Public Repository

Last pushed: a year ago
Short Description
GISpark is a Framework for GIS running on Spark.
Full Description

GISpark is a Framework for GIS running on Spark.
GISpark contain: Ubuntu + Jupyter notebook + Spark + GIScript + TensorFlow + Orange.

Setup

Create a directory: GISpark

mkdir GISpark

Then install anaconda(included).
Create a anconda virtual environment:

conda create -n GISpark anaconda

Run conda install to install packages(setup-gispark.sh)

conda install -n GISpark -y numpy  
conda install -n GISpark -y scipy  
conda install -n GISpark -y blaze  
conda install -n GISpark -y dask  
conda install -n GISpark -y libdynd  
conda install -n GISpark -y ipyparallel  

conda install -n GISpark -y -c https://conda.anaconda.org/jjhelmus tensorflow  
conda install -n GISpark -y -c https://conda.anaconda.org/anaconda-cluster spark  
conda install -n GISpark -y -c https://conda.anaconda.org/menpo opencv3  
conda install -n GISpark -y -c https://conda.anaconda.org/menpo ffmpeg

Setup GIScript for Geospatial process.

  1. Need the most libstdc++.so.6 and libgomp.so.1 in the GISpark\lib-giscript-x86-linux64/bin. already installed in the gispark/gispark images.
  2. Run setup-libpath.sh to set the giscript path for conda environment.
  3. After ./start.sh, goto browser to open giscript/giscript_test.ipynb, then shift+enter to test.

Usage

Get the GISpark image
docker pull gispark/gispark:v1
Run the container:
docker run -i -t --name gispark -p 9999:8888 gispark/gispark:v1 /bin/bash
Start jupyter notebook.
cd /root/GISpark
source activate GISpark
jupyter notebook

or run shell script: ./start.sh

To access jupyter-notebook service, open browser for view: http://<hostip>:9999
Docker running on Linux: http://localhost:9999.
Docker running on Mac OS X:
First get docker-host IP: docker-machine ip ,
then: http://<dockerhost-ip>:9999

Any other question please comment it as following.

Docker Pull Command
Owner
gispark

Comments (1)
gispark
a year ago

Currently, This project in early stage.
Can use Spark through Notebook by jupyter, TensorFlow and PySpark are functional.