and thus groupBy/cogroup are broken in Java APIs when Kryo is used).

@pwendell this should be merged into 1.0.1.

Thanks @sorenmacbeth for reporting this & helping out with the fix.

Author: Reynold Xin <rxin@apache.org>

Closes #1206 from rxin/kryo-iterable-2270 and squashes the following commits:

09da0aa [Reynold Xin] Updated the comment.
009bf64 [Reynold Xin] [SPARK-2270] Kryo cannot serialize results returned by asJavaIterable (and thus groupBy/cogroup are broken in Java APIs when Kryo is used).

**SPARK-2258.** Worker UI displays zombie processes if the executor throws an exception before a process is launched. This is because we only inform the Worker of the change if the process is already launched, which in this case it isn't.

**SPARK-2266.** We expose "Some(app-id)" on the log page. This is fairly minor.

Author: Andrew Or <andrewor14@gmail.com>

Closes #1213 from andrewor14/fix-worker-ui and squashes the following commits:

c1223fe [Andrew Or] Fix worker UI bugs

This reverts a change introduced in 38702487, which redirected all stderr to the OS pipe instead of directly to the `bin/pyspark` shell output. This causes a simple job to hang in two ways:

1. If the cluster is not configured correctly or does not have enough resources, the job hangs without producing any output, because the relevant warning messages are masked.
2. If the stderr volume is large, this could lead to a deadlock if we redirect everything to the OS pipe. From the [python docs](https://docs.python.org/2/library/subprocess.html):

```
Note Do not use stdout=PIPE or stderr=PIPE with this function as that can deadlock
based on the child process output volume. Use Popen with the communicate() method
when you need pipes.
```

Note that we cannot remove `stdout=PIPE` in a similar way, because we currently use it to communicate the py4j port. However, it should be fine (as it has been for a long time) because we do not produce a ton of traffic through `stdout`.

That commit was not merged in branch-1.0, so this fix is for master only.

Author: Andrew Or <andrewor14@gmail.com>

Closes #1178 from andrewor14/fix-python and squashes the following commits:

e68e870 [Andrew Or] Merge branch 'master' of github.com:apache/spark into fix-python
20849a8 [Andrew Or] Tone down stdout interference message
a09805b [Andrew Or] Return more than 1 line of error message to user
6dfbd1e [Andrew Or] Don't swallow original exception
0d1861f [Andrew Or] Provide more helpful output if stdout is garbled
21c9d7c [Andrew Or] Do not mask stderr from output

https://issues.apache.org/jira/browse/SPARK-2038

to differentiate with SparkConf object and at the same time keep the source level compatibility

Author: CodingCat <zhunansjtu@gmail.com>

Closes #1137 from CodingCat/SPARK-2038 and squashes the following commits:

11abeba [CodingCat] revise the comments
7ee5712 [CodingCat] to keep the source-compatibility
763975f [CodingCat] style fix
d91288d [CodingCat] rename "conf" parameters in the saveAsHadoop functions

The `BigDecimal` branch in `unwrap` matches to `scala.math.BigDecimal` rather than `java.math.BigDecimal`.

Author: Cheng Lian <lian.cs.zju@gmail.com>

Closes #1199 from liancheng/javaBigDecimal and squashes the following commits:

e9bb481 [Cheng Lian] Should match java.math.BigDecimal when wnrapping Hive output

JIRA issue: [SPARK-2263](https://issues.apache.org/jira/browse/SPARK-2263)

Map objects were not converted to Hive types before inserting into Hive tables.

Author: Cheng Lian <lian.cs.zju@gmail.com>

Closes #1205 from liancheng/spark-2263 and squashes the following commits:

c7a4373 [Cheng Lian] Addressed @concretevitamin's comment
784940b [Cheng Lian] SARPK-2263: support inserting MAP<K, V> to Hive tables

Author: witgo <witgo@qq.com>

Closes #1194 from witgo/SPARK-2248 and squashes the following commits:

6ac950b [witgo] spark.default.parallelism does not apply in local mode

Author: Michael Armbrust <michael@databricks.com>

Closes #1204 from marmbrus/nullPointerToString and squashes the following commits:

35b5fce [Michael Armbrust] Fix possible null pointer in acumulator toString

Author: Matthew Farrellee <matt@redhat.com>

Closes #1185 from mattf/master-1 and squashes the following commits:

42150fc [Matthew Farrellee] Autodetect JAVA_HOME on RPM-based systems

This will be helpful in join operators.

