/*
* Copyright 2014 Higher Frequency Trading http://www.higherfrequencytrading.com
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package net.openhft.chronicle.map;
import net.openhft.chronicle.TimeProvider;
import net.openhft.chronicle.map.serialization.BytesReader;
import net.openhft.chronicle.map.serialization.MetaBytesInterop;
import net.openhft.chronicle.map.serialization.MetaBytesWriter;
import net.openhft.chronicle.map.threadlocal.ThreadLocalCopies;
import net.openhft.lang.Maths;
import net.openhft.lang.collection.ATSDirectBitSet;
import net.openhft.lang.io.Bytes;
import net.openhft.lang.io.MultiStoreBytes;
import net.openhft.lang.io.NativeBytes;
import net.openhft.lang.model.Byteable;
import org.jetbrains.annotations.NotNull;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import java.io.*;
import java.util.Iterator;
import java.util.Set;
import java.util.concurrent.CopyOnWriteArraySet;
import java.util.concurrent.atomic.AtomicReferenceArray;
import static net.openhft.chronicle.map.serialization.Hasher.hash;
import static net.openhft.lang.collection.DirectBitSet.NOT_FOUND;
/**
* A Replicating Multi Master HashMap Each remote hash map, mirrors its changes over to another remote
* hash map, neither hash map is considered the master store of data, each hash map uses timestamps to
* reconcile changes. We refer to an instance of a remote hash-map as a node. A node will be connected to any
* number of other nodes, for the first implementation the maximum number of nodes will be fixed. The data
* that is stored locally in each node will become eventually consistent. So changes made to one node, for
* example by calling put() will be replicated over to the other node. To achieve a high level of performance
* and throughput, the call to put() won’t block, with concurrentHashMap, It is typical to check the return
* code of some methods to obtain the old value for example remove(). Due to the loose coupling and lock free
* nature of this multi master implementation, this return value will only be the old value on the nodes
* local data store. In other words the nodes are only concurrent locally. Its worth realising that another
* node performing exactly the same operation may return a different value. However reconciliation will ensure
* the maps themselves become eventually consistent.
Reconciliation
If two ( or more nodes )
* were to receive a change to their maps for the same key but different values, say by a user of the maps,
* calling the put(key, value). Then, initially each node will update its local store and each local store
* will hold a different value, but the aim of multi master replication is to provide eventual consistency
* across the nodes. So, with multi master when ever a node is changed it will notify the other nodes of its
* change. We will refer to this notification as an event. The event will hold a timestamp indicating the time
* the change occurred, it will also hold the state transition, in this case it was a put with a key and
* value. Eventual consistency is achieved by looking at the timestamp from the remote node, if for a given
* key, the remote nodes timestamp is newer than the local nodes timestamp, then the event from the remote
* node will be applied to the local node, otherwise the event will be ignored.
However there is an
* edge case that we have to concern ourselves with, If two nodes update their map at the same time with
* different values, we have to deterministically resolve which update wins, because of eventual consistency
* both nodes should end up locally holding the same data. Although it is rare two remote nodes could receive
* an update to their maps at exactly the same time for the same key, we have to handle this edge case, its
* therefore important not to rely on timestamps alone to reconcile the updates. Typically the update with the
* newest timestamp should win, but in this example both timestamps are the same, and the decision made to one
* node should be identical to the decision made to the other. We resolve this simple dilemma by using a node
* identifier, each node will have a unique identifier, the update from the node with the smallest identifier
* wins.
*
* @param the entries key type
* @param the entries value type
*/
class ReplicatedChronicleMap,
V, VW, MVW extends MetaBytesWriter>
extends VanillaChronicleMap
implements ChronicleMap,
Replica, Replica.EntryExternalizable, Replica.EntryResolver, Closeable {
// for file, jdbc and UDP replication
public static final int RESERVED_MOD_ITER = 8;
static final int MAX_UNSIGNED_SHORT = Character.MAX_VALUE;
private static final long serialVersionUID = 0L;
private static final Logger LOG = LoggerFactory.getLogger(ReplicatedChronicleMap.class);
private static final long LAST_UPDATED_HEADER_SIZE = 128L * 8L;
public static final int ADDITIONAL_ENTRY_BYTES = 10;
private final TimeProvider timeProvider;
private final byte localIdentifier;
transient Set closeables;
private transient Bytes identifierUpdatedBytes;
private transient ModificationDelegator modificationDelegator;
private transient long startOfModificationIterators;
public ReplicatedChronicleMap(@NotNull ChronicleMapBuilder builder)
throws IOException {
super(builder);
this.timeProvider = builder.timeProvider();
this.localIdentifier = builder.identifier();
if (localIdentifier==-1){
throw new IllegalStateException("localIdentifier should not be -1");
}
}
private int assignedModIterBitSetSizeInBytes() {
return (int) align64((127 + RESERVED_MOD_ITER) / 8);
}
@Override
Segment createSegment(NativeBytes bytes, int index) {
return new Segment(bytes, index);
}
Class segmentType() {
return Segment.class;
}
@Override
void initTransients() {
super.initTransients();
closeables = new CopyOnWriteArraySet();
}
long modIterBitSetSizeInBytes() {
return align64(bitsPerSegmentInModIterBitSet() * segments.length / 8);
}
private long bitsPerSegmentInModIterBitSet() {
// min 128 * 8 to prevent false sharing on updating bits from different segments
return Maths.nextPower2((long) entriesPerSegment, 128 * 8);
}
@Override
int multiMapsPerSegment() {
return 2;
}
@Override
long getHeaderSize() {
return super.getHeaderSize() + LAST_UPDATED_HEADER_SIZE +
(modIterBitSetSizeInBytes() * (128 + RESERVED_MOD_ITER)) +
assignedModIterBitSetSizeInBytes();
}
void setLastModificationTime(byte identifier, long timestamp) {
final long offset = identifier * 8L;
// purposely not volatile as this will impact performance,
// and the worst that will happen is we'll end up loading more data on a bootstrap
if (identifierUpdatedBytes.readLong(offset) < timestamp)
identifierUpdatedBytes.writeLong(offset, timestamp);
}
@Override
public long lastModificationTime(byte remoteIdentifier) {
assert remoteIdentifier != this.identifier();
// purposely not volatile as this will impact performance,
// and the worst that will happen is we'll end up loading more data on a bootstrap
return identifierUpdatedBytes.readLong(remoteIdentifier * 8L);
}
@Override
void onHeaderCreated() {
long offset = super.getHeaderSize();
identifierUpdatedBytes = ms.bytes(offset, LAST_UPDATED_HEADER_SIZE).zeroOut();
offset += LAST_UPDATED_HEADER_SIZE;
Bytes modDelBytes = ms.bytes(offset, assignedModIterBitSetSizeInBytes()).zeroOut();
offset += assignedModIterBitSetSizeInBytes();
startOfModificationIterators = offset;
modificationDelegator = new ModificationDelegator(eventListener, modDelBytes);
}
@Override
MapEventListener> eventListener() {
return modificationDelegator;
}
@Override
public V put(K key, V value) {
return put0(key, value, true, localIdentifier, timeProvider.currentTimeMillis());
}
/**
* Used in conjunction with map replication, all put events that originate from a remote node will be
* processed using this method.
*
* @param key key with which the specified value is to be associated
* @param value value to be associated with the specified key
* @param identifier a unique identifier for a replicating node
* @param timeStamp timestamp in milliseconds, when the put event originally occurred
* @return the previous value
* @see #put(Object, Object)
*/
V put(K key, V value, byte identifier, long timeStamp) {
assert identifier > 0;
return put0(key, value, true, identifier, timeStamp);
}
@Override
public V putIfAbsent(@net.openhft.lang.model.constraints.NotNull K key, V value) {
return put0(key, value, false, localIdentifier, timeProvider.currentTimeMillis());
}
/**
* @param key key with which the specified value is associated
* @param value value expected to be associated with the specified key
* @param replaceIfPresent set to false for putIfAbsent()
* @param identifier used to identify which replicating node made the change
* @param timeStamp the time that that change was made, this is used for replication
* @return the value that was replaced
*/
private V put0(K key, V value, boolean replaceIfPresent,
final byte identifier, long timeStamp) {
checkKey(key);
checkValue(value);
ThreadLocalCopies copies = keyInteropProvider.getCopies(null);
KI keyWriter = keyInteropProvider.get(copies, originalKeyInterop);
copies = metaKeyInteropProvider.getCopies(copies);
MKI metaKeyWriter =
metaKeyInteropProvider.get(copies, originalMetaKeyInterop, keyWriter, key);
long hash = metaKeyWriter.hash(keyWriter, key);
int segmentNum = getSegment(hash);
int segmentHash = segmentHash(hash);
return segment(segmentNum).put(copies, metaKeyWriter, keyWriter, key, value, segmentHash,
replaceIfPresent, identifier, timeStamp);
}
/**
* @param segmentNum a unique index of the segment
* @return the segment associated with the {@code segmentNum}
*/
private Segment segment(int segmentNum) {
return (Segment) segments[segmentNum];
}
@Override
V lookupUsing(K key, V value, boolean create) {
checkKey(key);
ThreadLocalCopies copies = keyInteropProvider.getCopies(null);
KI keyWriter = keyInteropProvider.get(copies, originalKeyInterop);
copies = metaKeyInteropProvider.getCopies(copies);
MKI metaKeyWriter =
metaKeyInteropProvider.get(copies, originalMetaKeyInterop, keyWriter, key);
long hash = metaKeyWriter.hash(keyWriter, key);
int segmentNum = getSegment(hash);
int segmentHash = segmentHash(hash);
return segment(segmentNum).acquire(copies, metaKeyWriter, keyWriter, key, value,
segmentHash, create, timeProvider.currentTimeMillis());
}
@Override
public V remove(final Object key) {
return removeIfValueIs(key, null, localIdentifier, timeProvider.currentTimeMillis());
}
/**
* Used in conjunction with map replication, all remove events that originate from a remote node will be
* processed using this method.
*
* @param key key with which the specified value is associated
* @param value value expected to be associated with the specified key
* @param identifier a unique identifier for a replicating node
* @param timeStamp timestamp in milliseconds, when the remove event originally occurred
* @return {@code true} if the entry was removed
* @see #remove(Object, Object)
*/
V remove(K key, V value, byte identifier, long timeStamp) {
assert identifier > 0;
return removeIfValueIs(key, value, identifier, timeStamp);
}
void addCloseable(Closeable closeable) {
closeables.add(closeable);
}
@Override
public void close() {
for (Closeable closeable : closeables) {
try {
closeable.close();
} catch (IOException e) {
LOG.error("", e);
}
}
try {
// give time for stuff to fully close
Thread.sleep(50);
} catch (InterruptedException e) {
e.printStackTrace();
}
super.close();
}
@Override
public byte identifier() {
return localIdentifier;
}
@Override
public Replica.ModificationIterator acquireModificationIterator
(short remoteIdentifier,
@NotNull final ModificationNotifier modificationNotifier) {
return modificationDelegator
.acquireModificationIterator(remoteIdentifier, modificationNotifier);
}
@Override
public boolean remove(@net.openhft.lang.model.constraints.NotNull final Object key, final Object value) {
if (value == null)
return false; // CHM compatibility; I would throw NPE
return removeIfValueIs(key, (V) value,
localIdentifier, timeProvider.currentTimeMillis()) != null;
}
private V removeIfValueIs(final Object k, final V expectedValue,
final byte identifier, final long timestamp) {
checkKey(k);
K key = (K) k;
ThreadLocalCopies copies = keyInteropProvider.getCopies(null);
KI keyWriter = keyInteropProvider.get(copies, originalKeyInterop);
copies = metaKeyInteropProvider.getCopies(copies);
MKI metaKeyWriter =
metaKeyInteropProvider.get(copies, originalMetaKeyInterop, keyWriter, key);
long hash = metaKeyWriter.hash(keyWriter, key);
int segmentNum = getSegment(hash);
int segmentHash = segmentHash(hash);
return segment(segmentNum).remove(copies, metaKeyWriter, keyWriter, key, expectedValue,
segmentHash, timestamp, identifier);
}
@Override
V replaceIfValueIs(@net.openhft.lang.model.constraints.NotNull final K key, final V existingValue, final V newValue) {
checkKey(key);
checkValue(newValue);
ThreadLocalCopies copies = keyInteropProvider.getCopies(null);
KI keyWriter = keyInteropProvider.get(copies, originalKeyInterop);
copies = metaKeyInteropProvider.getCopies(copies);
MKI metaKeyWriter =
metaKeyInteropProvider.get(copies, originalMetaKeyInterop, keyWriter, key);
long hash = metaKeyWriter.hash(keyWriter, key);
int segmentNum = getSegment(hash);
int segmentHash = segmentHash(hash);
return segment(segmentNum).replace(copies, metaKeyWriter, keyWriter, key, existingValue,
newValue, segmentHash, timeProvider.currentTimeMillis());
}
/**
* This method does not set a segment lock, A segment lock should be obtained before calling this method,
* especially when being used in a multi threaded context.
