Running PyCBC under Docker

The easiest way to start using PyCBC is to install one of our Docker containers. First, install the Docker Community Edition for your Mac or Windows desktop. Docker CE installations for Linux platforms are also available.

To start a Docker container with no graphics, type the commands:

docker pull pycbc/pycbc-el7:latest
docker run -it pycbc/pycbc-el7:latest /bin/bash -l

This example downloads current version of the code from the GitHub master branch. Replace the string latest with one of the PyCBC release tags (e.g. v1.7.0) to install a container containing a released version of PyCBC. The container includes all of the required software and dependencies to run PyCBC, including a compatible version of LALSuite.

Graphics on Mac OS X

To start a Docker container on Mac OS X that can display graphics on Mac OS X, you must be running XQuartz version 2.7.11 or higher. To start an X11 server, run the command:

open -a Xquartz

In the Xquartz preferences, make sure that Allow connections from network clients is checked, as shown below:

_images/xquartz_pref.png

Next, enable Docker to connect to your X11 server by running the commands:

ip=$(ifconfig en0 | grep inet | awk '$1=="inet" {print $2}')
xhost + $ip

Finally, start the Docker container with:

docker run -e DISPLAY=${ip}:0 -it pycbc/pycbc-el7:latest /bin/bash -l

Graphics on Linux

To start a Docker container on Linux that can display graphics, run the commands:

docker run -e DISPLAY=${DISPLAY} -v /tmp/.X11-unix:/tmp/.X11-unix/ -it pycbc/pycbc-el7:latest /bin/bash -l

If you are having problems with this method, try Using ssh to display graphics.

Sharing user files and SSH keys

It can be useful to share your SSH public/private key with the Docker container, for example to allow you to git push and pull from your repository on GitHub. To do this, add the argument -v ${HOME}/.ssh:/home/pycbc/.ssh to the docker run commands. You can also create e.g. a scratch directory and use the -v option to mount it in the container. This directory can be used to transfer files between the container and the host computer. The command below will start a Docker container with graphics on a Mac and share ssh keys and a scratch directory:

mkdir -p scratch
docker run -e DISPLAY=$(ifconfig en0 | grep inet | awk '$1=="inet" {print $2}'):0 -v `pwd`/scratch:/opt/pycbc/scratch -v ${HOME}/.ssh:/opt/pycbc/.ssh -it pycbc/pycbc-el7:latest /bin/bash -l

Using SSH to display graphics

If you do not have a recent version of Xquartz, or you are encountering other X11 issues, another way to run a docker container with graphics is to start a container that runs an SSH daemon and connect to that. This is also useful for trabsfering files to and from the Docker container using scp from your host computer.

If do not already have a personal ssh public/private key, first create one with the command:

ssh-keygen -t rsa
cat ~/.ssh/id_rsa.pub >> ~/.ssh/authorized_keys
chmod 600 ~/.ssh/authorized_keys

Now start the docker container with:

docker run --name pycbc -d -P -v ${HOME}/.ssh:/opt/pycbc/.ssh -t pycbc/pycbc-el7 tail -f /dev/null i
docker exec -u root -it pycbc /usr/bin/pycbc-sshd

And then you can connect to it with the command:

ssh -Y pycbc@127.0.0.1 -p `docker port pycbc 22 | awk -F: '{print $NF}'`

Using jupyter-notebook within docker

One can start a jupyter notebook within docker and then port forward to your computer environment.:

docker run -it -p 8888:8888 --name pycbc_test pycbc/pycbc-el7:latest /bin/bash -l

Within the docker environment:

jupyter notebook --no-browser --ip 0.0.0.0