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
|
package spark
import scala.actors._
import scala.actors.Actor._
import scala.actors.remote._
import scala.collection.mutable.HashMap
import scala.collection.mutable.HashSet
sealed trait CacheTrackerMessage
case class AddedToCache(rddId: Int, partition: Int, host: String, size: Long = 0L)
extends CacheTrackerMessage
case class DroppedFromCache(rddId: Int, partition: Int, host: String, size: Long = 0L)
extends CacheTrackerMessage
case class MemoryCacheLost(host: String) extends CacheTrackerMessage
case class RegisterRDD(rddId: Int, numPartitions: Int) extends CacheTrackerMessage
case class SlaveCacheStarted(host: String, size: Long) extends CacheTrackerMessage
case object GetCacheStatus extends CacheTrackerMessage
case object GetCacheLocations extends CacheTrackerMessage
case object StopCacheTracker extends CacheTrackerMessage
class CacheTrackerActor extends DaemonActor with Logging {
private val locs = new HashMap[Int, Array[List[String]]]
/**
* A map from the slave's host name to its cache size.
*/
private val slaveCapacity = new HashMap[String, Long]
private val slaveUsage = new HashMap[String, Long]
// TODO: Should probably store (String, CacheType) tuples
private def getCacheUsage(host: String): Long = slaveUsage.getOrElse(host, 0L)
private def getCacheCapacity(host: String): Long = slaveCapacity.getOrElse(host, 0L)
private def getCacheAvailable(host: String): Long = getCacheCapacity(host) - getCacheUsage(host)
def act() {
val port = System.getProperty("spark.master.port").toInt
RemoteActor.alive(port)
RemoteActor.register('CacheTracker, self)
logInfo("Registered actor on port " + port)
loop {
react {
case SlaveCacheStarted(host: String, size: Long) =>
logInfo("Started slave cache (size %s) on %s".format(
Utils.memoryBytesToString(size), host))
slaveCapacity.put(host, size)
slaveUsage.put(host, 0)
reply('OK)
case RegisterRDD(rddId: Int, numPartitions: Int) =>
logInfo("Registering RDD " + rddId + " with " + numPartitions + " partitions")
locs(rddId) = Array.fill[List[String]](numPartitions)(Nil)
reply('OK)
case AddedToCache(rddId, partition, host, size) =>
if (size > 0) {
slaveUsage.put(host, getCacheUsage(host) + size)
logInfo("Cache entry added: (%s, %s) on %s (size added: %s, available: %s)".format(
rddId, partition, host, Utils.memoryBytesToString(size),
Utils.memoryBytesToString(getCacheAvailable(host))))
} else {
logInfo("Cache entry added: (%s, %s) on %s".format(rddId, partition, host))
}
locs(rddId)(partition) = host :: locs(rddId)(partition)
reply('OK)
case DroppedFromCache(rddId, partition, host, size) =>
if (size > 0) {
logInfo("Cache entry removed: (%s, %s) on %s (size dropped: %s, available: %s)".format(
rddId, partition, host, Utils.memoryBytesToString(size),
Utils.memoryBytesToString(getCacheAvailable(host))))
slaveUsage.put(host, getCacheUsage(host) - size)
// Do a sanity check to make sure usage is greater than 0.
val usage = getCacheUsage(host)
if (usage < 0) {
logError("Cache usage on %s is negative (%d)".format(host, usage))
}
} else {
logInfo("Cache entry removed: (%s, %s) on %s".format(rddId, partition, host))
}
locs(rddId)(partition) = locs(rddId)(partition).filterNot(_ == host)
reply('OK)
case MemoryCacheLost(host) =>
logInfo("Memory cache lost on " + host)
// TODO: Drop host from the memory locations list of all RDDs
case GetCacheLocations =>
logInfo("Asked for current cache locations")
reply(locs.map{case (rrdId, array) => (rrdId -> array.clone())})
case GetCacheStatus =>
val status = slaveCapacity.map { case (host,capacity) =>
(host, capacity, getCacheUsage(host))
}.toSeq
reply(status)
case StopCacheTracker =>
reply('OK)
exit()
}
}
}
}
class CacheTracker(isMaster: Boolean, theCache: Cache) extends Logging {
// Tracker actor on the master, or remote reference to it on workers
var trackerActor: AbstractActor = null
val registeredRddIds = new HashSet[Int]
// Stores map results for various splits locally
val cache = theCache.newKeySpace()
if (isMaster) {
val tracker = new CacheTrackerActor
tracker.start()
trackerActor = tracker
} else {
val host = System.getProperty("spark.master.host")
val port = System.getProperty("spark.master.port").toInt
trackerActor = RemoteActor.select(Node(host, port), 'CacheTracker)
}
// Report the cache being started.