*/
@Override
public void writeExternalEntry(@NotNull Bytes entry, @NotNull Bytes destination,
int chronicleId) {
final long initialLimit = entry.limit();
final long keySize = keySizeMarshaller.readSize(entry);
final long keyPosition = entry.position();
entry.skip(keySize);
final long keyLimit = entry.position();
final long timeStamp = entry.readLong();
final byte identifier = entry.readByte();
if (identifier != localIdentifier) {
// although unlikely, this may occur if the entry has been updated
return;
}
final boolean isDeleted = entry.readBoolean();
long valueSize;
if (!isDeleted) {
valueSize = valueSizeMarshaller.readSize(entry);
} else {
valueSize = 0;
}
final long valuePosition = entry.position();
keySizeMarshaller.writeSize(destination, keySize);
valueSizeMarshaller.writeSize(destination, valueSize);
destination.writeStopBit(timeStamp);
// we store the isDeleted flag in the identifier
// ( when the identifier is negative it is deleted )
destination.writeByte(isDeleted ? -identifier : identifier);
// write the key
entry.position(keyPosition);
entry.limit(keyLimit);
destination.write(entry);
boolean debugEnabled = LOG.isDebugEnabled();
String message = null;
if (debugEnabled) {
if (isDeleted) {
LOG.debug("WRITING ENTRY TO DEST - into local-id={}, remove(key={})",
localIdentifier, entry.toString().trim());
} else {
message = String.format(
"WRITING ENTRY TO DEST - into local-id=%d, put(key=%s,",
localIdentifier, entry.toString().trim());
}
}
if (isDeleted)
return;
entry.limit(initialLimit);
entry.position(valuePosition);
// skipping the alignment, as alignment wont work when we send the data over the wire.
alignment.alignPositionAddr(entry);
// writes the value
entry.limit(entry.position() + valueSize);
destination.write(entry);
if (debugEnabled) {
LOG.debug(message + "value=" + entry.toString().trim() + ")");
}
}
/**
* This method does not set a segment lock, A segment lock should be obtained before calling this method,
* especially when being used in a multi threaded context.
*/
@Override
public void readExternalEntry(@NotNull Bytes source) {
final long keySize = keySizeMarshaller.readSize(source);
final long valueSize = valueSizeMarshaller.readSize(source);
final long timeStamp = source.readStopBit();
final byte id = source.readByte();
final byte remoteIdentifier;
final boolean isDeleted;
if (id < 0) {
isDeleted = true;
remoteIdentifier = (byte) -id;
} else if (id != 0) {
isDeleted = false;
remoteIdentifier = id;
} else {
throw new IllegalStateException("identifier can't be 0");
}
if (remoteIdentifier == ReplicatedChronicleMap.this.identifier()) {
// this may occur when working with UDP, as we will receive our own data
return;
}
final long keyPosition = source.position();
final long keyLimit = keyPosition + keySize;
source.limit(keyLimit);
long hash = hash(source);
int segmentNum = getSegment(hash);
int segmentHash = segmentHash(hash);
boolean debugEnabled = LOG.isDebugEnabled();
if (isDeleted) {
if (debugEnabled) {
LOG.debug(
"READING FROM SOURCE - into local-id={}, remote={}, remove(key={})",
localIdentifier, remoteIdentifier, source.toString().trim()
);
}
segment(segmentNum).remoteRemove(source, segmentHash, timeStamp, remoteIdentifier);
setLastModificationTime(remoteIdentifier, timeStamp);
return;
}
String message = null;
if (debugEnabled) {
message = String.format(
"READING FROM SOURCE - into local-id=%d, remote-id=%d, put(key=%s,",
localIdentifier, remoteIdentifier, source.toString().trim());
}
final long valuePosition = keyLimit;
final long valueLimit = valuePosition + valueSize;
segment(segmentNum).remotePut(source, segmentHash, remoteIdentifier, timeStamp,
valuePosition, valueLimit);
setLastModificationTime(remoteIdentifier, timeStamp);
if (debugEnabled) {
source.limit(valueLimit);
source.position(valuePosition);
LOG.debug(message + "value=" + source.toString().trim() + ")");
}
}
@NotNull
@Override
public Set> entrySet() {
return new EntrySet();
}
@Override
public K key(@NotNull Bytes entry, K usingKey) {
final long start = entry.position();
try {
long keySize = keySizeMarshaller.readSize(entry);
ThreadLocalCopies copies = keyReaderProvider.getCopies(null);
return keyReaderProvider.get(copies, originalKeyReader).read(entry, keySize);
} finally {
entry.position(start);
}
}
@Override
public V value(@NotNull Bytes entry, V usingValue) {
final long start = entry.position();
try {
entry.skip(keySizeMarshaller.readSize(entry));
//timeStamp
entry.readLong();
final byte identifier = entry.readByte();
if (identifier != localIdentifier) {
return null;
}
final boolean isDeleted = entry.readBoolean();
long valueSize;
if (!isDeleted) {
valueSize = valueSizeMarshaller.readSize(entry);
assert valueSize > 0;
} else {
return null;
}
alignment.alignPositionAddr(entry);
ThreadLocalCopies copies = valueReaderProvider.getCopies(null);
BytesReader valueReader = valueReaderProvider.get(copies, originalValueReader);
return valueReader.read(entry, valueSize, usingValue);
} finally {
