-
Notifications
You must be signed in to change notification settings - Fork 2.3k
/
SegmentedRaftLog.java
284 lines (244 loc) · 9.48 KB
/
SegmentedRaftLog.java
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
/*
* Copyright (c) 2002-2016 "Neo Technology,"
* Network Engine for Objects in Lund AB [http://neotechnology.com]
*
* This file is part of Neo4j.
*
* Neo4j is free software: you can redistribute it and/or modify
* it under the terms of the GNU Affero General Public License as
* published by the Free Software Foundation, either version 3 of the
* License, or (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU Affero General Public License for more details.
*
* You should have received a copy of the GNU Affero General Public License
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
package org.neo4j.coreedge.raft.log.segmented;
import java.io.File;
import java.io.IOException;
import java.time.Clock;
import org.neo4j.coreedge.helper.StatUtil;
import org.neo4j.coreedge.raft.log.EntryRecord;
import org.neo4j.coreedge.raft.log.RaftLog;
import org.neo4j.coreedge.raft.log.RaftLogCursor;
import org.neo4j.coreedge.raft.log.RaftLogEntry;
import org.neo4j.coreedge.raft.replication.ReplicatedContent;
import org.neo4j.coreedge.raft.state.ChannelMarshal;
import org.neo4j.cursor.IOCursor;
import org.neo4j.io.fs.FileSystemAbstraction;
import org.neo4j.kernel.lifecycle.LifecycleAdapter;
import org.neo4j.logging.Log;
import org.neo4j.logging.LogProvider;
import static java.lang.String.format;
/**
* The segmented RAFT log is an append only log supporting the operations required to support
* the RAFT consensus algorithm.
* <p>
* A RAFT log must be able to append new entries, but also truncate not yet committed entries,
* prune out old compacted entries and skip to a later starting point.
* <p>
* The RAFT log consists of a sequence of individual log files, called segments, with
* the following format:
* <p>
* [HEADER] [ENTRY]*
* <p>
* So a header with zero or more entries following it. Each segment file contains a consecutive
* sequence of appended entries. The operations of truncating and skipping in the log is implemented
* by switching to the next segment file, called the next version. A new segment file is also started
* when the threshold for a particular file has been reached.
*/
public class SegmentedRaftLog extends LifecycleAdapter implements RaftLog
{
public static final String SEGMENTED_LOG_DIRECTORY_NAME = "raft-log";
private final FileSystemAbstraction fileSystem;
private final File directory;
private final long rotateAtSize;
private final ChannelMarshal<ReplicatedContent> contentMarshal;
private final FileNames fileNames;
private final Log log;
private boolean needsRecovery;
private final LogProvider logProvider;
private final SegmentedRaftLogPruner pruner;
private State state;
private final ReaderPool readerPool;
private final StatUtil.StatContext scanStats; // this is temporary, for debugging purposes
public SegmentedRaftLog( FileSystemAbstraction fileSystem, File directory, long rotateAtSize,
ChannelMarshal<ReplicatedContent> contentMarshal, LogProvider logProvider,
String pruningConfig, int readerPoolSize, Clock clock )
{
this.fileSystem = fileSystem;
this.directory = directory;
this.rotateAtSize = rotateAtSize;
this.contentMarshal = contentMarshal;
this.logProvider = logProvider;
this.scanStats = StatUtil.create( "linear scans", logProvider.getLog( getClass() ), 10_000, true, false );
this.fileNames = new FileNames( directory );
this.readerPool = new ReaderPool( readerPoolSize, logProvider, fileNames, fileSystem, clock );
this.pruner = new SegmentedRaftLogPruner( pruningConfig, logProvider );
this.log = logProvider.getLog( getClass() );
}
@Override
public synchronized void start() throws IOException, DamagedLogStorageException, DisposedException
{
if ( !directory.exists() && !directory.mkdirs() )
{
throw new IOException( "Could not create: " + directory );
}
RecoveryProtocol recoveryProtocol = new RecoveryProtocol( fileSystem, fileNames, readerPool, contentMarshal, logProvider, scanStats );
state = recoveryProtocol.run();