Author: Cheng Hao <hao.cheng@intel.com>

Closes #1187 from chenghao-intel/joinedRow and squashes the following commits:

87c19e3 [Cheng Hao] Add base row set methods for JoinedRow

This is an alternative solution to #1124 . Before launching the executor backend, we first fetch driver's spark properties and use it to overwrite executor's spark properties. This should be better than #1124.

@pwendell Are there spark properties that might be different on the driver and on the executors?

Author: Xiangrui Meng <meng@databricks.com>

Closes #1132 from mengxr/akka-bootstrap and squashes the following commits:

77ff32d [Xiangrui Meng] organize imports
68e1dfb [Xiangrui Meng] use timeout from AkkaUtils; remove props from RegisteredExecutor
46d332d [Xiangrui Meng] fix a test
7947c18 [Xiangrui Meng] increase slack size for akka
4ab696a [Xiangrui Meng] bootstrap to retrieve driver spark conf

Author: Michael Armbrust <michael@databricks.com>

Closes #1201 from marmbrus/fixCacheTests and squashes the following commits:

9d87ed1 [Michael Armbrust] Use analyzer (which runs to fixed point) instead of manually removing analysis operators.

The assertJsonStringEquals method was missing an "assert" so
did not actually check that the strings were equal. This commit
adds the missing assert and fixes subsequently revealed problems
with the JsonProtocolSuite.

@andrewor14 I changed some of the test functionality to match what it
looks like you intended based on the expected strings -- let me know if
anything here looks wrong.

Author: Kay Ousterhout <kayousterhout@gmail.com>

Closes #1198 from kayousterhout/json_test_fix and squashes the following commits:

77f858f [Kay Ousterhout] Fix broken Json tests.

Don't check executor/host availability when creating a TaskSetManager. Because the executors may haven't been registered when the TaskSetManager is created, in which case all tasks will be considered "has no preferred locations", and thus losing data locality in later scheduling.

Author: Rui Li <rui.li@intel.com>
Author: lirui-intel <rui.li@intel.com>

Closes #892 from lirui-intel/delaySchedule and squashes the following commits:

8444d7c [Rui Li] fix code style
fafd57f [Rui Li] keep locality constraints within the valid levels
18f9e05 [Rui Li] restrict allowed locality
5b3fb2f [Rui Li] refine UT
99f843e [Rui Li] add unit test and fix bug
fff4123 [Rui Li] fix computing valid locality levels
685ed3d [Rui Li] remove delay shedule for pendingTasksWithNoPrefs
7b0177a [Rui Li] remove redundant code
c7b93b5 [Rui Li] revise patch
3d7da02 [lirui-intel] Update TaskSchedulerImpl.scala
cab4c71 [Rui Li] revised patch
539a578 [Rui Li] fix code style
cf0d6ac [Rui Li] fix code style
3dfae86 [Rui Li] re-compute pending tasks when new host is added
a225ac2 [Rui Li] SPARK-1937: fix issue with task locality

Found out about this from the Hacker News link to GraphX which was using HTTPs.

@mengxr

Author: Reynold Xin <rxin@apache.org>

Closes #1189 from rxin/mllib-doc and squashes the following commits:

5328be0 [Reynold Xin] [SPARK-2252] Fix MathJax for HTTPs.

This PR is a sub-task of SPARK-2044 to move the execution of aggregation into shuffle implementations.

I leave `CoGoupedRDD` and `SubtractedRDD` unchanged because they have their implementations of aggregation. I'm not sure is it suitable to change these two RDDs.

Also I do not move sort related code of `OrderedRDDFunctions` into shuffle, this will be solved in another sub-task.

Author: jerryshao <saisai.shao@intel.com>

Closes #1064 from jerryshao/SPARK-2124 and squashes the following commits:

4a05a40 [jerryshao] Modify according to comments
1f7dcc8 [jerryshao] Style changes
50a2fd6 [jerryshao] Fix test suite issue after moving aggregator to Shuffle reader and writer
1a96190 [jerryshao] Code modification related to the ShuffledRDD
308f635 [jerryshao] initial works of move combiner to ShuffleManager's reader and writer

Note that nothing gets printed to the console because we don't properly maintain session state right now.

I will have a followup PR that fixes it.

Author: Reynold Xin <rxin@apache.org>

Closes #1167 from rxin/commands and squashes the following commits:

56f04f8 [Reynold Xin] [SPARK-2227] Support dfs command in SQL.