entry.position(start);
}
}
@Override
public boolean wasRemoved(@NotNull Bytes entry) {
final long start = entry.position();
try {
return entry.readBoolean(keySizeMarshaller.readSize(entry) + 9);
} finally {
entry.position(start);
}
}
class Segment extends VanillaChronicleMap.Segment {
private volatile IntIntMultiMap hashLookupLiveAndDeleted;
private volatile IntIntMultiMap hashLookupLiveOnly;
Segment(NativeBytes bytes, int index) {
super(bytes, index);
}
@Override
void createHashLookups(long start) {
hashLookupLiveAndDeleted = createMultiMap(start);
start += align64(sizeOfMultiMap() + sizeOfMultiMapBitSet());
hashLookupLiveOnly = createMultiMap(start);
}
public IntIntMultiMap getHashLookup() {
return hashLookupLiveOnly;
}
private long entrySize(long keySize, long valueSize) {
long result = alignment.alignAddr(metaDataBytes +
keySizeMarshaller.sizeEncodingSize(keySize) + keySize + ADDITIONAL_ENTRY_BYTES +
valueSizeMarshaller.sizeEncodingSize(valueSize)) + valueSize;
// replication enforces that the entry size will never be larger than an unsigned short
if (result > MAX_UNSIGNED_SHORT)
throw new IllegalStateException("ENTRY WRITE_BUFFER_SIZE TOO LARGE : Replicated " +
"ChronicleMap's" +
" are restricted to an " +
"entry size of " + MAX_UNSIGNED_SHORT + ", " +
"your entry size=" + result);
return result;
}
/**
* @see VanillaChronicleMap.Segment#acquire
*/
V acquire(ThreadLocalCopies copies, MKI metaKeyWriter, KI keyWriter,
K key, V usingValue, int hash2, boolean create, long timestamp) {
lock();
try {
long keySize = metaKeyWriter.size(keyWriter, key);
MultiStoreBytes entry = tmpBytes;
long offset = searchKey(keyWriter, metaKeyWriter, key, keySize, hash2, entry,
hashLookupLiveOnly);
if (offset >= 0) {
// skip the timestamp, identifier and is deleted flag
entry.skip(ADDITIONAL_ENTRY_BYTES);
return onKeyPresentOnAcquire(copies, key, usingValue, offset, entry);
} else {
boolean usingValuePassed = usingValue != null;
usingValue = tryObtainUsingValueOnAcquire(key, usingValue, create);
if (usingValue != null) {
// see VanillaChronicleMap.Segment.acquire() for explanation
// why `usingValue` is `create`.
offset = putEntryOnAcquire(copies, metaKeyWriter, keyWriter, key, keySize,
hash2, usingValue, create, timestamp);
incrementSize();
if (usingValuePassed || !create)
notifyPut(offset, true, key, usingValue, posFromOffset(offset));
return usingValue;
} else {
return null;
}
}
} finally {
unlock();
}
}
private long putEntryOnAcquire(ThreadLocalCopies copies,
MKI metaKeyWriter, KI keyWriter, K key,
long keySize, int hash2, V value, boolean usingValue,
long timestamp) {
return putEntry(copies, metaKeyWriter, keyWriter, key, keySize, hash2, value,
usingValue, localIdentifier, timestamp, hashLookupLiveOnly);
}
/**
* called from a remote node as part of replication
*/
private void remoteRemove(Bytes keyBytes, int hash2,
final long timestamp, final byte identifier) {
lock();
try {
long keySize = keyBytes.remaining();
hashLookupLiveAndDeleted.startSearch(hash2);
for (int pos; (pos = hashLookupLiveAndDeleted.nextPos()) >= 0; ) {
long offset = offsetFromPos(pos);
NativeBytes entry = entry(offset);
if (!remoteKeyEquals(keyBytes, keySize, entry))
continue;
// key is found
entry.skip(keySize);
long timeStampPos = entry.position();
if (shouldIgnore(entry, timestamp, identifier)) {
return;
}
// skip the is deleted flag
boolean wasDeleted = entry.readBoolean();
if (!wasDeleted) {
hashLookupLiveOnly.remove(hash2, pos);
decrementSize();
}
entry.position(timeStampPos);
entry.writeLong(timestamp);
assert identifier > 0;
entry.writeByte(identifier);
// was deleted
entry.writeBoolean(true);
}
// key is not found
if (LOG.isDebugEnabled())
LOG.debug("Segment.remoteRemove() : key=" + keyBytes.toString().trim() + " was not " +
"found");
} finally {
unlock();
}
}
/**
* called from a remote node when it wishes to propagate a remove event
*/
private void remotePut(@NotNull final Bytes inBytes, int hash2,
final byte identifier, final long timestamp,
long valuePos, long valueLimit) {
lock();
try {
// inBytes position and limit correspond to the key
final long keySize = inBytes.remaining();
hashLookupLiveAndDeleted.startSearch(hash2);
for (int pos; (pos = hashLookupLiveAndDeleted.nextPos()) >= 0; ) {
long offset = offsetFromPos(pos);
NativeBytes entry = entry(offset);
if (!remoteKeyEquals(inBytes, keySize, entry))
continue;
// key is found
entry.skip(keySize);
final long timeStampPos = entry.positionAddr();
if (shouldIgnore(entry, timestamp, identifier)) {
return;
}
boolean wasDeleted = entry.readBoolean();
entry.positionAddr(timeStampPos);
entry.writeLong(timestamp);
assert identifier > 0;
entry.writeByte(identifier);
// deleted flag
entry.writeBoolean(false);
long valueSizePos = entry.position();
long valueSize = readValueSize(entry);
long entryEndAddr = entry.positionAddr() + valueSize;
// write the value
inBytes.limit(valueLimit);
inBytes.position(valuePos);
putValue(pos, offset, entry, valueSizePos, entryEndAddr, null, null, inBytes,
