James-Crean/cdatweb

Web visualization framework of UV-CDAT

================= CDATWeb 2.0 alpha

##Please read the Devlopment Guidelines

This is an early version of CDAT Web 2.0 with the basic organizational structure in place. Work is ongoing to restore features and document the development process. For the moment, the application relies on starting two web servers:

1. A standard django test server with no difficult to install dependencies.

2. A visualization server running vtkWeb. This server requires a UV-CDAT installation with vtkWeb enabled. You can check if vktWeb is enabled by sourcing your setup_runtime.sh file and running the following:

   python -c "import vtk.web"

   If that gives you an import error, then you need to rebuild VTK or enable the ParaView option for UV-CDAT. It is also possible to compile without ParaView but to enable vtkWeb support. For now this requires patching the UV-CDAT source.

Django installation

Installing the django components is relatively simple. It is recommended you aren't in your UV-CDAT environment for this section. It is also recommended that you run inside a virtual environment as follows:

pip install virtualenv
virtualenv django
source django/bin/activate

You can leave the virtual environment at any time by running deactivate.

To install the dependencies, just run:

pip install -r requirements.txt -r requirements-dev.txt 

Now install the application into your virtualenv, create the database, and start up test server:

python setup.py develop
manage.py syncdb --noinput
manage.py runserver

The front end is now browseable at http://localhost:8000/.

Visualization installation

Refer to one of the items below for starting up the visualization server. Once the server is running, browse to http://localhost:8000/vtk/viewer.html. In case of trouble, see the server logs in tmp/logs. By default, CDATWeb is configured to use a development visualization server at LLNL. In order to use a local visualization server, you will need to modify your django configuration at cdatweb/settings/development.py. The value of VISUALIZATION_LAUNCHER should be set to your launcher service endpoint. By default this is

VISUALIZATION_LAUNCHER = 'http://localhost:7000/vtk'

1. Running vtkWeb from a docker image (Mac OS or Linux ony)

You first must have Docker installed. See the installation guide for more information. On OSX, the setup here assumes you have docker-machine on you PATH and that you have a docker machine named docker (see the docker setup instructions below).

First, get the latest docker image from dockerhub by running

docker pull uvcdat/cdatweb-vtkweb

Now go to the vis_server directory and install the requirements for the vtkweb launcher

pip install -r requirements.txt

You should now be able to run the launcher

python launcher.py config-docker.json

This will serve the launcher at port 7000 on your machine.

At this point, you can start up the django server from a different terminal window as described above. If everything is working correctly, you should see a list of files when you browse to http://localhost:8000/vtk/viewer.html.

2. Running a local vtkWeb instance

As noted above, you may be able use a prebuilt UV-CDAT, but if you run into troubles, try to rebuild the latest version with the following option: cmake -D -DCDAT_BUILD_WEB=ON /path/to/uvcdat. Make sure you are not in your django virtual environment and source your UV-CDAT setup script. Source uvcdat environment: . install/bin/setup_runtime.sh and make sure netcdf is built with DAP support by running nc-config --all and checking if --has-dap is on. If that is not the case you won't be able to open files from http://test.opendap.org If everything is setup correctly, you should be able to run the following:

cd vis_server
./start.sh

3. Running vtkWeb on a remote machine

The procedure for serving vtkWeb from a remote machine (or a cluster) is similar to running locally; however, you must edit the file vis_server/config.json. The details of making this work are beyond the scope of this document. See ParaViewWeb documentation for more details.

Docker setup on OS X

On Mac OS, docker containers must run inside a light weight virtual machine. There is a command line utility for managing instances of these VM's called docker-machine. Docker-machine replaces and earlier tool called boot2docker and supports provisioning on to VirtualBox, VMWare, as well as many cloud providers. For local development, the following command will provision a docker host in VirtualBox with the machine name of docker.

docker-machine create docker

To use that host in you're current shell, you need to run

eval "$(docker-machine env docker)"

after which you can use docker as normal. For the docker.py script in this repository, it is assumed that the docker host is called "docker".

Building a new docker container

If you make changes to the visualization server component, you will need to rebuild the docker container. This can be done simply by going into the vis_server directory and running the following command:

docker build -t uvcdat/cdatweb-vtkweb .