Commit graph

641 commits

Author SHA1 Message Date
Charles Reiss 24d7aa2d15 Extra whitespace in ExecutorExitCode 2012-12-13 18:39:23 -08:00
Charles Reiss 829206f1a7 Explain slaveLost calls made by StandaloneSchedulerBackend 2012-12-13 16:23:36 -08:00
Charles Reiss a4041dd87f Log duplicate slaveLost() calls in ClusterScheduler. 2012-12-13 16:23:36 -08:00
Charles Reiss fa9df4a45d Normalize executor exit statuses and report them to the user. 2012-12-13 16:23:31 -08:00
Charles Reiss 1d8e2e6cff Call slaveLost on executor death for standalone clusters. 2012-12-12 21:15:34 -08:00
Reynold Xin 21b271f5bd Suppress shuffle block updates when a slave node comes back. 2012-12-10 20:36:03 -08:00
Matei Zaharia a1a2daa7ef Merge pull request #317 from woggling/block-manager-heartbeat
Implement block manager heartbeat
2012-12-10 11:03:55 -08:00
Charles Reiss b6b62d774f Decrease BlockManagerMaster logging verbosity 2012-12-10 00:31:55 -08:00
Charles Reiss 5d3e917d09 Use Akka scheduler for BlockManager heart beats.
Adds required ActorSystem argument to BlockManager constructors.
2012-12-10 00:31:50 -08:00
Charles Reiss b53dd28c90 Changed default block manager heartbeat interval to 5 s 2012-12-09 23:03:34 -08:00
Matei Zaharia beb440089e Merge pull request #310 from tomdz/master-mavenized
Maven build setup
2012-12-09 21:40:05 -08:00
Matei Zaharia e1d7cd2276 Search for a non-loopback address in Utils.getLocalIpAddress 2012-12-08 00:33:11 -08:00
Charles Reiss 714c8d32d5 Don't divide by milliseconds by 1000 more. 2012-12-06 18:38:34 -08:00
Charles Reiss 8f0819520c map -> foreach 2012-12-06 18:29:50 -08:00
Charles Reiss 7a033fd795 Make LocalSparkCluster use distinct IPs 2012-12-06 00:03:08 -08:00
Charles Reiss a2a94fdbc7 Tests for block manager heartbeats. 2012-12-05 23:36:05 -08:00
Charles Reiss d21ca010ac Add block manager heart beats.
Renames old message called 'HeartBeat' to 'BlockUpdate'.