hashLookupLiveAndDeleted);
if (wasDeleted) {
// remove() would have got rid of this so we have to add it back in
hashLookupLiveOnly.put(hash2, pos);
incrementSize();
}
return;
}
// key is not found
long valueSize = valueLimit - valuePos;
int pos = alloc(inBlocks(entrySize(keySize, valueSize)));
long offset = offsetFromPos(pos);
clearMetaData(offset);
NativeBytes entry = entry(offset);
keySizeMarshaller.writeSize(entry, keySize);
// write the key
entry.write(inBytes);
entry.writeLong(timestamp);
entry.writeByte(identifier);
entry.writeBoolean(false);
valueSizeMarshaller.writeSize(entry, valueSize);
alignment.alignPositionAddr(entry);
// write the value
inBytes.limit(valueLimit);
inBytes.position(valuePos);
entry.write(inBytes);
hashLookupLiveAndDeleted.putAfterFailedSearch(pos);
hashLookupLiveOnly.put(hash2, pos);
incrementSize();
} finally {
unlock();
}
}
private boolean remoteKeyEquals(Bytes keyBytes, long keySize, Bytes entry) {
return keySize == keySizeMarshaller.readSize(entry) && entry.startsWith(keyBytes);
}
V put(ThreadLocalCopies copies, MKI metaKeyWriter, KI keyWriter,
K key, V value, int hash2, boolean replaceIfPresent,
final byte identifier, final long timestamp) {
lock();
try {
IntIntMultiMap hashLookup = hashLookupLiveAndDeleted;
long keySize = metaKeyWriter.size(keyWriter, key);
hashLookup.startSearch(hash2);
int pos;
while ((pos = hashLookup.nextPos()) >= 0) {
long offset = offsetFromPos(pos);
NativeBytes entry = entry(offset);
if (!keyEquals(keyWriter, metaKeyWriter, key, keySize, entry))
continue;
// key is found
entry.skip(keySize);
final long timeStampPos = entry.positionAddr();
if (shouldIgnore(entry, timestamp, identifier))
return null;
boolean wasDeleted = entry.readBoolean();
// if wasDeleted==true then we even if replaceIfPresent==false we can treat it
// the same way as replaceIfPresent true
if (replaceIfPresent || wasDeleted) {
entry.positionAddr(timeStampPos);
entry.writeLong(timestamp);
entry.writeByte(identifier);
// deleted flag
entry.writeBoolean(false);
final V prevValue = replaceValueOnPut(copies, key, value, entry, pos,
offset, !wasDeleted && !putReturnsNull, hashLookup);
if (wasDeleted) {
// remove() would have got rid of this so we have to add it back in
hashLookupLiveOnly.put(hash2, pos);
incrementSize();
return null;
} else {
return prevValue;
}
} else {
// we wont replaceIfPresent and the entry is not deleted
return putReturnsNull ? null : readValue(copies, entry, null);
}
}
// key is not found
long offset = putEntry(copies, metaKeyWriter, keyWriter, key, keySize, hash2, value,
false, identifier, timestamp, hashLookup);
incrementSize();
notifyPut(offset, true, key, value, posFromOffset(offset));
return null;
} finally {
unlock();
}
}
/**
* Used only with replication, its sometimes possible to receive an old ( or stale update ) from a
* remote map. This method is used to determine if we should ignore such updates. We can
* reject put() and removes() when comparing times stamps with remote systems
*
* @param entry the maps entry
* @param timestamp the time the entry was created or updated
* @param identifier the unique identifier relating to this map
* @return true if the entry should not be processed
*/
private boolean shouldIgnore(@NotNull final NativeBytes entry, final long timestamp, final byte identifier) {
final long lastModifiedTimeStamp = entry.readLong();
// if the readTimeStamp is newer then we'll reject this put()
// or they are the same and have a larger id
if (lastModifiedTimeStamp < timestamp) {
entry.skip(1); // skip the byte used for the identifier
return false;
}
if (lastModifiedTimeStamp > timestamp)
return true;
// check the identifier
return entry.readByte() > identifier;
}
/**
* Puts entry. If {@code value} implements {@link Byteable} interface and {@code usingValue} is {@code
* true}, the value is backed with the bytes of this entry.
*
* @param hash2 a hash of the {@code keyBytes}. Caller was searching for the key in the
* {@code searchedHashLookup} using this hash.
* @param value the value to put
* @param usingValue {@code true} if the value should be backed with the bytes of the entry,
* if it implements {@link Byteable} interface, {@code false} if it should
* put itself
* @param identifier the identifier of the outer CHM node
* @param timestamp the timestamp when the entry was put (this could be later if it was a
* remote put) this method is called only from usual put or acquire
* @param searchedHashLookup the hash lookup that used to find the entry based on the key
* @return offset of the written entry in the Segment bytes
* @see VanillaChronicleMap.Segment#putEntry
*/
private long putEntry(ThreadLocalCopies copies, MKI metaKeyWriter,
KI keyWriter, K key, long keySize, int hash2, V value,
boolean usingValue, final int identifier, final long timestamp,
IntIntMultiMap searchedHashLookup) {
// "if-else polymorphism" is not very beautiful, but allows to
// reuse the rest code of this method and doesn't hurt performance.