trackerActor !? SlaveCacheStarted(Utils.getHost, cache.getCapacity)
// Remembers which splits are currently being loaded (on worker nodes)
val loading = new HashSet[(Int, Int)]
// Registers an RDD (on master only)
def registerRDD(rddId: Int, numPartitions: Int) {
registeredRddIds.synchronized {
if (!registeredRddIds.contains(rddId)) {
logInfo("Registering RDD ID " + rddId + " with cache")
registeredRddIds += rddId
trackerActor !? RegisterRDD(rddId, numPartitions)
}
}
}
// Get a snapshot of the currently known locations
def getLocationsSnapshot(): HashMap[Int, Array[List[String]]] = {
(trackerActor !? GetCacheLocations) match {
case h: HashMap[_, _] => h.asInstanceOf[HashMap[Int, Array[List[String]]]]
case _ => throw new SparkException("Internal error: CacheTrackerActor did not reply with a HashMap")
}
}
// Get the usage status of slave caches. Each tuple in the returned sequence
// is in the form of (host name, capacity, usage).
def getCacheStatus(): Seq[(String, Long, Long)] = {
(trackerActor !? GetCacheStatus) match {
case h: Seq[(String, Long, Long)] => h.asInstanceOf[Seq[(String, Long, Long)]]
case _ =>
throw new SparkException(
"Internal error: CacheTrackerActor did not reply with a Seq[Tuple3[String, Long, Long]")
}
}
// Gets or computes an RDD split
def getOrCompute[T](rdd: RDD[T], split: Split)(implicit m: ClassManifest[T]): Iterator[T] = {
logInfo("Looking for RDD partition %d:%d".format(rdd.id, split.index))
val cachedVal = cache.get(rdd.id, split.index)
if (cachedVal != null) {
// Split is in cache, so just return its values
logInfo("Found partition in cache!")
return cachedVal.asInstanceOf[Array[T]].iterator
} else {
// Mark the split as loading (unless someone else marks it first)
val key = (rdd.id, split.index)
loading.synchronized {
while (loading.contains(key)) {
// Someone else is loading it; let's wait for them
try { loading.wait() } catch { case _ => }
}
// See whether someone else has successfully loaded it. The main way this would fail
// is for the RDD-level cache eviction policy if someone else has loaded the same RDD
// partition but we didn't want to make space for it. However, that case is unlikely
// because it's unlikely that two threads would work on the same RDD partition. One
// downside of the current code is that threads wait serially if this does happen.
val cachedVal = cache.get(rdd.id, split.index)
if (cachedVal != null) {
return cachedVal.asInstanceOf[Array[T]].iterator
}
// Nobody's loading it and it's not in the cache; let's load it ourselves
loading.add(key)
}
// If we got here, we have to load the split
// Tell the master that we're doing so
// TODO: fetch any remote copy of the split that may be available
logInfo("Computing partition " + split)
var array: Array[T] = null
var putResponse: CachePutResponse = null
try {
array = rdd.compute(split).toArray(m)
putResponse = cache.put(rdd.id, split.index, array)
} finally {
// Tell other threads that we've finished our attempt to load the key (whether or not
// we've actually succeeded to put it in the map)
loading.synchronized {
loading.remove(key)
loading.notifyAll()
}
}
putResponse match {
case CachePutSuccess(size) => {
// Tell the master that we added the entry. Don't return until it
// replies so it can properly schedule future tasks that use this RDD.
trackerActor !? AddedToCache(rdd.id, split.index, Utils.getHost, size)
}
case _ => null
}
return array.iterator
}
}
// Called by the Cache to report that an entry has been dropped from it
def dropEntry(datasetId: Any, partition: Int) {
datasetId match {
//TODO - do we really want to use '!!' when nobody checks returned future? '!' seems to enough here.
case (cache.keySpaceId, rddId: Int) => trackerActor !! DroppedFromCache(rddId, partition, Utils.getHost)
}
}
def stop() {
trackerActor !? StopCacheTracker
registeredRddIds.clear()
trackerActor = null
}
}
|