The BlockManager periodically sends a heart beat message to the master.
If the manager is currently not registered. The master responds to the
heart beat by indicating whether the BlockManager is currently registered
with the master. Additionally, the master now also responds to block
updates by indicating whether the BlockManager in question is registered.
When the BlockManager detects (by heart beat or failed block update)
that it stopped being registered, it reregisters and sends block
updates for all its blocks.
2012-12-05 23:35:20 -08:00
Charles Reiss c9e54a6755 Track block managers by hostname; handle manager removal. 2012-12-05 23:35:20 -08:00
Charles Reiss 5afa2ee9e9 Actually put millis in _lastSeenMs 2012-12-05 23:35:20 -08:00
Charles Reiss 813ac71459 Don't use bogus port number in notifyADeadHost(). 2012-12-05 23:35:20 -08:00
Josh Rosen cdaa0fad51 Use external addresses in standalone WebUI on EC2. 2012-12-01 18:19:13 -08:00
Thomas Dudziak 84e584fa8c Code review feedback fix 2012-11-28 19:46:06 -08:00
Matei Zaharia f86960cba9 Merge pull request #313 from rxin/pde_size_compress
Added a partition preserving flag to MapPartitionsWithSplitRDD.
2012-11-27 22:39:25 -08:00
Matei Zaharia 3ebd8e1885 Added zip to Java API 2012-11-27 22:38:09 -08:00
Matei Zaharia 27e43abd19 Added a zip() operation for RDDs with the same shape (number of
partitions and number of elements in each partition)
2012-11-27 22:27:47 -08:00
Matei Zaharia f410a111ad Merge branch 'master' of github.com:mesos/spark 2012-11-27 20:51:58 -08:00
Josh Rosen 7d71b9a56a Fix NullPointerException caused by unregistered map outputs. 2012-11-27 20:51:51 -08:00
Matei Zaharia 935c468b71 Merge pull request #311 from woggling/map-output-npe
Fix NullPointerException when map output unregistered from MapOutputTracker twice
2012-11-27 20:50:48 -08:00
Reynold Xin bd6dd1a3a6 Added a partition preserving flag to MapPartitionsWithSplitRDD. 2012-11-27 19:43:30 -08:00
Reynold Xin f24bfd2dd1 For size compression, compress non zero values into non zero values. 2012-11-27 19:20:45 -08:00
Thomas Dudziak 3b643e86bc Updated versions in the pom.xml files to match current master 2012-11-27 17:50:42 -08:00
Charles Reiss cf79de425d Fix NullPointerException when unregistering a map output twice. 2012-11-27 16:12:05 -08:00
Charles Reiss 5fa868b98b Tests for MapOutputTracker. 2012-11-27 16:05:36 -08:00
Thomas Dudziak 69297c64be Addressed code review comments 2012-11-27 15:45:16 -08:00
Thomas Dudziak 811a32257b Added maven and debian build files 2012-11-20 16:19:51 -08:00
Matei Zaharia 3ff6f4bdee Merge pull request #304 from mbautin/configurable_local_ip
SPARK-624: make the default local IP customizable
2012-11-19 13:23:39 -08:00
mbautin 00f4e3ff9c Addressing Matei's comment: SPARK_LOCAL_IP environment variable 2012-11-19 11:52:10 -08:00
Charles Reiss 12c24e786c Set default uncaught exception handler to exit.
Among other things, should prevent OutOfMemoryErrors in some daemon threads
(such as the network manager) from causing a spark executor to enter a state
where it cannot make progress but does not report an error.
2012-11-16 20:12:31 -08:00
mbautin 1f5a7e0e64 SPARK-624: make the default local IP customizable 2012-11-15 13:57:47 -08:00
Matei Zaharia c23a74df0a Use DNS names instead of IP addresses in standalone mode, to allow
matching with data locality hints from storage systems.
2012-11-15 00:10:52 -08:00
Matei Zaharia 173e0354c0 Detect correctly when one has disconnected from a standalone cluster.
SPARK-617 #resolve
2012-11-11 21:06:57 -08:00
root acf8272324 Fix K-means example a little 2012-11-10 23:07:21 -08:00
Tathagata Das 9915989bfa Incorporated Matei's suggestions. Tested with 5 producer(consumer) threads each doing 50k puts (gets), took 15 minutes to run, no errors or deadlocks. 2012-11-09 15:46:15 -08:00
Tathagata Das de00bc63db Fixed deadlock in BlockManager.
1. Changed the lock structure of BlockManager by replacing the 337 coarse-grained locks to use BlockInfo objects as per-block fine-grained locks.
2. Changed the MemoryStore lock structure by making the block putting threads lock on a different object (not the memory store) thus making sure putting threads minimally blocks to the getting treads.
3. Added spark.storage.ThreadingTest to stress test the BlockManager using 5 block producer and 5 block consumer threads.
2012-11-09 14:09:37 -08:00
Matei Zaharia 6607f546cc Added an option to spread out jobs in the standalone mode. 2012-11-08 23:13:12 -08:00
Matei Zaharia 66cbdee941 Fix for connections not being reused (from Josh Rosen) 2012-11-08 09:53:40 -08:00
Imran Rashid 809b2bb1fe fix bug in getting slave id out of mesos 2012-11-08 00:34:28 -08:00
Matei Zaharia bb1bce7924 Various fixes to standalone mode and web UI:
- Don't report a job as finishing multiple times
- Don't show state of workers as LOADING when they're running
- Show start and finish times in web UI
- Sort web UI tables by ID and time by default
2012-11-07 16:49:53 -08:00
Matei Zaharia e2b8477487 Made Akka timeout and message frame size configurable, and upped the defaults 2012-11-06 15:58:05 -08:00
Shivaram Venkataraman a7d967a1ca Remove unnecessary hash-map put in MemoryStore 2012-11-01 10:46:38 -07:00