boolean byteableValue = usingValue && value instanceof Byteable;
long valueSize;
MetaBytesWriter metaValueWriter = null;
VW valueWriter = null;
Byteable valueAsByteable = null;
if (!byteableValue) {
copies = valueWriterProvider.getCopies(copies);
valueWriter = valueWriterProvider.get(copies, originalValueWriter);
copies = metaValueWriterProvider.getCopies(copies);
metaValueWriter = metaValueWriterProvider.get(
copies, originalMetaValueWriter, valueWriter, value);
valueSize = metaValueWriter.size(valueWriter, value);
} else {
valueAsByteable = (Byteable) value;
valueSize = valueAsByteable.maxSize();
}
long entrySize = entrySize(keySize, valueSize);
int pos = alloc(inBlocks(entrySize));
long offset = offsetFromPos(pos);
clearMetaData(offset);
NativeBytes entry = entry(offset);
keySizeMarshaller.writeSize(entry, keySize);
metaKeyWriter.write(keyWriter, entry, key);
entry.writeLong(timestamp);
entry.writeByte(identifier);
entry.writeBoolean(false);
writeValueOnPutEntry(valueSize, metaValueWriter, valueWriter, value, valueAsByteable,
entry);
// we have to add it both to the live
if (searchedHashLookup == hashLookupLiveAndDeleted) {
hashLookupLiveAndDeleted.putAfterFailedSearch(pos);
hashLookupLiveOnly.put(hash2, pos);
} else {
hashLookupLiveOnly.putAfterFailedSearch(pos);
hashLookupLiveAndDeleted.put(hash2, pos);
}
return offset;
}
/**
* @see VanillaChronicleMap.Segment#remove
*/
V remove(ThreadLocalCopies copies, MKI metaKeyWriter, KI keyWriter,
K key, V expectedValue, int hash2,
final long timestamp, final byte identifier) {
assert identifier > 0;
lock();
try {
long keySize = metaKeyWriter.size(keyWriter, key);
final IntIntMultiMap multiMap = hashLookupLiveAndDeleted;
multiMap.startSearch(hash2);
for (int pos; (pos = multiMap.nextPos()) >= 0; ) {
long offset = offsetFromPos(pos);
NativeBytes entry = entry(offset);
if (!keyEquals(keyWriter, metaKeyWriter, key, keySize, entry))
continue;
// key is found
entry.skip(keySize);
long timeStampPos = entry.position();
if (shouldIgnore(entry, timestamp, identifier)) {
return null;
}
// skip the is deleted flag
final boolean wasDeleted = entry.readBoolean();
if (wasDeleted) {
// this caters for the case when the entry in not in our hashLookupLiveOnly
// map but maybe in our hashLookupLiveAndDeleted,
// so we have to send the deleted notification
entry.position(timeStampPos);
entry.writeLong(timestamp);
entry.writeByte(identifier);
// was deleted is already true
entry.skip(1);
notifyRemoved(offset, key, null, pos);
return null;
}
long valueSize = readValueSize(entry);
V valueRemoved = expectedValue != null || !removeReturnsNull
? readValue(copies, entry, null, valueSize) : null;
if (expectedValue != null && !expectedValue.equals(valueRemoved))
return null;
hashLookupLiveOnly.remove(hash2, pos);
decrementSize();
entry.position(timeStampPos);
entry.writeLong(timestamp);
entry.writeByte(identifier);
entry.writeBoolean(true);
notifyRemoved(offset, key, valueRemoved, pos);
return valueRemoved;
}
// key is not found
return null;
} finally {
unlock();
}
}
@Override
IntIntMultiMap containsKeyHashLookup() {
return hashLookupLiveOnly;
}
/**
* @see VanillaChronicleMap.Segment#replace
*/
V replace(ThreadLocalCopies copies, MKI metaKeyWriter, KI keyWriter,
K key, V expectedValue, V newValue, int hash2, long timestamp) {
lock();
try {
long keySize = metaKeyWriter.size(keyWriter, key);
hashLookupLiveOnly.startSearch(hash2);
for (int pos; (pos = hashLookupLiveOnly.nextPos()) >= 0; ) {
long offset = offsetFromPos(pos);
NativeBytes entry = entry(offset);
if (!keyEquals(keyWriter, metaKeyWriter, key, keySize, entry))
continue;
// key is found
entry.skip(keySize);
if (shouldIgnore(entry, timestamp, localIdentifier))
return null;
// skip the is deleted flag
entry.skip(1);
return onKeyPresentOnReplace(copies, key, expectedValue, newValue, pos, offset,
entry, hashLookupLiveOnly);
}
// key is not found
return null;
} finally {
unlock();
}
}
@Override
void replacePosInHashLookupOnRelocation(IntIntMultiMap searchedHashLookup, int prevPos, int pos) {
searchedHashLookup.replacePrevPos(pos);
int hash = searchedHashLookup.getSearchHash();
IntIntMultiMap anotherLookup = searchedHashLookup == hashLookupLiveAndDeleted ?
hashLookupLiveOnly : hashLookupLiveAndDeleted;
anotherLookup.replace(hash, prevPos, pos);
}
public void dirtyEntries(final long timeStamp,
final ModificationIterator.EntryModifiableCallback callback) {
this.lock();
try {
final int index = Segment.this.getIndex();
hashLookupLiveAndDeleted.forEach(new IntIntMultiMap.EntryConsumer() {
@Override
public void accept(int hash, int pos) {
final NativeBytes entry = entry(offsetFromPos(pos));
long keySize = keySizeMarshaller.readSize(entry);
entry.skip(keySize);
final long entryTimestamp = entry.readLong();
if (entryTimestamp >= timeStamp &&
entry.readByte() == ReplicatedChronicleMap.this.identifier())
callback.set(index, pos);
}
});
} finally {
unlock();
}
}
@Override
public Entry getEntry(long pos) {
bytes.position(offsetFromPos(pos) + metaDataBytes);
long keySize = keySizeMarshaller.readSize(bytes);
ThreadLocalCopies copies = keyReaderProvider.getCopies(null);
K key = keyReaderProvider.get(copies, originalKeyReader).read(bytes, keySize);
long timestamp = bytes.readLong();
bytes.skip(2L); // identifier and isDeleted flag
long valueSize = valueSizeMarshaller.readSize(bytes);
alignment.alignPositionAddr(bytes);
copies = valueReaderProvider.getCopies(copies);
V value = valueReaderProvider.get(copies, originalValueReader).read(bytes, valueSize);
return new TimestampTrackingEntry(key, value, timestamp);
}
@Override
void clear() {
// we have to make sure that every calls notifies on remove,
// so that the replicators can pick it up
for (K k : keySet()) {
ReplicatedChronicleMap.this.remove(k);
}
}
}
class TimestampTrackingEntry extends SimpleEntry {
private static final long serialVersionUID = 0L;
transient long timestamp;
