The ShinyStudio project is an orchestration of various open-source solutions with the goal of providing:
- a secured, collaborative development environment for R, Python, PowerShell, and more.
- a secured, convenient way to share apps and documents written in Shiny, RMarkdown, plain Markdown, or HTML.
- easily reproducible, cross-platform setup leveraging Docker for all components.
There are two distributions of ShinyStudio, the image and the stack, explained below.
The ShinyStudio image, hosted on DockerHub, builds upon the Rocker project to include:
- ShinyProxy
- RStudio Server
- VS Code, modified by Coder.com
- Shiny Server
The image is great for a personal instance, a quick demo, or the building blocks for a very customized setup.
The ShinyStudio stack builds upon the image to incorporate:
Each component of the stack is run in a Docker container for reproducibility, scalability, and security. Only the NGINX port is exposed on the host system; all communication between ShinyProxy and other components happens inside an isolated Docker network.
The setup has been verified to work on each of Docker (for Linux) and Docker Desktop (for Mac and Windows).
Note: when upgrading ShinyStudio, please setup from scratch and migrate existing content/settings afterward.
Note: Setup must be run as a non-root user.
To download and run the ShinyStudio image from
DockerHub, first, create
a docker network named shinystudio-net
:
docker network create shinystudio-net
Then, execute docker run
in the terminal for your OS:
- Bash (Linux/Mac)
docker run -d --restart always --name shinyproxy \
--network shinystudio-net \
-v /var/run/docker.sock:/var/run/docker.sock \
-e USERID=$USERID \
-e USER=$USER \
-e PASSWORD=password \
-e CONTENT_PATH="${HOME}/ShinyStudio" \
-e SITE_NAME=shinystudio \
-p 8080:8080 \
dm3ll3n/shinystudio
- PowerShell (Windows)
docker run -d --restart always --name shinyproxy `
--network shinystudio-net `
-v /var/run/docker.sock:/var/run/docker.sock `
-e USERID=1000 `
-e USER=$env:USERNAME `
-e PASSWORD=password `
-e CONTENT_PATH="/host_mnt/c/Users/$env:USERNAME/ShinyStudio" `
-e SITE_NAME=shinystudio `
-p 8080:8080 `
dm3ll3n/shinystudio
Notice the unique form of the path for the
CONTENT_PATH
variable in the Windows setup.
Once complete, open a web browser and navigate to
http://<hostname>:8080
. Log in with your username and the password
password
.
The stack distribution of ShinyStudio is delivered through the GitHub repo and introduces two additional requirements:
- docker-compose (ships with Docker Desktop)
- Git
HTTPS is configured by default, so SSL/TLS certs are required in order
for the stack to operate. Use the provided script certify.sh
(certify.ps1
for Windows) to create a self-signed certificate, or to
request one from LetsEncrypt (more on that).
# copy the setup files.
git clone https://github.com/dm3ll3n/ShinyStudio
# enter the directory.
cd ShinyStudio
# run certify to generate self-signed cert.
./certify.[sh/ps1]
Now, browse to http://<hostname>
(e.g., http://localhost
) to access
ShinyStudio. On first launch, you will need to accept the warning about
an untrusted certificate. See the customized setup to see how to request
a trusted cert from LetsEncrypt.
The default logins are below. See the customized setup to see how to add/remove accounts.
username | password |
---|---|
user | user |
admin | admin |
superadmin | superadmin |
There are three files essential to a customized configuration:
.env
The docker-compose environment file. The project name, content path, and HTTP ports can be changed here.
Note that Docker volume names are renamed along with the project name, so be prepared to migrate or recreate data stored in Docker volumes when changing the project name.
application.yml
The ShinyProxy config file. Users can be added/removed here. Other configurations are available too, such as the site title and the ability to provide a non-standard landing page.
Using the provided template, you can assign users to the following groups with tiered access:
- readers: can only view content from “Apps & Reports”, “Documents”, and “Personal”.
- admins: can view all site content and develop content with RStudio and VS Code.
- superadmins: can view and develop site content across multiple instances of ShinyStudio. Can also manage all user files.
Review the ShinyProxy configuration documentation for all options.
nginx.conf
The NGINX config file. Defines the accepted site name and what ports to listen on.
If you change the ports here, you must also change the ports defined in
the .env
file. Also, if you change the domain name, you must
provide/generate a new certificate for it.
certify.[sh/ps1]
The script used to generate a self-signed cert, or to request a trusted cert from LetsEncrypt.
With no parameters, certify
generates a self-signed cert for
example.com
(the default domain name defined in nginx.conf
).
To generate a self-signed cert with another domain name, first edit the
domain name in nginx.conf
. Afterward, generate a new cert with:
./certify.sh <domain name>
# e.g., ./certify.sh www.shinystudio.com
If your server is accessible from the web, you can request a trusted
certificate from LetsEncrypt. First, edit nginx.conf
with your domain
name, then request a new cert from LetsEncrypt like so:
./certify.sh <domain name> <email>
# e.g., ./certify.sh www.shinystudio.com [email protected]
CertBot, included in the stack, will automatically renew your LetsEncrypt certificate.
To manage the services in the stack, use the native docker-compose commands, e.g.:
# stop all services.
docker-compose down
# start all services.
docker-compose up -d
Open either RStudio or VS Code and notice two important directories:
- __ShinyStudio__
- __Personal__
Files must be saved in either of these two directories in order to persist between sessions.
These two folders are shared between instances RStudio, VS Code, and Shiny Server. So, creating new content is as simple as saving a file to the appropriate directory.
The ShinyStudio image comes with…
- R
- Python 3
- PowerShell
…and ODBC drivers for:
- SQL Server
- PostgresSQL
- Cloudera Impala.
These are persistent because they are built into the image.
Persistent | |
---|---|
__ShinyStudio__ directory | Yes |
__Personal__ directory | Yes |
Other directories | No |
R Libraries | Yes |
Python Packages | Yes |
PowerShell Modules | Yes |
RStudio User Settings | Yes |
VS Code User Settings | Yes |
Installed Apps | No |
Installed Drivers | No |