With this change, users can submit a Flink job to a YARN cluster without having a local client monitoring the Application Master or job. You can basically fire and forget a Flink job to YARN.
For supporting this, the ApplicationMaster can now monitor the status of a job and shutdown itself once it is in a terminal state.

The change also verifies that various ways of setting the parallelism on YARN are passed through the system correctly (per job, session).

There was a bug in YARN container creation which made the configuration values for the heap offset useless. This change fixes this error.

All mentioned features and bugs are covered by the flink-yarn-tests.

This closes #542

[FLINK-1801] [FLINK-1465] Network environment can start prior to TaskManager in "disassociated" mode.

NetworkEnvironment allocates heavy network buffer pool on startup and supports
multiple associations / disassociations with the TaskManager actor.

Fix negative memory report by replacing overflowing ints with longs.

This closes #522

[FLINK-1779] [runtime] Rename the function name from getCurrentyActiveConnections to getCurrentActiveConnections in org.apache.flink.runtime.blob

This closes #529

This closes #452

This closes #496

Also clean up some unused imports.

[FLINK-1796] [jobmanager] In local mode, the embedded TaskManager is watched by a process reaper as well.

Package name is now flink-table. ExpressionOperation is renamed to
Table.

This also adds more JavaDoc and ScalDoc.

Adjusts the comments inside the POM file to encourage use ov "mvn clean package",
rather than "mvn clean install".

Also exclude more large transitive dependencies.

Remove the redundant import code in TaskManager

Author: hongsibao <hongsibao@huawei.com>

Closes #538 from matadorhong/RemoveRedundantImport and squashes the following commits:

1f80c0d [hongsibao] Remove the redunant import code

This closes #518

This closes #421

This closes #533

[gelly] corrected community detection usage description and moved it from the library method to the example;
removed unnecessary @SuppressWarnings(serial) from SSSP

This closes #505

Move the hadoop-hdfs and hadoop-common test jars dependency before Tachyon.

Looks like Maven respect the ordering of dependencies and first declarations win.
Seemed like Tachyon brings transitive dependencies of older hadoop versions.

Author: Henry Saputra <hsaputra@apache.org>

Closes #536 from hsaputra/FLINK-1785_add_hadoop_deps_to_test_scope and squashes the following commits:

f266a6c [Henry Saputra] [FLINK-1785] [test] Master tests in flink-tachyon fail with java.lang.NoSuchFieldError: IBM_JAVA.

Closes #534

Closes #523

This commit squashes:

 - [runtime] Rename create read view methods to be consistent
 - [runtime] Fix spillable subpartition view read offset after spilling
 - [tests] Add comment to ScheduleOrUpdateConsumersTest

It was possible that a concurrent release and enqueue operation from the
network I/O thread leave a buffer queued at a released channel, resulting in a
resource leak.

This commit fixes this and adds a test, which successfully provoked the race
condition before the fix.

The visibility of input channel operations is reduced to package private.

Moves the partition consumable notification out of ResultPartition. This makes
it easier to setup unit tests, where no job manager is available to notify.