionTimeout but lastLeaderMessageInterval {} < election timeout {}",
 521                         logName(), lastLeaderMessageInterval, context.getConfigParams().getElectionTimeOutInterval());
 522                 scheduleElection(electionDuration());
 523             }
 524         } else if (message instanceof ElectionTimeout) {
 525             if (noLeaderMessageReceived) {
 526                 setLeaderId(null);
 527             }
 528
 529             scheduleElection(electionDuration());
 530         }
 531
 532         return this;
 533     }
 534
 535     private boolean isLeaderAvailabilityKnown() {
 536         if (leaderId == null) {
 537             return false;
 538         }
 539
 540         Optional<Cluster> cluster = context.getCluster();
 541         if (!cluster.isPresent()) {
 542             return false;
 543         }
 544
 545         ActorSelection leaderActor = context.getPeerActorSelection(leaderId);
 546         if (leaderActor == null) {
 547             return false;
 548         }
 549
 550         Address leaderAddress = leaderActor.anchorPath().address();
 551
 552         CurrentClusterState state = cluster.orElseThrow().state();
 553         Set<Member> unreachable = state.getUnreachable();
 554
 555         log.debug("{}: Checking for leader {} in the cluster unreachable set {}", logName(), leaderAddress,
 556                 unreachable);
 557
 558         for (Member m: unreachable) {
 559             if (leaderAddress.equals(m.address())) {
 560                 log.info("{}: Leader {} is unreachable", logName(), leaderAddress);
 561                 return false;
 562             }
 563         }
 564
 565         for (Member m: state.getMembers()) {
 566             if (leaderAddress.equals(m.address())) {
 567                 if (m.status() == MemberStatus.up() || m.status() == MemberStatus.weaklyUp()) {
 568                     log.debug("{}: Leader {} cluster status is {} - leader is available", logName(),
 569                             leaderAddress, m.status());
 570                     return true;
 571                 } else {
 572                     log.debug("{}: Leader {} cluster status is {} - leader is unavailable", logName(),
 573                             leaderAddress, m.status());
 574                     return false;
 575                 }
 576             }
 577         }
 578
 579         log.debug("{}: Leader {} not found in the cluster member set", logName(), leaderAddress);
 580
 581         return false;
 582     }
 583
 584     private boolean isThisFollowerIsolated() {
 585         final Optional<Cluster> maybeCluster = context.getCluster();
 586         if (!maybeCluster.isPresent()) {
 587             return false;
 588         }
 589
 590         final Cluster cluster = maybeCluster.orElseThrow();
 591         final Member selfMember = cluster.selfMember();
 592
 593         final CurrentClusterState state = cluster.state();
 594         final Set<Member> unreachable = state.getUnreachable();
 595         final Iterable<Member> members = state.getMembers();
 596
 597         log.debug("{}: Checking if this node is isolated in the cluster unreachable set {},"
 598                         + "all members {} self member: {}", logName(), unreachable, members, selfMember);
 599
 600         // no unreachable peers means we cannot be isolated
 601         if (unreachable.isEmpty()) {
 602             return false;
 603         }
 604
 605         final Set<Member> membersToCheck = new HashSet<>();
 606         members.forEach(membersToCheck::add);
 607
 608         membersToCheck.removeAll(unreachable);
 609
 610         // check if the only member not unreachable is us
 611         return membersToCheck.size() == 1 && membersToCheck.iterator().next().equals(selfMember);
 612     }
 613
 614     private void handleInstallSnapshot(final ActorRef sender, final InstallSnapshot installSnapshot) {
 615
 616         log.debug("{}: handleInstallSnapshot: {}", logName(), installSnapshot);
 617
 618         leaderId = installSnapshot.getLeaderId();
 619
 620         if (snapshotTracker == null) {
 621             snapshotTracker = new SnapshotTracker(log, installSnapshot.getTotalChunks(), installSnapshot.getLeaderId(),
 622                     context);
 623         }
 624
 625         updateInitialSyncStatus(installSnapshot.getLastIncludedIndex(), installSnapshot.getLeaderId());
 626
 627         try {
 628             final InstallSnapshotReply reply = new InstallSnapshotReply(
 629                     currentTerm(), context.getId(), installSnapshot.getChunkIndex(), true);
 630
 631             if (snapshotTracker.addChunk(installSnapshot.getChunkIndex(), installSnapshot.getData(),
 632                     installSnapshot.getLastChunkHashCode())) {
 633
 634                 log.info("{}: Snapshot installed from leader: {}", logName(), installSnapshot.getLeaderId());
 635
 636                 Snapshot snapshot = Snapshot.create(
 637                         context.getSnapshotManager().convertSnapshot(snapshotTracker.getSnapshotBytes()),
 638                         List.of(),
 639                         installSnapshot.getLastIncludedIndex(),
 640                         installSnapshot.getLastIncludedTerm(),
 641                         installSnapshot.getLastIncludedIndex(),
 642                         installSnapshot.getLastIncludedTerm(),
 643                         context.getTermInformation().getCurrentTerm(),
 644                         context.getTermInformation().getVotedFor(),
 645                         installSnapshot.getServerConfig().orElse(null));
 646
 647                 ApplySnapshot.Callback applySnapshotCallback = new ApplySnapshot.Callback() {
 648                     @Override
 649                     public void onSuccess() {
 650                         log.debug("{}: handleInstallSnapshot returning: {}", logName(), reply);
 651
 652                         sender.tell(reply, actor());
 653                     }
 654
 655                     @Override
 656                     public void onFailure() {
 657                         sender.tell(new InstallSnapshotReply(currentTerm(), context.getId(), -1, false), actor());
 658                     }
 659                 };
 660
 661                 actor().tell(new ApplySnapshot(snapshot, applySnapshotCallback), actor());
 662
 663                 closeSnapshotTracker();
 664             } else {
 665                 log.debug("{}: handleInstallSnapshot returning: {}", logName(), reply);
 666
 667                 sender.tell(reply, actor());
 668             }
 669         } catch (IOException e) {
 670             log.debug("{}: Exception in InstallSnapshot of follower", logName(), e);
 671
 672             sender.tell(new InstallSnapshotReply(currentTerm(), context.getId(), -1, false), actor());
 673
 674             closeSnapshotTracker();
 675         }
 676     }
 677
 678     private void closeSnapshotTracker() {
 679         if (snapshotTracker != null) {
 680             snapshotTracker.close();
 681             snapshotTracker = null;
 682         }
 683     }
 684
 685     @Override
 686     public void close() {
 687         closeSnapshotTracker();
 688         stopElection();
 689         appendEntriesMessageAssembler.close();
 690     }
 691
 692     @VisibleForTesting
 693     SnapshotTracker getSnapshotTracker() {
 694         return snapshotTracker;
 695     }
 696 }