Cleanup on Connection, ConnectionManagerId, and ConnectionManager classes part 2 while I was working at the code there to help IDE:
1. Remove unused imports
2. Remove parentheses in method calls that do not have side affect.
3. Add parentheses in method calls that do have side effect or not simple get to object properties.
4. Change if-else check (via isInstanceOf) for Connection class type with Scala expression for consistency and cleanliness.
5. Remove semicolon
6. Remove extra spaces.
7. Remove redundant return for consistency

Author: Henry Saputra <henry.saputra@gmail.com>

Closes #1157 from hsaputra/cleanup_connection_classes_part2 and squashes the following commits:

4be6906 [Henry Saputra] Fix Spark Scala style for line over 100 chars.
85b24f7 [Henry Saputra] Cleanup on Connection and ConnectionManager classes part 2 while I was working at the code there to help IDE: 1. Remove unused imports 2. Remove parentheses in method calls that do not have side affect. 3. Add parentheses in method calls that do have side effect. 4. Change if-else check (via isInstanceOf) for Connection class type with Scala expression for consitency and cleanliness. 5. Remove semicolon 6. Remove extra spaces.

Two improvements to the history server:

- Separate the HTTP handling from history fetching, so that it's easy to add
  new backends later (thinking about SPARK-1537 in the long run)

- Avoid loading all UIs in memory. Do lazy loading instead, keeping a few in
  memory for faster access. This allows the app limit to go away, since holding
  just the listing in memory shouldn't be too expensive unless the user has millions
  of completed apps in the history (at which point I'd expect other issues to arise
  aside from history server memory usage, such as FileSystem.listStatus()
  starting to become ridiculously expensive).

I also fixed a few minor things along the way which aren't really worth mentioning.
I also removed the app's log path from the UI since that information may not even
exist depending on which backend is used (even though there is only one now).

Author: Marcelo Vanzin <vanzin@cloudera.com>

Closes #718 from vanzin/hist-server and squashes the following commits:

53620c9 [Marcelo Vanzin] Add mima exclude, fix scaladoc wording.
c21f8d8 [Marcelo Vanzin] Feedback: formatting, docs.
dd8cc4b [Marcelo Vanzin] Standardize on using spark.history.* configuration.
4da3a52 [Marcelo Vanzin] Remove UI from ApplicationHistoryInfo.
2a7f68d [Marcelo Vanzin] Address review feedback.
4e72c77 [Marcelo Vanzin] Remove comment about ordering.
249bcea [Marcelo Vanzin] Remove offset / count from provider interface.
ca5d320 [Marcelo Vanzin] Remove code that deals with unfinished apps.
6e2432f [Marcelo Vanzin] Second round of feedback.
b2c570a [Marcelo Vanzin] Make class package-private.
4406f61 [Marcelo Vanzin] Cosmetic change to listing header.
e852149 [Marcelo Vanzin] Initialize new app array to expected size.
e8026f4 [Marcelo Vanzin] Review feedback.
49d2fd3 [Marcelo Vanzin] Fix a comment.
91e96ca [Marcelo Vanzin] Fix scalastyle issues.
6fbe0d8 [Marcelo Vanzin] Better handle failures when loading app info.
eee2f5a [Marcelo Vanzin] Ensure server.stop() is called when shutting down.
bda2fa1 [Marcelo Vanzin] Rudimentary paging support for the history UI.
b284478 [Marcelo Vanzin] Separate history server from history backend.

Author: Prashant Sharma <prashant.s@imaginea.com>

Closes #1068 from ScrapCodes/SPARK-2118/tools-jar-check and squashes the following commits:

29e768b [Prashant Sharma] Code Review
5cb6f7d [Prashant Sharma] [SPARK-2118] spark class should complaing if tools jar is missing.

JIRA issue: [SPARK-1669](https://issues.apache.org/jira/browse/SPARK-1669)

Caching the same table multiple times should end up with only 1 in-memory columnar representation of this table.

Before:

```
scala> loadTestTable("src")
...
scala> cacheTable("src")
...
scala> cacheTable("src")
...
scala> table("src")
...
== Query Plan ==
InMemoryColumnarTableScan [key#2,value#3], (InMemoryRelation [key#2,value#3], false, (InMemoryColumnarTableScan [key#2,value#3], (InMemoryRelation [key#2,value#3], false, (HiveTableScan [key#2,value#3], (MetastoreRelation default, src, None), None))))
```

After:

```
scala> loadTestTable("src")
...
scala> cacheTable("src")
...
scala> cacheTable("src")
...
scala> table("src")
...
== Query Plan ==
InMemoryColumnarTableScan [key#2,value#3], (InMemoryRelation [key#2,value#3], false, (HiveTableScan [key#2,value#3], (MetastoreRelation default, src, None), None))
```

Author: Cheng Lian <lian.cs.zju@gmail.com>

Closes #1183 from liancheng/spark-1669 and squashes the following commits:

68f8a20 [Cheng Lian] Removed an unused import
51bae90 [Cheng Lian] Made cacheTable idempotent

When mvn is not detected (not in executor's path), 'set -e' causes the
detection to terminate the script before the helpful error message can
be displayed.

