ODIn/spark-instrumented-optimizer
1
0
Fork 0
Code Issues Pull requests Projects Releases Wiki Activity
2040 commits 6 branches 181 tags 397 MiB
Commit graph

229 commits

Author SHA1 Message Date
Matei Zaharia 8815aeba0c Take executor environment vars as an arguemnt to SparkContext 2012-10-13 15:31:11 -07:00
Josh Rosen 33cd3a0c12 Remove map-side combining from ShuffleMapTask. 
This separation of concerns simplifies the 
ShuffleDependency and ShuffledRDD interfaces.

Map-side combining can be performed in a
mapPartitions() call prior to shuffling the RDD.

I don't anticipate this having much of a 
performance impact: in both approaches, each tuple
is hashed twice: once in the bucket partitioning
and once in the combiner's hashtable.  The same
steps are being performed, but in a different
order and through one extra Iterator.
 2012-10-13 14:59:20 -07:00
Josh Rosen 10bcd217d2 Remove mapSideCombine field from Aggregator. 
Instead, the presence or absense of a ShuffleDependency's aggregator
will control whether map-side combining is performed.
 2012-10-13 14:59:20 -07:00
Josh Rosen 4775c55641 Change ShuffleFetcher to return an Iterator. 2012-10-13 14:59:20 -07:00
Matei Zaharia 682b2d9329 Added a test for when an RDD only partially fits in memory 2012-10-12 14:58:26 -07:00
Shivaram Venkataraman 8577523f37 Add test to verify if RDD is computed even if block manager has insufficient 
memory
 2012-10-12 14:14:57 -07:00
Shivaram Venkataraman 2cf40c5fd5 Change block manager to accept a ArrayBuffer instead of an iterator to ensure 
that the computation can proceed even if we run out of memory to cache the
block. Update CacheTracker to use this new interface
 2012-10-11 00:42:46 -07:00
Matei Zaharia efc5423210 Made compression configurable separately for shuffle, broadcast and RDDs 2012-10-07 11:30:53 -07:00
Reynold Xin 80f59e17e2 Fixed a bug in addFile that if the file is specified as "file:///", the 
symlink is created wrong for local mode.
 2012-10-07 00:54:38 -07:00
Matei Zaharia eca570f66a Removed the need to sleep in tests due to waiting for Akka to shut down 2012-10-07 00:17:59 -07:00
Matei Zaharia dc28a3ac0a Modified shuffle to limit the maximum outstanding data size in bytes, 
instead of the maximum number of outstanding fetches. This should make
it faster when there are many small map output files, as well as more
robust to overallocating memory on large map outputs.
 2012-10-06 20:07:10 -07:00
Matei Zaharia 9a3b3f32a3 Pass sizes of map outputs back to MapOutputTracker 2012-10-06 18:46:04 -07:00
Matei Zaharia 716e10ca32 Minor formatting fixes 2012-10-05 22:03:06 -07:00
Andy Konwinski a242cdd0a6 Factor subclasses of RDD out of RDD.scala into their own classes 
in the rdd package.
 2012-10-05 19:53:54 -07:00
Andy Konwinski e0067da082 Moves all files in core/src/main/scala/ that have RDD in them from 
package spark to package spark.rdd and updates all references to them.
 2012-10-05 19:23:45 -07:00
Shivaram Venkataraman b6e4f46a96 Fix SizeEstimator tests to work with String classes in JDK 6 and 7 
Conflicts:

	core/src/test/scala/spark/BoundedMemoryCacheSuite.scala
 2012-10-05 16:58:57 -07:00
Imran Rashid e0698f8f26 change tests to show utility of localValue 2012-10-04 23:05:42 -07:00
Imran Rashid 82a3327862 make accumulator.localValue public, add tests 
Conflicts:
	core/src/test/scala/spark/AccumulatorSuite.scala
 2012-10-04 23:05:01 -07:00
Matei Zaharia 97cbd699d7 Merge branch 'dev' of github.com:mesos/spark into dev 2012-10-02 17:31:01 -07:00
Matei Zaharia 5fda59ab99 Added a test for overly large blocks in memory store 2012-10-02 17:30:40 -07:00
Matei Zaharia 6098f7e87a Fixed cache replacement behavior of BlockManager: 
- Partitions that get dropped to disk will now be loaded back into RAM
  after they're accessed again
- Same-RDD rule for cache replacement is now implemented (don't drop
  partitions from an RDD to make room for other partitions from itself)
- Items stored as MEMORY_AND_DISK go into memory only first, instead of
  being eagerly written out to disk
- MemoryStore.ensureFreeSpace is called within a lock on the writer
  thread to prevent race conditions (this can still be optimized to
  allow multiple concurrent calls to it but it's a start)
- MemoryStore does not accept blocks larger than its limit
 2012-10-02 17:25:38 -07:00
Reynold Xin 0898a21b95 Merge branch 'dev' of https://github.com/mesos/spark into dev 2012-10-02 13:08:01 -07:00
Matei Zaharia 22684653a5 Revert "Place Spray repo ahead of Cloudera in Maven search path" 
This reverts commit 42e0a68082.
 2012-10-02 12:01:32 -07:00
Reynold Xin b8cd681169 Allow whitespaces in cluster URL configuration for local cluster. 2012-10-02 11:52:12 -07:00
Matei Zaharia 42e0a68082 Place Spray repo ahead of Cloudera in Maven search path 2012-10-02 11:37:19 -07:00
Matei Zaharia 74a9244255 Write all unit test output to a file 2012-10-01 15:07:42 -07:00
Matei Zaharia 0b84871dbc Remove some printlns in tests 2012-10-01 10:57:26 -07:00
Matei Zaharia 2314132d57 Added a (failing) test for LRU with MEMORY_AND_DISK. 2012-09-30 22:52:16 -07:00
Matei Zaharia 83143f9a5f Fixed several bugs that caused weird behavior with files in spark-shell: 
- SizeEstimator was following through a ClassLoader field of Hadoop
  JobConfs, which referenced the whole interpreter, Scala compiler, etc.
  Chaos ensued, giving an estimated size in the tens of gigabytes.
- Broadcast variables in local mode were only stored as MEMORY_ONLY and
  never made accessible over a server, so they fell out of the cache when
  they were deemed too large and couldn't be reloaded.
 2012-09-30 21:19:39 -07:00
Matei Zaharia fd0374b9de Comment 2012-09-29 21:43:06 -07:00
Matei Zaharia 143ef4f90d Added a CoalescedRDD class for reducing the number of partitions in an RDD. 2012-09-29 21:30:52 -07:00
Matei Zaharia c45758ddde Comment 2012-09-29 20:27:54 -07:00
Matei Zaharia 9b326d01e9 Made BlockManager unmap memory-mapped files when necessary to reduce the 
number of open files. Also optimized sending of disk-based blocks.
 2012-09-29 20:21:54 -07:00
Matei Zaharia 009b0e37e7 Added an option to compress blocks in the block store 2012-09-27 18:45:44 -07:00
Matei Zaharia 7bcb08cef5 Renamed storage levels to something cleaner; fixes #223. 2012-09-27 17:50:59 -07:00
Matei Zaharia 920fab23c3 Merge pull request #222 from rxin/dev 
Added MapPartitionsWithSplitRDD.
 2012-09-26 23:16:45 -07:00
Matei Zaharia 1ef4f0fbd2 Allow controlling number of splits in sortByKey. 2012-09-26 19:18:47 -07:00
Reynold Xin 1ad1331a34 Added MapPartitionsWithSplitRDD. 2012-09-26 17:11:28 -07:00
Matei Zaharia d71a358c46 Fixed a test that was getting extremely lucky before, and increased the 
number of samples used for sorting
 2012-09-26 00:25:34 -07:00
Matei Zaharia 6eeb379cf8 Fix some test issues 2012-09-24 15:39:58 -07:00
Reynold Xin 397d3816e1 Separated ShuffledRDD into multiple classes: RepartitionShuffledRDD, 
ShuffledSortedRDD, and ShuffledAggregatedRDD.
 2012-09-19 12:31:45 -07:00
