app is built using the rocker image rocker/verse:3.6.1. This image contains tidyverse, Tex packages and a web-based RStudio application. Based on the Dockerfile, app creates an OS(docker container) where ACMT runs. The 'workspace' folder contains files needed for running ACMT.
It is estimated that an hour of set-up time is needed.
rocker/verse:3.6.1 contains the basic OS that we need for running ACMT, but we still need to install the OS software and R packages. For the operating systems, we install libudunits2-dev and libgdal-dev; they are needed for installing R packages geosphere, lwgeom, raster, sf, tidycensus, tigris, units, and USAboundaries.
When we need to make adjustment to the OS, we can run app individually with the following instructions.
In terminal, go to src/app.
Run docker build -t app . to build the docker image for app, following the instructions specified in Dockerfile.
Run docker run --rm -v /Users/wzhou87/Desktop/epiProject/docker-acmt/src/app/workspace:/home/rstudio/workspace -e PASSWORD=acmtPassword -p 8787:8787 app to create a container from the app image we just built.
Open a browser in your local computer and go to localhost:8787. In the log-in window, enter rstudio for username and acmtPassword for password.
The RStudio opened in the browser is now connected from the docker container you just created.
docker-acmt is a project to make ACMT more portable. It creates a docker network for data querying and it runs the ACMT R scripts on a local computer to query data (see https://github.com/smooney27/docker-acmt). The disavantage of this method is that, although the Docker network is isolated from the local computer, the R scripts are not. Now, we have incorporated the running of the R scripts into the Docker network (docker-acmt) as well and created acmt-network, a fully isolated environment for running ACMT. We achieve this by creating a Docker container that runs the R scripts and connects the container to the original docker network.
When the R scripts are run in a Docker container, the original way of querying data: pinging something like http://localhost:5433/ failed. To fix this, we made a small tweak by adding the internal to the URL and changed it to http://host.docker.internal:5433. The host.docker.internal replaces localhost and allow the app container query other containers in docker-acmt as if the app container was run locally like before.
Recording some codes used during development.
curl http://localhost:5433/latlong?q=1410+NE+Campus+Parkway%2c+Seattle%2c+WA+98195
curl http://localhost:8787
curl http://host.docker.internal:5000
curl http://localhost:5000