aws.ec2-furrr.R

#devtools::install_github("cloudyr/aws.ec2")
# devtools::install_github("cloudyr/aws.ec2", ref = devtools::github_pull(38))
# ^ Must actually install PR 38. Some broken code in the main repo
#devtools::install_github("cloudyr/aws.signature")

library(aws.signature)
library(aws.ec2)
library(furrr)
aws.signature::use_credentials()

# Louis's AMI
image <- "ami-fd2ffe87"
describe_images(image)

# Check your VPC and Security Group settings
#s <- describe_subnets()
g <- describe_sgroups("sg-7497f03c") # <- Your preferred security group here
kp <- describe_keypairs("dvaughan") # <- Your keypair here

# Launch the instance using appropriate settings
i <- run_instances(image = image,
                   type = "t2.medium",
                   sgroup = g,
                   min = 2L, # <- Launching 2 medium instances
                   keypair = kp$dvaughan)

##### Start an as.cluster.ec2_instance function around here
# It should accept the i and the .pem file path
# Below is the possible implementation


# Ensure we are running and initialized
is_running <- vector("logical", length(i))
is_initialized <- vector("logical", length(i))

while(!all(is_initialized)) {

  for(ii in seq_along(i)) {

    # Current instance
    i_ii <- instance_status(i[[ii]])

    # Initially, we don't get any information
    if(length(i_ii) == 0) {
      next()
    }

    # First check if we are at least running
    if(!is_running[ii]) {
      if(unlist(i_ii$item$instanceState$name) == "running") {
        is_running[ii] <- TRUE
        message("Instance ", ii, " is running. Now initialzing.")
      }
    }

    # Then check if we are initialized
    if(!is_initialized[ii]) {
      if(unlist(i_ii$item$instanceStatus$status) == "ok") {
        is_initialized[ii] <- TRUE
        message("Instance ", ii, " is initialized.")
      }
    }

  }

}

# Get the public IPs
ips <- vapply(
  i,
  function(i_ii) {
    i_di <- describe_instances(i_ii)
    i_di[[1]]$instancesSet[[1]]$networkInterfaceSet$privateIpAddressesSet$association$publicIp
  },
  FUN.VALUE = character(1)
)

public_ip <- ips

# This is where my pem file lives (password to connect essentially).
ssh_private_key_file <- "~/Desktop/programming/AWS/key-pair/dvaughan.pem"

# Connect!
cl <- makeClusterPSOCK(

  ## Public IP number of EC2 instance
  public_ip,

  ## User name (always 'ubuntu')
  user = "ubuntu",

  ## Use private SSH key registered with AWS
  rshopts = c(
    "-o", "StrictHostKeyChecking=no",
    "-o", "IdentitiesOnly=yes",
    "-i", ssh_private_key_file
  ),

  ## Set up .libPaths() for the 'ubuntu' user and
  ## install future/purrr/furrr packages
  rscript_args = c(
    "-e", shQuote("local({p <- Sys.getenv('R_LIBS_USER'); dir.create(p, recursive = TRUE, showWarnings = FALSE); .libPaths(p)})"),
    "-e", shQuote("install.packages(c('future', 'purrr', 'furrr'))")
  ),

  dryrun = FALSE
)

############## Now we have a cluster object we can use with future

# # Run all future / furrr code here!
plan(cluster, workers = cl)

future_map(1:2, ~.x)

#future_map(1:2, ~Sys.sleep(10))


# stop_instances(i)
# start_instances(i)

# Shutdown permanently
parallel::stopCluster(cl)
terminate_instances(i)