log.info( "log started with recovered state %s", state );
}
@Override
public synchronized void stop() throws DisposedException, IOException
{
readerPool.disposeAll();
state.segments.close();
}
@Override
public synchronized long append( RaftLogEntry... entries ) throws IOException
{
ensureOk();
try
{
for ( RaftLogEntry entry : entries )
{
state.appendIndex++;
updateTerm( entry );
state.segments.last().write( state.appendIndex, entry );
}
state.segments.last().flush();
}
catch ( Throwable e )
{
needsRecovery = true;
throw e;
}
if ( state.segments.last().position() >= rotateAtSize )
{
rotateSegment( state.appendIndex, state.appendIndex, state.currentTerm );
}
return state.appendIndex;
}
private void ensureOk()
{
if ( needsRecovery )
{
throw new IllegalStateException( "Raft log requires recovery" );
}
}
private void updateTerm( RaftLogEntry entry )
{
if ( entry.term() >= state.currentTerm )
{
state.currentTerm = entry.term();
}
else
{
throw new IllegalStateException(
format( "Non-monotonic term %d for entry %s in term %d", entry.term(), entry.toString(),
state.currentTerm ) );
}
}
@Override
public synchronized void truncate( long fromIndex ) throws IOException
{
if ( state.appendIndex < fromIndex )
{
throw new IllegalArgumentException( "Cannot truncate at index " + fromIndex + " when append index is " +
state.appendIndex );
}
long newAppendIndex = fromIndex - 1;
long newTerm = readEntryTerm( newAppendIndex );
truncateSegment( state.appendIndex, newAppendIndex, newTerm );
state.appendIndex = newAppendIndex;
state.currentTerm = newTerm;
}
private void rotateSegment( long prevFileLastIndex, long prevIndex, long prevTerm ) throws IOException
{
state.segments.last().closeWriter();
state.segments.rotate( prevFileLastIndex, prevIndex, prevTerm );
}
private void truncateSegment( long prevFileLastIndex, long prevIndex, long prevTerm ) throws IOException
{
state.segments.last().closeWriter();
state.segments.truncate( prevFileLastIndex, prevIndex, prevTerm );
}
private void skipSegment( long prevFileLastIndex, long prevIndex, long prevTerm ) throws IOException
{
state.segments.last().closeWriter();
state.segments.skip( prevFileLastIndex, prevIndex, prevTerm );
}
@Override
public long appendIndex()
{
return state.appendIndex;
}
@Override
public long prevIndex()
{
return state.prevIndex;
}
@Override
public RaftLogCursor getEntryCursor( long fromIndex ) throws IOException
{
final IOCursor<EntryRecord> inner = new EntryCursor( state.segments, fromIndex );
return new SegmentedRaftLogCursor( fromIndex, inner );
}
@Override
public synchronized long skip( long newIndex, long newTerm ) throws IOException
{
log.info( "Skipping from {index: %d, term: %d} to {index: %d, term: %d}", state.appendIndex, state.currentTerm, newIndex, newTerm );
if ( state.appendIndex < newIndex )
{
skipSegment( state.appendIndex, newIndex, newTerm );
state.prevTerm = newTerm;
state.currentTerm = newTerm;
state.prevIndex = newIndex;
state.appendIndex = newIndex;
}
return state.appendIndex;
}
private RaftLogEntry readLogEntry( long logIndex ) throws IOException
{
try ( IOCursor<EntryRecord> cursor = new EntryCursor( state.segments, logIndex ) )
{
return cursor.next() ? cursor.get().logEntry() : null;
}
}
@Override
public long readEntryTerm( long logIndex ) throws IOException
{
log.info( "reading entry term at %d. prevIndex:%d, prevTerm:%d", logIndex, state.prevIndex, state.prevTerm );
if ( logIndex == state.prevIndex )
{
return state.prevTerm;
}
else if ( logIndex < state.prevIndex || logIndex > state.appendIndex )
{
return -1;
}
RaftLogEntry entry = readLogEntry( logIndex );
return entry == null ? -1 : entry.term();
}
@Override
public long prune( long safeIndex ) throws IOException
{
long pruneIndex = pruner.getIndexToPruneFrom( safeIndex, state.segments );
SegmentFile oldestNotDisposed = state.segments.prune( pruneIndex );
long newPrevIndex = oldestNotDisposed.header().prevIndex();
long newPrevTerm = oldestNotDisposed.header().prevTerm();
if ( newPrevIndex > state.prevIndex )
{
state.prevIndex = newPrevIndex;
}
if ( newPrevTerm > state.prevTerm )
{
state.prevTerm = newPrevTerm;
}
return state.prevIndex;
}
}