Author: Matthew Farrellee <matt@redhat.com>

Closes #1181 from mattf/master-0 and squashes the following commits:

506549f [Matthew Farrellee] Fix mvn detection

The backslash is needed for multiline command

Author: Vlad <frolvlad@gmail.com>

Closes #1158 from frol/patch-1 and squashes the following commits:

e258044 [Vlad] Fixed small running on YARN docs typo

Recent changes ignored the fact that path may be defined with "local:"
URIs, which means they need to be explicitly added to the classpath
everywhere a remote process is started. This change fixes that by:

- Using the correct methods to add paths to the classpath
- Creating SparkConf settings for the Spark jar itself and for the
  user's jar
- Propagating those two settings to the remote processes where needed

This ensures that both in client and in cluster mode, the driver has
the necessary info to build the executor's classpath and have things
still work when they contain "local:" references.

The change also fixes some confusion in ClientBase about whether
to use SparkConf or system properties to propagate config options to
the driver and executors, by standardizing on using data held by
SparkConf.

On the cleanup front, I removed the hacky way that log4j configuration
was being propagated to handle the "local:" case. It's much more cleanly
(and generically) handled by using spark-submit arguments (--files to
upload a config file, or setting spark.executor.extraJavaOptions to pass
JVM arguments and use a local file).

Author: Marcelo Vanzin <vanzin@cloudera.com>

Closes #560 from vanzin/yarn-local-2 and squashes the following commits:

4e7f066 [Marcelo Vanzin] Correctly propagate SPARK_JAVA_OPTS to driver/executor.
6a454ea [Marcelo Vanzin] Use constants for PWD in test.
6dd5943 [Marcelo Vanzin] Fix propagation of config options to driver / executor.
b2e377f [Marcelo Vanzin] Review feedback.
93c3f85 [Marcelo Vanzin] Fix ClassCastException in test.
e5c682d [Marcelo Vanzin] Fix cluster mode, restore SPARK_LOG4J_CONF.
1dfbb40 [Marcelo Vanzin] Add documentation for spark.yarn.jar.
bbdce05 [Marcelo Vanzin] [SPARK-1395] Fix "local:" URI support in Yarn mode (again).

Will list the EC2 instances before detroying the cluster.
This was added because it can be scary to destroy EC2
instances without knowing which one will be impacted.

Author: Jean-Martin Archer <jeanmartin.archer@pulseenergy.com>

This patch had conflicts when merged, resolved by
Committer: Patrick Wendell <pwendell@gmail.com>

Closes #270 from j-martin/master and squashes the following commits:

826455f [Jean-Martin Archer] [SPARK-2611] Implementing recommendations
27b0a36 [Jean-Martin Archer] Listing of instances to be terminated before the prompt Will list the EC2 instances before detroying the cluster. This was added because it can be scary to destroy EC2 instances without knowing which one will be impacted.

To preserve quoted command line args (in case options have space in them).

Author: Ori Kremer <ori.kremer@gmail.com>

Closes #1169 from orikremer/quote_cmd_line_args and squashes the following commits:

67e2aa1 [Ori Kremer] quote command line args

Author: witgo <witgo@qq.com>

Closes #1174 from witgo/SPARK-2229 and squashes the following commits:

f85f321 [witgo] FileAppender throw anIllegalArgumentException in JDK6
e1a8da8 [witgo] SizeBasedRollingPolicy throw an java.lang.IllegalArgumentException in JDK6

Commons IO is actually barely used, and is not a declared dependency. This just replaces with equivalents from the JDK and Guava.

