1) Java SE Platform (JDK) 1.7:   http://www.oracle.com/technetwork/indexes/downloads/index.html
2) Scala:                        http://www.scala-lang.org/download/install.html
3) Scala Build Tool (sbt):       http://www.scala-sbt.org/0.13/docs/Setup.html

1. Checkout the EDAS sources:

    >> cd <prefix>
    >> git clone https://github.com/nasa-nccs-cds/EDAS.git 
    
2. Configure the EDAS server (please note that the build process copies the edas.properties file to the cache dir, which defaults to ~/.edas/cache.  The EDAS app will access it from there):

    >> cd EDAS/project
    >> cp edas.properties.template edas.properties
    >> emacs edas.properties
    
3. Parameter Descriptions
    
    inputs.methods.allowed:  Comma-separated list of allowed input methods, possible values:  collection file  http
    kernels.visibility:      Kernel visibility level, possible values:  experimental developmental restricted public
    spark.master:            Address of Spark master node, examples:  spark://cldralogin101:7077   local[4]
    num.cluster.nodes:       Number of nodes in your cluster (defaults to 1)
    partition.size           Optimal spark partition size for your cluster, recommended value:  200m
    record.size              Optimal spark record size for your cluster, recommended value:  200m
    parts.per.node           Number of cores to be utilized on each worker node of your cluster   
    wps.shared.data.dir      Denotes the path of a data directory that is shared between the EDAS server and portal.  Should not be included if a shared directory does not exist.
    
4. Optional EDAS environment variable:
    EDAS_CACHE_DIR:             EDAS Cacahe dir (defaults to ~/.edas/cache).
    EDAS_UNMANAGED_JARS:        Directory containing external jars for kernel development
    YARN_CONF_DIR:              Yarn config directory (Yarn usage discouraged- recommend running Spark in standalone mode).

5. Build the application (for a clean build one can execute "sbt clean" before "sbt package"):

    >> cd EDAS
    >> sbt package

6. Run unit tests:

    >> sbt test

7. Source the setup file to configure the runtime environment:

    >> source ~/.edas/sbin/setup_runtime.sh

8. Startup the EDAS server:
 
    Cluster mode:   >> startup_edas_standalone.sh
    Local mode:     >> startup_edas_local.sh

9. Access demos:

    Designed to be deployed with the CDWPS framework (https://github.com/nasa-nccs-cds/CDWPS)

1) Install Anaconda: https://github.com/UV-CDAT/uvcdat/wiki/Install-using-Anaconda

2) Create EDAS conda environment:
    
    >> conda create -n edas -c conda-forge -c cdat cdat
    >> source activate edas
    >> conda install pyzmq psutil lxml requests urllib3 six defusedxml
           
3) Initialize shell enviromnment for edas:

    >> source <prefix>/EDAS/bin/setup_runtime.sh
    
4) Build EDAS python pacakges:

    >> cd EDAS
    >> python setup.py install

1) Install IntelliJ IDEA CE from https://www.jetbrains.com/idea/download/ with Scala plugin enabled.

2) Start IDEA and import the EDAS Project from Version Control (github) using the address https://github.com/nasa-nccs-cds/EDAS.git.

 The EDAS framework utilzes a number of network connections:
    1) A ZeroMQ connection from the web service to the EDAS server.
    2) Connections from the EDAS master node to the EDAS workers, managed by Apache Spark.
    3) A ZeroMQ connection on each worker node between the spark worker and a python slave process.
    
 It is the responsibility of the installer to ensure that these connections are suitably secured.  Information on 
 ZeroMQ authentication and encryption, should it be needed, can be found at: http://www.evilpaul.org/wp/2017/05/02/authentication-encryption-zeromq.
 
 Apache Spark currently supports authentication via a shared secret. Authentication can be configured to be on via the spark.authenticate configuration parameter. 
 This parameter controls whether the Spark communication protocols do authentication using the shared secret. This authentication is a basic handshake to 
 make sure both sides have the same shared secret and are allowed to communicate. If the shared secret is not identical they will not be allowed to communicate.
 The Spark parameter spark.authenticate.secret should be configured on each of the nodes. This secret will be used by all the Master/Workers and applications.
 See:  https://spark.apache.org/docs/latest/security.html