azure machine learning and R using azuremlsdk - supported R version and custom_docker_image

Question

So we have to move away from using SQL Server Machine Learning services as it only supports R 3.5.2 even for SQL Server 2019!

I am trying hard to go all 21st century and deploy some of our on prem R trained models as web service as described by one of the Microsoft evangelists David Smith (see code below).

Looking at r_environment I noticed to my horror that, if I do not use a custom docker image, predefine images only support R 3.6?! Is this correct? If so, how do I create a custom docker image and why does Microsoft suggest using Azure ML where there are also restrictions in terms of the R version!

PS:

Some code to possibly replicate my issues:

Train model locally:

library(datasets)
library(caret)

data(iris)

setwd("C:/Data")

index <- createDataPartition(iris$Species, p=0.80, list=FALSE)
testset <- iris[-index,]
trainset <- iris[index,]

model = train(Species ~ ., 
                  data=trainset, 
                  method="rpart", 
                  trControl = trainControl(method = "cv"))

saveRDS(model, "model.rds")

I can deploy this model in Azure ML:

Scoring script score.r

library(jsonlite)

init <- function()
{
  model_path <- Sys.getenv("AZUREML_MODEL_DIR")
  model <- readRDS(file.path(model_path, "model.rds"))
  message("iris classfication model loaded")
  
  function(data)
  {
    vars <- as.data.frame(fromJSON(data))
    prediction <- predict(model, newdata=vars)
    toJSON(prediction)
  }
}

Failing code:

library(azuremlsdk)

interactive_auth <- interactive_login_authentication(tenant_id="xxx")

ws <- get_workspace(
        name = "amazing_work_space", 
        subscription_id = "xxx", 
        resource_group ="xxx", 
        auth = interactive_auth
)

model <- get_model(ws, name = "iris_classification")

r_env <- r_environment(name = 'myr_env',
                       version = '1')

inference_config <- inference_config(
  entry_script = "score.R",
  source_directory = ".",
  environment = r_env)

aci_config <- aci_webservice_deployment_config(cpu_cores = 1, memory_gb = 0.5)

aci_service <- deploy_model(ws, 
                            'xxx', 
                            list(model), 
                            inference_config, 
                            aci_config)

wait_for_deployment(aci_service, show_output = TRUE)