Shiny app deployment error - Out of memory(oom)

I keep encountering this error when running my app in -> Container event from container-8169095: oom (out of memory). I am using the basic plan. I am working with large raster datasets(I am trying to create a shiny app that calculates the total population of people within FAO farming system boundaries for all the countries in the world. I am also calculating mean farming yields for 42 crops which are all 1Gb in size), performing calculations that require a lot of RAM. Here is my code. If you are able to help, kindly assist. I've been stuck for days.


Read Africa countries shapefile and transform to WGS84

gadm_sf <- st_read("Africa_countries.shp") %>%
st_transform("+init=EPSG:4326") %>%

Load farming system boundaries from FAO and transform to WGS84

fao_sf <- st_read("fao_gadm_intersect.shp") %>%
st_transform("+init=EPSG:4326") %>%

Load WorldPop population dataset and transform to WGS84

wp_path <- "H:\CIMMYT Shiny project\Data_2\AFR_PPP_2020_adj_v2.tif"
worldpop_pop <- terra::rast(wp_path)

Set the CRS of fao_sf to match worldpop_pop

fao_sf <- st_transform(fao_sf, st_crs(worldpop_pop))

Set the CRS of fao_sf to match worldpop_pop

gadm_sf <- st_transform(gadm_sf, st_crs(worldpop_pop))

#load spam dataset

physical area from SPAM 2010

mylist <- list.files(pattern="._A.tif$")

r <- rast(mylist)

crop extent to gadm_sf

spam_yield <- terra::crop(r, gadm_sf)

sum crop area for all 42 crops

agext <- sum(spam_yield)

#Define the UI
ui <- bootstrapPage(
theme = shinytheme("flatly"), collapsible = TRUE, id = "nav",
HTML('Agri population mapper'),
windowTitle = "Agricultural population calculator",
tabPanel("Calculate Population",
width = 4,
inputId = "country",
label = tags$span("Select Country",
style = "cursor: help;",
title = "Select the country for which you want to calculate agricultural population. The data available includes information on the farming system and the population density of different administrative levels."),
choices = unique(gadm_sf$ADM0_NAME)
width = 4,
inputId = "farming_system",
label = tags$span("Select Farming System",
style = "cursor: help;",
title = "Select the farming system for which you want to calculate agricultural population. The available options depend on the country selected."),
choices = NULL
width = 4,
inputId = "calculate",
label = "Calculate Population"
tableOutput(outputId = "result_table"),
leafletOutput(outputId = "result_map")

         numericInput(inputId = "maxrows", label = "Rows to show", value = 25),
         "This tool uses data provided by ",
         tags$a(href = "", "GADM"),
         ", ",
         tags$a(href = "", "FAO"),
         ", ",
         tags$a(href = "", "WorldPop"),
         ", and ",
         tags$a(href = "", "SPAM"),

tabPanel("About this site",
           tags$h4("What is the Agricultural Population Mapper?"),
           "The Agricultural Population Mapper is a tool that allows you to estimate the population living in rural areas and engaged in different farming systems in a given country.",
           tags$br(), tags$br(),
           tags$h4("How can I use it?"),
           "To use the Agricultural Population Mapper, simply select the country and farming system you're interested in and click the 'Calculate Population' button. The tool will return a table and a map showing the population and population density of different administrative levels within the selected country and farming system.",
           tags$br(), tags$br(),
           "Code and input data used to generate this Shiny mapping tool are available on ",
           tags$a(href="", "Github."),
           "This app was developed by CIMMYT as part of Ex Ante Evaluation Support to Prospective Agronomy Interventions."


Define Server

Define Server

server <- function(input, output, session) {

Update farming system choices based on selected country

observeEvent(input$country, {
country_name <- input$country
farming_system_choices <- unique(fao_sf$DESCRIPTIO[fao_sf$ADM0_NAME == country_name])
updateSelectInput(session, "farming_system", choices = farming_system_choices)

Calculate agricultural population and yield based on user selections

observeEvent(input$calculate, {
country_name <- input$country
farming_system_name <- input$farming_system
farming_system_geom <- fao_sf[fao_sf$DESCRIPTIO == farming_system_name & fao_sf$ADM0_NAME == country_name, ]

country_geom <- gadm_sf[gadm_sf$ADM0_NAME == country_name, ]
farming_system_geom <- fao_sf[fao_sf$DESCRIPTIO == farming_system_name & fao_sf$ADM0_NAME == country_name, ]
intersection_extent <- st_intersection(country_geom, farming_system_geom)

##create a new raster object rr using the same resolution and extent as the worldpop_pop raster, but based on the geometry of the farming_system_geom.
rr <- terra::rast(intersection_extent, resolution = res(worldpop_pop), ext = ext(worldpop_pop))

##rasterize the farming system geometry to be used in masking
farming_system_raster <- terra::rasterize(intersection_extent, field = "gridcode", rr, fun = "sum", overwrite = TRUE)

#mask population based on farming system selected
worldpop_pop_mask <- terra::mask(worldpop_pop, farming_system_raster)

# Resample spam_yield to have the same extent as worldpop_pop
spam_yield_resampled <- terra::resample(agext, rr)

spam_yield_mask <- terra::mask(spam_yield_resampled, farming_system_raster) 

total_population_sum <- sum(as.vector(worldpop_pop_mask), na.rm = TRUE)
mean_population <- mean(as.vector(worldpop_pop_mask), na.rm = TRUE, FUN = mean)
mean_yield <- mean(as.vector(spam_yield_mask), na.rm = TRUE, FUN = mean)

output$result_table <- renderTable({
  data.frame("Total_Population" = total_population_sum,
             "mean_Population" = mean_population,
             "mean_SPAM_Yield" = mean_yield)

# convert SpatRaster to rasterLayer object
worldpop_pop_mask_raster <- raster(worldpop_pop_mask)

pal0 <- colorNumeric(c("RdYlBu"), na.omit(values(worldpop_pop_mask_raster)),
                     na.color = "transparent")

output$result_map <- renderLeaflet({
  leaflet() %>% 
    addProviderTiles("CartoDB.Positron") %>% 
    addRasterImage(worldpop_pop_mask_raster, colors = pal0, opacity = 0.9) %>%
    addLegend(pal = pal0, values = values(worldpop_pop_mask_raster), 
              title = "Population",
              position = "bottomright")


Run the app

shinyApp(ui = ui, server = server)

did get any solution

there are two routes to take to solve an outofmemory issue, and 1 supporting activity.

the supporting activity is to profile your application so as to have a good knowledge of how it is using memory, and so that if you make a change in the hope of improving your memory use, you have a method for judging whether you succeeded or failed.

the two solutions are

  1. find a way to be more efficient and so use less peak memory. If your current approach is perfectly efficient this may be impossible, though often memory use has not been optimised and there will be improvements possible. You can potentially get elaborate with all sorts of chunking schemes, it very much depends what you are doing.
  2. get (i.e. pay for) more memory to be available.
1 Like

Thanks for your response.

I found there are four types of paid plans. So how can I decide which plan is suitable? I am new to the Shiny app. How can I find the amount of memory used by the app, and can we get details of memory available on various plans Shiny provides?

I mentioned profiling, which is exactly this, understanding how your app is working.
There are materials out there you can search for. In particular look for profvis

Thanks @nirgrahamuk

Yes, I did. I had to find a nother way to handle the data. it was very huge hence out of memory error.

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.

If you have a query related to it or one of the replies, start a new topic and refer back with a link.