Denny 5e4076e3f2 Merge branch 'dev' into feature/fileserver 
Conflicts:
	core/src/main/scala/spark/SparkContext.scala
 2012-09-11 16:57:17 -07:00
Matei Zaharia 6d7f907e73 Manually merge pull request #175 by Imran Rashid 2012-09-11 16:00:06 -07:00
Denny 4d3471dd07 Fix serialization bugs and added local cluster tests 2012-09-10 15:39:58 -07:00
Denny b864c36a30 Dynamically adding jar files and caching fileSets. 2012-09-10 12:49:09 -07:00
Denny f275fb07da General FileServer 
A general fileserver for both JARs and regular files.
 2012-09-10 12:48:59 -07:00
Matei Zaharia a13780670d Added a unit test for local-cluster mode and simplified some of the code involved in that 2012-09-10 12:48:58 -07:00
Matei Zaharia 995982b3c9 Added a unit test for local-cluster mode and simplified some of the code involved in that 2012-09-07 17:08:36 -07:00
Reynold Xin c308fbcb79 Removed cache add/remove log messages from CacheTracker. 
Added log messages on BlockManagerMaster to reflect block add/remove.
Also did some minor cleanup of storage package code.
 2012-09-05 15:59:48 -07:00
Reynold Xin a8a2a08a1a Added a test for testing map-side combine on/off switch. 2012-08-30 12:34:28 -07:00
Matei Zaharia 2c16ae36d7 Set log level in tests to WARN 2012-08-23 20:38:14 -07:00
Matei Zaharia deedb9e7b7 Fix further issues with tests and broadcast. 
The broadcast fix is to store values as MEMORY_ONLY_DESER instead of
MEMORY_ONLY, which will save substantial time on serialization.
 2012-08-23 20:31:49 -07:00
Shivaram Venkataraman 0f4fbb057b Change BlockManagerSuite test cases to use a deterministic size estimator and 
update the results to match the new estimates
 2012-08-13 13:32:23 -07:00
Shivaram Venkataraman 22ba3a3f77 Add test-cases for 32-bit and no-compressed oops scenarios. 2012-08-13 13:32:10 -07:00
Shivaram Venkataraman 1f68c4b03b Update test cases to match the new size estimates. Uses 64-bit and compressed 
oops setting to get deterministic results
 2012-08-13 13:31:54 -07:00
Matei Zaharia 6ae3c375a9 Renamed apply() to call() in Java API and allowed it to throw Exceptions 2012-08-12 23:10:19 +02:00
Matei Zaharia e463e7a333 Merge pull request #167 from JoshRosen/piped-rdd-fixes 
Detect non-zero exit status from PipedRDD process
 2012-08-10 00:56:42 -07:00
Shivaram Venkataraman ce3444d2cb Fix testcheckpoint to reuse spark context defined in the class 2012-08-03 18:52:26 -07:00
Matei Zaharia 62898b631f Made range partition balance tests more aggressive. 
This is because we pull out such a large sample (10x the number of
partitions) that we should expect pretty good balance. The tests are
also deterministic so there's no worry about them failing irreproducibly.
 2012-08-03 16:46:48 -04:00
Matei Zaharia 6601a6212b Added a unit test for cross-partition balancing in sort, and changes to 
RangePartitioner to make it pass. It turns out that the first partition
was always kind of small due to how we picked partition boundaries.
 2012-08-03 16:40:45 -04:00
Matei Zaharia 3ee2530c0c Merge branch 'block-manager-fix' into dev 2012-07-30 13:58:46 -07:00
Matei Zaharia 400221f851 Merge branch 'dev' of git://github.com/tdas/spark into dev 2012-07-30 13:54:57 -07:00
Matei Zaharia ed1b0f8388 Made BlockManagerMaster no longer be a singleton. 
Also cleaned up a few formatting things throughout block manager code.
 2012-07-30 13:53:47 -07:00
Matei Zaharia d7f089323a Fixed AccumulatorSuite to clean up SparkContext with BeforeAndAfter 2012-07-28 20:25:42 -07:00
Imran Rashid f7149c5e46 tasks cannot access value of accumulator 2012-07-28 20:16:17 -07:00
Imran Rashid f1face1ea9 rename addToAccum to addAccumulator 2012-07-28 20:16:01 -07:00
Imran Rashid 2d666b9d76 add some functionality to Vector, delete copy in AccumulatorSuite 2012-07-28 20:15:51 -07:00
Imran Rashid 83659af11c Accumulator now inherits from Accumulable, whcih simplifies a bunch of other things (eg., no +:=) 
Conflicts:

	core/src/main/scala/spark/Accumulators.scala
 2012-07-28 20:13:51 -07:00
Imran Rashid ae07f3864c add Accumulatable, add corresponding docs & tests for accumulators 2012-07-28 20:12:41 -07:00
Matei Zaharia f6f917bd00 Add a sleep to prevent a failing test. 
The BlockManager's put seems to be slightly asynchronous, which can
cause it to fail this test by not removing stuff from the cache before
we put the next value. We should probably change the semantics of put()
in this case but it's hard right now. It will also be hard for
asynchronously replicated puts.
 2012-07-27 16:59:36 -07:00
Matei Zaharia c0c78d2119 Renamed test more descriptively 2012-07-27 16:28:18 -07:00
Matei Zaharia dee8ff1b9d Added a second version of union() without varargs. 2012-07-27 16:27:52 -07:00
Matei Zaharia b51d733a57 Fixed Java union methods having same erasure. 
Changed union() methods on lists to take a separate "first element"
argument in order to differentiate them to the compiler, because Java 7
considered it an error to have them all take Lists parameterized with
different types.
 2012-07-27 12:23:27 -07:00
Tathagata Das 024905f682 Added BlockRDD and a first-cut version of checkpoint() to RDD class. 2012-07-27 12:00:49 -07:00
Tathagata Das 0426769f89 Modified the block dropping code for better performance. 2012-07-26 20:53:45 -07:00
Matei Zaharia 5c5aa2ff81 Merge pull request #153 from JoshRosen/new-java-api 
Java API
 2012-07-26 17:20:52 -07:00
Josh Rosen c5e2810dc7 Add persist(), splits(), glom(), and mapPartitions() to Java API. 2012-07-26 12:46:47 -07:00
Josh Rosen bf61c10072 Detect non-zero exit status from PipedRDD process. 2012-07-26 11:32:59 -07:00
Denny 4f4a34c025 Stlystic changes 
Conflicts:

