Awesome

analogsea

Use our link https://m.do.co/c/1d5a471e5f54, you'll get 100 usd in credits to try DO!

analogsea is an R client for version 2 of the Digital Ocean API. See ?droplet_functions after loading analogsea. It allows you to programatically create and destroy droplets (remote computers), and install various R related tools:

R (done)
RStudio Server (done)
RStudio Shiny Server (done)
OpenCPU (not yet)
Use packrat to move a project to a droplet (not yet)

In addition, it allows you to use a readily available image with RStudio Server, Shiny Server and fully tidyverse from DigitalOcean Marketplace.

Docs: https://pacha.dev/analogsea/.

Use cases

Andrew Heiss: Create a cheap, disposable supercomputer with R, DigitalOcean, and future

Install

Stable version from CRAN

install.packages("analogsea")

Development version from GitHub

remotes::install_github("pachadotdev/analogsea")

library("analogsea")

Create a DO account

If you don't already have one, create a DO account. By using this link, you'll start with $100 in credits with a 2 months limit. This is enough for 1440 hours of computing on a machine with 4 GB memory and 2 dedicated CPUs.

Authenticate

The best way to authenticate is to generate a personal access token (https://cloud.digitalocean.com/settings/tokens/new) and save it in an environment variable called DO_PAT. If you don't do this, you'll be prompted to authenticate in your browser the first time you use analogsea.

Make sure you provide DigitalOcean your public key at https://cloud.digitalocean.com/ssh_keys. Github has some good advice on creating a new public key if you don't already have one: https://help.github.com/articles/generating-ssh-keys/.

It is highly recommended to use DO_PAT, it's easier and allows (semi) automated tasks.

To use ssh keys install the recommended ssh package: r install.packages("ssh").

Get droplets

All droplets

droplets()

<r> droplets()
$unintrenchable
<droplet>unintrenchable (2724525)
  Status: off
  Region: San Francisco 1
  Image: Ubuntu 14.04 x64
  Size: 512mb ($0.00744 / hr)

$basipterygium
<droplet>basipterygium (2724526)
  Status: active
  Region: San Francisco 1
  Image: Ubuntu 14.04 x64
  Size: 512mb ($0.00744 / hr)

A single droplet. Pass in a single droplet id.

droplet(2724525)

<droplet>unintrenchable (2724525)
  Status: off
  Region: San Francisco 1
  Image: Ubuntu 14.04 x64
  Size: 512mb ($0.00744 / hr)

Spin up a new droplet

To make this as dead simple as possible, you just use one function, without any parameters.

droplet_create()

Using default ssh key: Jane Doe
NB: This costs $0.00744 / hour  until you droplet_delete() it
<droplet>sabaoth (2727258)
  Status: new
  Region: San Francisco 1
  Image: Ubuntu 14.04 x64
  Size: 512mb ($0.00744 / hr)

You can of course pass in lots of options for name of the droplet, RAM size, disk size, ssh keys, etc. See regions() and sizes() to list available regions and sizes.

Delete a droplet

droplet_delete(1707487)

Chaining

Most of the droplet_* functions can be chained together using the %>% function. For example, you can turn a droplet off, snapshot, and then turn it back on with:

d %>%
   droplet_power_off() %>%
   droplet_snapshot() %>%
   droplet_power_on() %>%

R/RStudio

We're still working on these, but would love feedback.

By default, analogsea creates an Ubuntu 20.04 image. Let's say you've got ten students from an R workshop, then you can run

droplet_create("rstudio-20-04")

users <- list(
  user = paste0("student", 1:5),
  password = sapply(rep(8,5), create_password)
)
  
for (i in seq_along(users$user)) {
  ubuntu_create_user(x, users$user[i], users$password[i], keyfile = "~/.ssh/id_rsa")
}

Then, each student shall be able to open RStudio from the browser, by visiting http://123.456.789:8787 and accessing with studentX (X = 1,...,10) and the password created.

To install more R packages, you can make them readily available for all the users that have access to your droplet with

ubuntu_install_r() # ohlf if you didn't use the RStudio image
install_r_package("arrow")

Shiny

The RStudio Server image also features readily available Shiny Server, but we provide the functions ubuntu_install_shiny() and docklet_shinyserver() (requires to use the Docker image, i.e. docklet_create()) to configure Shiny from scratch.

Install RStudio Server

For the standard Ubuntu image you can use ubuntu_install_rstudio().

For dockerized RStudio you have the next option:

docklet_create() %>%
  docklet_rstudio()

This will install R, RStudio Server and it's dependencies. It will automatically pop open the RStudio server instance in your default browser, with default username/password of rstudio/server.