public TimestampTrackingEntry(K key, V value, long timestamp) {
super(key, value);
this.timestamp = timestamp;
}
@Override
public V setValue(V value) {
long newTimestamp = timestamp = timeProvider.currentTimeMillis();
put(getKey(), value, localIdentifier, newTimestamp);
return super.setValue(value);
}
}
class EntryIterator extends VanillaChronicleMap.EntryIterator {
@Override
void removePresent(VanillaChronicleMap.Segment seg, int pos) {
@SuppressWarnings("unchecked")
Segment segment = (Segment) seg;
final long offset = segment.offsetFromPos(pos);
final NativeBytes entry = segment.entry(offset);
final long limit = entry.limit();
final long keySize = keySizeMarshaller.readSize(entry);
long keyPosition = entry.position();
entry.skip(keySize);
long timestamp = entry.readLong();
entry.position(keyPosition);
if (timestamp > ((TimestampTrackingEntry) returnedEntry).timestamp) {
// The entry was updated after being returned from iterator.next()
// Check that it is still the entry with the same key
K key = returnedEntry.getKey();
ThreadLocalCopies copies = keyInteropProvider.getCopies(null);
KI keyWriter = keyInteropProvider.get(copies, originalKeyInterop);
copies = metaKeyInteropProvider.getCopies(copies);
MKI metaKeyWriter =
metaKeyInteropProvider.get(copies, originalMetaKeyInterop, keyWriter, key);
long returnedKeySize = metaKeyWriter.size(keyWriter, key);
if (returnedKeySize != keySize ||
!metaKeyWriter.startsWith(keyWriter, entry, key)) {
// The case:
// 1. iterator.next() - thread 1
// 2. map.put() which cause relocation of the key, returned above - thread 2
// OR map.remove() which remove this key - thread 2
// 3. map.put() which place a new key on the `pos` in current segment - thread 3
// 4. iterator.remove() - thread 1
ReplicatedChronicleMap.this.remove(key);
return;
}
}
final int segmentHash = segmentHash(hash(entry, keyPosition, keyPosition + keySize));
segment.hashLookupLiveOnly.remove(segmentHash, pos);
segment.decrementSize();
entry.skip(keySize);
entry.writeLong(timeProvider.currentTimeMillis());
entry.writeByte(localIdentifier);
entry.writeBoolean(true);
segment.notifyRemoved(offset, returnedEntry.getKey(), returnedEntry.getValue(), pos);
}
}
class EntrySet extends VanillaChronicleMap.EntrySet {
@NotNull
@Override
public Iterator> iterator() {
return new EntryIterator();
}
}
/**
* receive an update from the map, via the MapEventListener and delegates the changes to the currently
* active modification iterators
*/
class ModificationDelegator extends MapEventListener> {
private static final long serialVersionUID = 0L;
// the assigned modification iterators
private final ATSDirectBitSet modIterSet;
private final MapEventListener> nextListener;
private final AtomicReferenceArray modificationIterators =
new AtomicReferenceArray(127 + RESERVED_MOD_ITER);
public ModificationDelegator(@NotNull final MapEventListener> nextListener,
final Bytes bytes) {
this.nextListener = nextListener;
modIterSet = new ATSDirectBitSet(bytes);
}
public ModificationIterator acquireModificationIterator(
final short remoteIdentifier,
@NotNull final ModificationNotifier modificationNotifier) {
final ModificationIterator modificationIterator = modificationIterators.get(remoteIdentifier);
if (modificationIterator != null)
return modificationIterator;
synchronized (modificationIterators) {
final ModificationIterator modificationIterator0 = modificationIterators.get(remoteIdentifier);
if (modificationIterator0 != null)
return modificationIterator0;
final Bytes bytes = ms.bytes(startOfModificationIterators + (modIterBitSetSizeInBytes()
* remoteIdentifier),
modIterBitSetSizeInBytes());
final ModificationIterator newModificationIterator = new ModificationIterator(
bytes, modificationNotifier);
modificationIterators.set(remoteIdentifier, newModificationIterator);
modIterSet.set(remoteIdentifier);
return newModificationIterator;
}
}
@Override
public void onPut(ChronicleMap map, Bytes entry, int metaDataBytes,
boolean added, K key, V value, long pos, SharedSegment segment) {
assert ReplicatedChronicleMap.this == map :
"ModificationIterator.onPut() is called from outside of the parent map";
try {
nextListener.onPut(map, entry, metaDataBytes, added, key, value, pos, segment);
} catch (Exception e) {
LOG.error("", e);
}
for (long next = modIterSet.nextSetBit(0); next > 0; next = modIterSet.nextSetBit(next + 1)) {
try {
final ModificationIterator modificationIterator = modificationIterators.get((int) next);
modificationIterator.onPut(map, entry, metaDataBytes,
added, key, value, pos, segment);
} catch (Exception e) {
LOG.error("", e);
}
}
}
@Override
public void onRemove(ChronicleMap map, Bytes entry, int metaDataBytes,
K key, V value, int pos, SharedSegment segment) {
assert ReplicatedChronicleMap.this == map :
"ModificationIterator.onRemove() is called from outside of the parent map";
try {
nextListener.onRemove(map, entry, metaDataBytes, key, value, pos, segment);
} catch (Exception e) {
LOG.error("", e);
}
for (long next = modIterSet.nextSetBit(0); next > 0; next = modIterSet.nextSetBit(next + 1)) {
try {
final ModificationIterator modificationIterator = modificationIterators.get((int) next);
modificationIterator.onRemove(map, entry, metaDataBytes, key, value, pos, segment);
} catch (Exception e) {
LOG.error("", e);
}
}
}
@Override
void onRelocation(int pos, SharedSegment segment) {
for (long next = modIterSet.nextSetBit(0); next > 0; next = modIterSet.nextSetBit(next + 1)) {
try {
final ModificationIterator modificationIterator = modificationIterators.get((int) next);
modificationIterator.onRelocation(pos, segment);
} catch (Exception e) {
LOG.error("", e);
}
}
}
@Override
public void onGetFound(ChronicleMap map, Bytes entry, int metaDataBytes,
K key, V value) {
nextListener.onGetFound(map, entry, metaDataBytes, key, value);
}
/**
* Always throws {@code NotSerializableException} since instances of this class are not intended to be
* serializable.