	core/src/test/scala/spark/MesosSchedulerSuite.scala
 2012-07-23 16:32:20 -07:00
Denny 866e6949df Always destroy SparkContext in after block for the unit tests. 
Conflicts:

	core/src/test/scala/spark/ShuffleSuite.scala
 2012-07-23 16:29:17 -07:00
Josh Rosen 042dcbde33 Add type annotations to Java API methods. 
Add missing Scala Map to java.util.Map conversions.
 2012-07-22 17:35:29 -07:00
Josh Rosen 01dce3f569 Add Java API 
Add distinct() method to RDD.

Fix bug in DoubleRDDFunctions.
 2012-07-18 17:34:29 -07:00
Matei Zaharia 408b5a1332 More work on deploy code (adding Worker class) 2012-06-30 16:45:57 -07:00
Matei Zaharia 2fb6e7d71e Initial framework to get a master and web UI up. 2012-06-30 14:45:55 -07:00
Matei Zaharia c53670b9bf Various code style fixes, mostly from IntelliJ IDEA 2012-06-29 18:47:12 -07:00
Matei Zaharia 3920189932 Upgraded to Akka 2 and fixed test execution (which was still parallel 
across projects).
 2012-06-28 23:51:28 -07:00
Tathagata Das e896a505e2 Added testcase for ByteBufferInputStream bugs. 2012-06-17 16:11:12 -07:00
Matei Zaharia f58da6164e Merge branch 'master' into dev 2012-06-15 23:47:11 -07:00
Tathagata Das c6156da9e2 Multiple bug fixes to pass the testsuites ShuffleSuite and BlockManagerSuite. 2012-06-13 16:26:49 -04:00
Matei Zaharia e75b1b5cb4 Change the default broadcast implementation to a simple HTTP-based 
broadcast. Fixes #139.
 2012-06-09 15:58:07 -07:00
Matei Zaharia a96558caa3 Performance improvements to shuffle operations: in particular, preserve 
RDD partitioning in more cases where it's possible, and use iterators
instead of materializing collections when doing joins.
 2012-06-09 14:44:18 -07:00
Matei Zaharia c2c7299d7a Added BlockManagerSuite, which I'd forgotten to merge. 2012-06-07 13:47:10 -07:00
Matei Zaharia 63051dd2bc Merge in engine improvements from the Spark Streaming project, developed 
jointly with Tathagata Das and Haoyuan Li. This commit imports the changes
and ports them to Mesos 0.9, but does not yet pass unit tests due to
various classes not supporting a graceful stop() yet.
 2012-06-07 12:45:38 -07:00
Matei Zaharia 6ae2746d1e Handle arrays that contain the same element many times better in 
SizeEstimator. Also added a test for SizeEstimator. Fixes #136.
 2012-06-06 16:13:02 -07:00
Matei Zaharia 0a617958d1 Some refactoring to make BoundedMemoryCache test similar to others 2012-06-06 16:12:08 -07:00
Matei Zaharia e141f644ca Merge pull request #132 from Benky/rb-first-iteration 
Little refactoring and unit tests for CacheTrackerActor
 2012-05-26 13:15:06 -07:00
Richard Benkovsky ae64920337 MesosScheduler refactoring 2012-05-22 11:04:54 +02:00
Richard Benkovsky 3a1bcd4028 Added tests for CacheTrackerActor 2012-05-22 11:04:54 +02:00
Richard Benkovsky 518506a7c5 Added tests for Utils.copyStream 2012-05-22 11:04:51 +02:00
Richard Benkovsky 565245871f BoundedMemoryCache.put fails when estimated size of 'value' is larger than cache capacity 2012-05-20 22:13:35 +02:00
Reynold Xin 16461e2eda Updated Cache's put method to use a case class for response. Previously 
it was pretty ugly that put() should return -1 for failures.
 2012-05-15 00:31:52 -07:00
Reynold Xin 019e48833f Added the capacity to report cache usage status back to the cache 
trackor. This is essential for building a dashboard to see the status of
caches on all slaves.
 2012-05-14 18:39:04 -07:00
Reynold Xin 761ea65a98 Added a test for the previous commit (failing to serialize task results 
would throw an exception for local tasks).
 2012-04-24 15:14:35 -07:00
Reynold Xin e601b3b9e5 Added the ability to set environmental variables in piped rdd. 2012-04-17 16:40:56 -07:00
Matei Zaharia c7af538ac1 Some fixes to sorting for when the RDD has fewer elements than the 
number of partitions we ask to partition it into. Also, removed a test
that was taking way too long to run.
 2012-03-17 13:08:36 -07:00
Matei Zaharia 1e10df0a46 Merge pull request #111 from alupher/master 
Adding sorting to RDDs
 2012-02-24 15:50:14 -08:00
Antonio 0d93d95bcf Removed unnecessary import 2012-02-21 19:57:12 -08:00
Antonio 2990298f71 Added sorting testing suite 2012-02-21 19:54:21 -08:00
Matei Zaharia aa04f87cd2 Added support for parallel execution of jobs in DAGScheduler. 2012-02-19 22:50:23 -08:00
Matei Zaharia a766780f4c Added some tests for multithreaded access to Spark. 2012-02-09 22:27:53 -08:00
Matei Zaharia 43a3335090 Simplifying test 2012-02-05 22:46:51 -08:00
Matei Zaharia eb05154b7a Fixed a failure recovery bug and added some tests for fault recovery. 2012-01-13 19:08:25 -08:00
Matei Zaharia e269f6f7ea Register RDDs with the MapOutputTracker even if they have no partitions. 
Fixes #105.
 2012-01-05 15:59:20 -05:00
Matei Zaharia 735843a049 Merge remote-tracking branch 'origin/charles-newhadoop' 2011-12-02 21:59:30 -08:00
Charles Reiss 66f05f383e Add new Hadoop API reading support. 2011-12-01 14:02:10 -08:00
Charles Reiss 02d43e6986 Add new Hadoop API writing support. 2011-12-01 14:01:28 -08:00
Matei Zaharia 22b8fcf632 Added fold() and aggregate() operations that reuse an object to 
merge results into rather than requiring a new object allocation
for each element merged. Fixes #95.
 2011-11-30 11:37:47 -08:00
Matei Zaharia 9e4c79a4d3 Closure cleaner unit test 2011-11-08 00:40:15 -08:00
Matei Zaharia c2b7fd6899 Make parallelize() work efficiently for ranges of Long, Double, etc 
(splitting them into sub-ranges). Fixes #87.
 2011-11-02 15:16:02 -07:00
Matei Zaharia d12122502b Various improvements to Kryo serializer: 
- Replaced modified Kryo version with the standard one augmented with
  the kryo-serializers package, which includes support for classes with
  no-arg constructors (that was why we had a modified Kryo before)
- The kryo-serializers version also fixes issue #72.
- Added a bunch of tests.
- Serialize maps and a few other common types properly by default.
 2011-07-21 22:09:33 -07:00
Matei Zaharia e4c3402d2d Renamed ParallelArray to ParallelCollection 2011-07-14 14:47:01 -04:00
Matei Zaharia 2604939f64 Simplified and documented code a little and added test 2011-07-14 00:19:00 -04:00
Matei Zaharia 9c0069188b Updated save code to allow non-file-based OutputFormats and added a test 
for file-related stuff
 2011-07-13 23:04:06 -04:00
Matei Zaharia 842e14d567 Added mapPartitions operation and a bunch of tests for RDD ops 2011-07-13 00:19:52 -04:00
Olivier Grisel 2e3531d8bf Implemented RDD.leftOuterJoin and RDD.rightOuterJoin 2011-06-24 11:00:51 +02:00
Olivier Grisel 005d1605a4 add missing test for RDD.groupWith 2011-06-23 02:10:52 +02:00
Ismael Juma 1396678baa Move REPL classes to separate module. 2011-05-27 11:22:50 +01:00
Matei Zaharia 4db50e26c7 Fixed unit tests by making them clean up the SparkContext after use and 
thus clean up the various singletons (RDDCache, MapOutputTracker, etc).
This isn't perfect yet (ideally we shouldn't use singleton objects at
all) but we can fix that later.
 2011-05-13 12:03:58 -07:00
Matei Zaharia e5c4cd8a5e Made examples and core subprojects 2011-02-01 15:11:08 -08:00