Author: Sean Owen <sowen@cloudera.com>

Closes #1173 from srowen/SPARK-1316 and squashes the following commits:

2eb53db [Sean Owen] Reorder Guava import
8fde404 [Sean Owen] Remove use of Commons IO, which is not actually a dependency

Tobias noted today on the mailing list:

========

I am trying to use Spark Streaming with Kafka, which works like a
charm – except for shutdown. When I run my program with "sbt
run-main", sbt will never exit, because there are two non-daemon
threads left that don't die.
I created a minimal example at
<https://gist.github.com/tgpfeiffer/b1e765064e983449c6b6#file-kafkadoesntshutdown-scala>.
It starts a StreamingContext and does nothing more than connecting to
a Kafka server and printing what it receives. Using the `future
Unknown macro: { ... }
` construct, I shut down the StreamingContext after some seconds and
then print the difference between the threads at start time and at end
time. The output can be found at
<https://gist.github.com/tgpfeiffer/b1e765064e983449c6b6#file-output1>.
There are a number of threads remaining that will prevent sbt from
exiting.
When I replace `KafkaUtils.createStream(...)` with a call that does
exactly the same, except that it calls `consumerConnector.shutdown()`
in `KafkaReceiver.onStop()` (which it should, IMO), the output is as
shown at <https://gist.github.com/tgpfeiffer/b1e765064e983449c6b6#file-output2>.
Does anyone have any idea what is going on here and why the program
doesn't shut down properly? The behavior is the same with both kafka
0.8.0 and 0.8.1.1, by the way.

========

Something similar was noted last year:

http://mail-archives.apache.org/mod_mbox/spark-dev/201309.mbox/%3C1380220041.2428.YahooMailNeo@web160804.mail.bf1.yahoo.com%3E

KafkaInputDStream doesn't close `ConsumerConnector` in `onStop()`, and does not close the `Executor` it creates. The latter leaves non-daemon threads and can prevent the JVM from shutting down even if streaming is closed properly.

Author: Sean Owen <sowen@cloudera.com>

Closes #980 from srowen/SPARK-2034 and squashes the following commits:

9f31a8d [Sean Owen] Restore ClassTag to private class because MIMA flags it; is the shadowing intended?
2d579a8 [Sean Owen] Close ConsumerConnector in onStop; shutdown() the local Executor that is created so that its threads stop when done; close the Zookeeper client even on exception; fix a few typos; log exceptions that otherwise vanish

...rsion

Author: Patrick Wendell <pwendell@gmail.com>

Closes #1175 from pwendell/test-hadoop-version and squashes the following commits:

9210ef4 [Patrick Wendell] SPARK-2231: dev/run-tests should include YARN and use a recent Hadoop version

Just following up Matei's suggestion to remove the Akka repo references. Builds and the audit-release script appear OK.

Author: Sean Owen <sowen@cloudera.com>

Closes #1170 from srowen/SPARK-1996 and squashes the following commits:

5ca2930 [Sean Owen] Remove outdated Akka repository references

The single file was getting very long (500+ loc).

Author: Reynold Xin <rxin@apache.org>

Closes #1166 from rxin/hiveOperators and squashes the following commits:

5b43068 [Reynold Xin] [SQL] Break hiveOperators.scala into multiple files.

This makes it easier to use config options in operators.

Author: Reynold Xin <rxin@apache.org>

Closes #1164 from rxin/sqlcontext and squashes the following commits:

797b2fd [Reynold Xin] Pass SQLContext instead of SparkContext into physical operators.

- JavaAPISuite was trying to compare a bare path with a URI. Fix by
  extracting the path from the URI, since we know it should be a
  local path anyway/

- b9be1609 excluded the ASM dependency everywhere, but easymock needs
  it (because cglib needs it). So re-add the dependency, with test
  scope this time.

The second one above actually uncovered a weird situation: the maven
test target works, even though I can't find the class sbt complains
about in its classpath. sbt complains with:

  [error] Uncaught exception when running org.apache.spark.util
  .random.RandomSamplerSuite: java.lang.NoClassDefFoundError:
  org/objectweb/asm/Type

To avoid more weirdness caused by that, I explicitly added the asm
dependency to both maven and sbt (for tests only), and verified
the classes don't end up in the final assembly.

Author: Marcelo Vanzin <vanzin@cloudera.com>

Closes #917 from vanzin/flaky-tests and squashes the following commits:

d022320 [Marcelo Vanzin] Fix some tests.

The jira for the issue can be found at: https://issues.apache.org/jira/browse/SPARK-2061
Most of spark has used over to consistently using `partitions` instead of `splits`. We should do likewise and add a `partitions` method to JavaRDDLike and have `splits` just call that. We should also go through all cases where other API's (e.g. Python) call `splits` and we should change those to use the newer API.

Author: Anant <anant.asty@gmail.com>

Closes #1062 from anantasty/SPARK-2061 and squashes the following commits:

b83ce6b [Anant] Fixed syntax issue
21f9210 [Anant] Fixed version number in deprecation string
9315b76 [Anant] made related changes to use partitions in python api
8c62dd1 [Anant] Made splits deprecated in JavaRDDLike