*/
private void writeObject(ObjectOutputStream out) throws IOException {
throw new NotSerializableException(getClass().getCanonicalName());
}
/**
* Always throws {@code NotSerializableException} since instances of this class are not intended to be
* serializable.
*/
private void readObject(ObjectInputStream in)
throws IOException, ClassNotFoundException {
throw new NotSerializableException(getClass().getCanonicalName());
}
}
/**
* Once a change occurs to a map, map replication requires that these changes are picked up by another
* thread, this class provides an iterator like interface to poll for such changes. In most cases
* the thread that adds data to the node is unlikely to be the same thread that replicates the data over
* to the other nodes, so data will have to be marshaled between the main thread storing data to the map,
* and the thread running the replication.
One way to perform this marshalling, would be to pipe
* the data into a queue. However, This class takes another approach. It uses a bit set, and marks bits
* which correspond to the indexes of the entries that have changed. It then provides an iterator like
* interface to poll for such changes.
*
* @author Rob Austin.
*/
class ModificationIterator extends MapEventListener>
implements Replica.ModificationIterator {
private static final long serialVersionUID = 0L;
private final ModificationNotifier modificationNotifier;
private final ATSDirectBitSet changes;
private final int segmentIndexShift;
private final long posMask;
private final EntryModifiableCallback entryModifiableCallback = new EntryModifiableCallback();
// records the current position of the cursor in the bitset
private long position = -1;
/**
* @param bytes the back the bitset, used to mark which entries have changed
* @param modificationNotifier called when ever there is a change applied
*/
public ModificationIterator(@NotNull final Bytes bytes,
@NotNull final ModificationNotifier modificationNotifier) {
this.modificationNotifier = modificationNotifier;
long bitsPerSegment = bitsPerSegmentInModIterBitSet();
segmentIndexShift = Long.numberOfTrailingZeros(bitsPerSegment);
posMask = bitsPerSegment - 1;
changes = new ATSDirectBitSet(bytes);
}
/**
* used to merge multiple segments and positions into a single index used by the bit map
*
* @param segmentIndex the index of the maps segment
* @param pos the position within this {@code segmentIndex}
* @return and index the has combined the {@code segmentIndex} and {@code pos} into a single value
*/
private long combine(int segmentIndex, long pos) {
return (((long) segmentIndex) << segmentIndexShift) | pos;
}
@Override
public void onPut(ChronicleMap map, Bytes entry, int metaDataBytes,
boolean added, K key, V value, long pos, SharedSegment segment) {
assert ReplicatedChronicleMap.this == map :
"ModificationIterator.onPut() is called from outside of the parent map";
changes.set(combine(segment.getIndex(), pos));
modificationNotifier.onChange();
}
@Override
public void onRemove(ChronicleMap map, Bytes entry, int metaDataBytes,
K key, V value, int pos, SharedSegment segment) {
assert ReplicatedChronicleMap.this == map :
"ModificationIterator.onRemove() is called from outside of the parent map";
changes.set(combine(segment.getIndex(), pos));
modificationNotifier.onChange();
}
/**
* Ensures that garbage in the old entry's location won't be broadcast as changed entry.
*/
@Override
void onRelocation(int pos, SharedSegment segment) {
changes.clear(combine(segment.getIndex(), pos));
// don't call nextListener.onRelocation(),
// because no one event listener else overrides this method.
}
/**
* you can continue to poll hasNext() until data becomes available. If are are in the middle of
* processing an entry via {@code nextEntry}, hasNext will return true until the bit is cleared
*
* @return true if there is an entry
*/
public boolean hasNext() {
final long position = this.position;
return changes.nextSetBit(position == NOT_FOUND ? 0 : position) != NOT_FOUND ||
(position > 0 && changes.nextSetBit(0) != NOT_FOUND);
}
/**
* @param entryCallback call this to get an entry, this class will take care of the locking
* @return true if an entry was processed
*/
public boolean nextEntry(@NotNull final EntryCallback entryCallback, final int chronicleId) {
long position = this.position;
while (true) {
long oldPosition = position;
position = changes.nextSetBit(oldPosition + 1);
if (position == NOT_FOUND) {
if (oldPosition == NOT_FOUND) {
this.position = NOT_FOUND;
return false;
}
continue;
}
this.position = position;
final SharedSegment segment = segment((int) (position >>> segmentIndexShift));
segment.lock();
try {
if (changes.clearIfSet(position)) {
entryCallback.onBeforeEntry();
final long segmentPos = position & posMask;
final NativeBytes entry = segment.entry(segment.offsetFromPos(segmentPos));
// if the entry should be ignored, we'll move the next entry
final boolean success = entryCallback.onEntry(entry, chronicleId);
entryCallback.onAfterEntry();
if (success) {
return true;
}
}
// if the position was already cleared by another thread
// while we were trying to obtain segment lock (for example, in onReplication()),
// go to pick up next (next iteration in while (true) loop)
} finally {
segment.unlock();
}
}
}
@Override
public void dirtyEntries(long fromTimeStamp) {
// iterate over all the segments and mark bit in the modification iterator
// that correspond to entries with an older timestamp
for (final Segment segment : (Segment[]) segments) {
segment.dirtyEntries(fromTimeStamp, entryModifiableCallback);
}
}
/**
* Always throws {@code NotSerializableException} since instances of this class are not intended to be
* serializable.
*/
private void writeObject(ObjectOutputStream out) throws IOException {
throw new NotSerializableException(getClass().getCanonicalName());
}
/**
* Always throws {@code NotSerializableException} since instances of this class are not intended to be
* serializable.
*/
private void readObject(ObjectInputStream in)
throws IOException, ClassNotFoundException {
throw new NotSerializableException(getClass().getCanonicalName());
}
/**
* details about when a modification to an entry was made
*/
class EntryModifiableCallback {
/**
* set the bit related to {@code segment} and {@code pos}
*
* @param segmentIndex the segment relating to the bit to set
* @param pos the position relating to the bit to set
*/
public void set(int segmentIndex, int pos) {
final long combine = combine(segmentIndex, pos);
changes.set(combine);
}
}
}
}