This closes #5001.

This was causing the snapshot deployment to fail.

Update to the latest checkstyle version and fix the errors not
previously detected.

This closes #5044.

Before, the "hadoop2" profile would create a binary release for whatever
happens to be the default hadoop.version.

This closes #4981.

This uses traps to ensure that we properly do cleanups, remove config
values and shutdown things.

- do not shade everything, especially not JDK classes!
-> instead define include patterns explicitly
- do not shade core Flink classes (only those imported from flink-hadoop-fs)
- hack around Hadoop loading (unshaded/non-relocated) classes based on names in
  the core-default.xml by overwriting the Configuration class (we may need to
  extend this for the mapred-default.xml and hdfs-defaults.xml):
-> provide a core-default-shaded.xml file with shaded class names and copy and
  adapt the Configuration class of the respective Hadoop version to load this
  file instead of core-default.xml.

Add checkstyle suppression pattern for the Hadoop Configuration classes

Also fix the (integration) tests not working because they tried to load the
relocated classes which are apparently not available there

Remove minimizeJar from shading of flink-s3-fs-presto because this was
causing "java.lang.ClassNotFoundException:
org.apache.flink.fs.s3presto.shaded.org.apache.commons.logging.impl.LogFactoryImpl"
since these classes are not statically imported and thus removed when
minimizing.

Fix s3-fs-presto not shading org.HdrHistogram

Fix log4j being relocated in the S3 fs implementations

Add shading checks to travis

This closes #4605.

This closes #4927

This closes #3748.

The QS module is split into core and client. The core should
be put in the lib folder to enable queryable state, while the
client is the one that the user will program against. The
reason for the restructuring in mainly to remove the dependency
on the flink-runtime from the user's program.

Prior to this commit, there is no flushing of KPL outstanding records on
checkpoints in the FlinkKinesisProducer. Likewise to the at-least-once
issue on the Flink Kafka producer before, this may lead to data loss if
there are asynchronous failing records after a checkpoint which the
records was part of was completed.

This closes #4871.

This closes #4886.

This closes #4867.

This adds two implementations of a file system that write to S3.
Both are not actual re-implementations but wrap other implementations and shade dependencies.

(1) A wrapper around Hadoop's s3a file system. By pulling a smaller dependency tree and
    shading all dependencies away, this keeps the appearance of Flink being Hadoop-free,
    from a dependency perspective.

(2) The second S3 file system is from the Presto Project.
    Initial simple tests seem to indicate that it responds slightly faster
    and in a bit more lightweight manner to write/read/list requests, compared
    to the Hadoop s3a FS, but it has some semantical differences.

This closes #4818

We deliberately ignore redundant modifiers for now since we want `final`
modifiers on `final` classes for increased future-proofness.

This changes the discovery mechanism of file from static class name configurations
to a service mechanism (META-INF/services).

As part of that, it factors HDFS and MapR FS implementations into separate modules.

With this change, users can add new filesystem implementations and make them available
by simply adding them to the class path.

The commits after this one move consolidated tests to this file which
breaks the 3000 lines limit.

This also adds an end-to-end test that verifies correct order for both
classes and resources.

This fixes the netty check and makes all of them more robust against failures of
the executed commands counting the number of dependencies.

This closes #4653.

This closes #4503.

This closes #4505.

This closes #4494.

This brings it in line with the monolithic create_release_files.sh.

This closes #4452.