This closes #4278.

This closes #4096.

This closes #3720.

This closes #4064.

This closes #4067.

This closes #4066.

This closes #4248.

Add dependencies for batch and streaming WordCount programs and copies
the jar files into a new target/programs directory. The integration
tests now directly references the program jar files rather than the
prior brittle search.

This removes the flink-yarn-tests build-time dependency on the examples
modules (there remains a build-time dependency on flink-dist).

This closes #4264

This closes #4219

[FLINK-7117] [ha] Add HighAvailabilityServices.getJobManagerLeaderRetriever(jobId, defaultJobManagerAddress)

The getJobManagerLeaderRetriever method with the defaultJobManagerAddress parameter is necessary for
the standalone HA services case. The TaskExecutor don't know at creation time of the HA services,
where the JobManager will be executed. Therefore, we need this way to pass in the JobManager
address at runtime.

This closes #4276.

The RpcService should expose its port it is bound to. That way it is easier to connect to a
remote RpcService.

This closes #4275.

[FLINK-7109] [hadoop] Remove GlobalConfiguration.loadConfiguration from HadoopUtils#getHadoopConfiguration

The HadoopUtils#getHadoopConfiguration should not load the global configuration. Instead
we pass it in as parameter.

This closes #4265.

This closes #4202.

This closes #4206.

This closes #4173.

This closes #4232.

This closes #4229.

This closes #4244.

Add implementations of Byte/Char/Double/Float/ShortValueArray. Along
with the existing implementations of Int/Long/Null/StringValueArray this
covers all 10 CopyableValue types.

This closes #4203

Gelly algorithms commonly have a Result class extending a Tuple type and
implementing one of the Unary/Binary/TertiaryResult interfaces.

Add a Unary/Binary/TertiaryResultBase class implementing each interface
and convert the Result classes to POJOs extending the base result
classes.

This closes #4201

This closes #4097

This closes #4230

This closes #4138

The taskmanager.sh script now supports to start a TaskExecutor by providing flip6 as
a second argument to the script.

This closes #4253.

The entry point can be used by the standalone mode to run a TaskManager. Moreover, the
YarnTaskExecutorRunner now reuses some of the start up logic of the TaskManagerRunner.

This closes #4252.

The TaskManager sends the SlotReport as part of the heartbeat payload to the ResourceManager.
That way, the ResourceManager can sync its internal view on the slot allocation with the actual
allocation state.

This closes #4251.

This closes #4242

This closes #4153

This closes #4068.

This closes  #4221.

This closes #3583.

- [docs] update some config options to the new, non-deprecated ones
- [docs] improvements in the BlobService docs

- use Preconditions.checkArgument in BlobClient
- refactor BlobCache#getURL()
- do not fail the BlobServer if delete fails
- fix concurrent job directory creation
- do not guard a delete() call with a check for existence
- cleanup javadocs in (Blob)LibraryCacheManager
- cleanup tests for BlobLibraryCacheManager

This closes #4146.

This closes #4065.

The switch from the container-based to sudo-enabled environment in
TravisCI has increased available memory from 4 GB to 7.5 GB so use a
forkCount of 2 in all packages including flink-test.

This closes #4222.

This closes #4063.

This closes #4095.