Skip to content

Latest commit

 

History

History
111 lines (58 loc) · 3.78 KB

Docker.md

File metadata and controls

111 lines (58 loc) · 3.78 KB

To use Docker

Note: In the instructions below,

  • $ refers to the native bash shell prompt on your computer,
  • root@...# refers to the bash shell prompt within the Docker container, once you have that running.

First install Docker and start the Docker daemon.

Then download the Dockerfile from this directory, either as a single file via:

$ wget https://raw.githubusercontent.com/clawpack/geoclaw_tutorial_tulane2017/master/Dockerfile

or by cloning this repository:

$ git clone https://github.com/clawpack/geoclaw_tutorial_tulane2017

Then go into the directory containing the Dockerfile and build a Docker image:

$ docker build -t geoclaw_tutorial_dockerimage -f Dockerfile .

Don't forget the last . on this line!

Then do:

$ docker run -i -t -p 8889:8889 --name geoclaw_tutorial_container geoclaw_tutorial_dockerimage

This starts a virtual machine (container) named geoclaw_tutorial_container and gives a Unix prompt like:

root@...#

You can go into the geoclaw_tutorial_tulane2017/geoclaw_examples/chile2010a directory, for example, and run the code from the command line, e.g. via

root@...# cd geoclaw_tutorial_tulane2017/geoclaw_examples/chile2010a
root@...# make topo
root@...# make .plots

See below for how to view the resulting plots.

Notebook server:

If you want to work with the notebooks, start the notebook server via

root@...# jupyter notebook --notebook-dir=/geoclaw_tutorial_tulane2017 --ip='*' --port=8889 --no-browser

Then open a browser (on your laptop) to the URL printed out when the Jupyter server starts via the command above. This might be of the form:

http://localhost:8889/tree

or perhaps will include a token, something like:

http://localhost:8889/?token=9542fb1ed940f873f28e6a371c6334c5b1a0d8656121905c

This should open a web page with the list of directories. Navigate to notebooks/Index.ipynb to get started.

Use ctrl-C to exit the Jupyter notebook server.

See http://jupyter.org/ for more documentation on Jupyter.

Connecting with a second bash shell

If you have the notebook server running and also want another window open with a bash shell (to run GeoClaw from the command line), in another shell on your laptop you can do:

$ docker exec -it geoclaw_tutorial_container bash

Exiting a shell / halting a container

Use Ctrl-p Ctrl-q to exit from a shell without halting the docker container.

You can exit the container (after using ctrl-C to quit the jupyter server if one is running) via::

root...# exit

Restarting a container

You can restart the container via::

$ docker start -a -i geoclaw_tutorial_container

The external port should still work for serving notebooks.

Viewing plots

If you run GeoClaw from the command line and want to view the resulting plots (generated e.g. via make .plots), you can point your browser to, e.g.:

http://localhost:8889/files/geoclaw_examples/chile2010a/_plots/_PlotIndex.html

This assumes you have the Jupyter notebook server running (from a different shell).

Removing a container

This gets rid of the container and any data that you might have created when running this container:

$ docker rm geoclaw_tutorial_container

Removing the image

You might want to do this if you need to rebuild an image with an updated Dockerfile or to incorporate a newer version of software:

$ docker rmi geoclaw_tutorial_dockerimage

More resources: