Shiny - Display Dataset - Unable to connect to worker after 60.00 seconds

shiny
, , ,
1

Hi,

I'm relatively new to Shiny apps and am currently encountering an issue when deploying a dashboard that is supposed to display a data frame. While the dashboard works perfectly locally, I encounter the error "Unable to connect to worker after 60.00 seconds; startup took too long." when I access online after deploying the app.

Here is the code snippet used to load the data frame, build the UI and server logic, and deploy the app:

# Load packages 
library(shiny)
library(tidyverse)
library(data.table)
library(DT)
library(shinydashboard)

# Load data 
df_full <- readRDS("df_full.rds")

# Create Dashboard
ui <- fluidPage(
  titlePanel("Construct Validity Data Explorer"),
  sidebarLayout(
    sidebarPanel(
      selectInput("countryInput", "Choose a Country:", choices = c("All" = "All", unique(df_full$Country))),
      selectInput("numPartiesInput", "Number of Parties:", choices = c("All" = "All", unique(df_full$`Number of Parties`)))
    ),
    mainPanel(dataTableOutput("validityTable"))
  )
)

server <- function(input, output, session) {
  filtered_data <- reactive({
    temp_data <- df_full
    if (input$countryInput != "All") {
      temp_data <- temp_data[Country == input$countryInput]
    }
    if (input$numPartiesInput != "All") {
      temp_data <- temp_data[`Number of Parties` == as.numeric(input$numPartiesInput)]
    }
    temp_data
  })
  
  output$validityTable <- renderDataTable({
    filtered_data() %>% select(Country, `Number of Parties`, starts_with("Construct Validity"))
  })
}

shinyApp(ui = ui, server = server)

# Deploy app 
# Here, I just replaced the info for the purposes of this post
rsconnect::setAccountInfo(name='XXXX', token='XXX', secret='XXX') 
rsconnect::deployApp(forceUpdate = "TRUE")

Attempts to diagnose and fix the issue include:

  1. Reviewing logs, which haven't indicated that there is a problem.
  2. Deploying a simpler app without data loading, which worked, suggesting the deployment process itself is not the issue.
  3. Testing with a significantly smaller data frame (1KB) to rule out issues related to data size of the original data frame (315KB), which also didn't work.
  4. Increasing the start-up timeout to 300s - the app still does not load.

I would appreciate any help with this! Thank you very much in advance!

Cheers,
Jakob

In case it helps, this is what the data structure looks like:

> head(df_full)
# A tibble: 6 × 7
  Country        `Number of Parties` `Party Combination`                                                              Construct Validity (Unw…¹ Construct Validity (…² Construct Validity (…³ Construct Validity (…⁴
  <chr>                        <dbl> <chr>                                                                                                <dbl>                  <dbl>                  <dbl>                  <dbl>
1 Albania (2017)                   2 Partia Socialiste e Shqiperise (PS), Partia Demokratike e Shqiperise (PD)                            0.715                  0.915                  0.810                  0.858
2 Albania (2017)                   2 Partia Socialiste e Shqiperise (PS), Levizja Socialiste per Integrim (LSI)                           0.664                  0.743                  0.675                  0.593
3 Albania (2017)                   2 Partia Socialiste e Shqiperise (PS), Partia Drejtesi, Integrim dhe Unitet (PDIU)                     0.635                  0.632                  0.652                  0.503
4 Albania (2017)                   2 Partia Socialiste e Shqiperise (PS), Lista e Barabarte (LIBRA)                                       0.668                  0.637                  0.662                  0.446
5 Albania (2017)                   2 Partia Socialiste e Shqiperise (PS), Partia Socialdemokrate e Shqiperise (PSD)                       0.590                  0.554                  0.605                  0.361
6 Albania (2017)                   2 Partia Demokratike e Shqiperise (PD), Levizja Socialiste per Integrim (LSI)                          0.458                  0.353                  0.405                  0.405
# ℹ abbreviated names: ¹​`Construct Validity (Unweighted Spread)`, ²​`Construct Validity (Unweighted Weighted)`, ³​`Construct Validity (Unweighted Mean Distance)`, ⁴​`Construct Validity (Weighted Mean Distance)````
2

When you deployed the app, did you include the .rds file in the set of files being uploaded to the server, and was it in the same directory as the app itself?

3

Thank you very much for the quick response! I think so - I have stored it in the same directory and the deployment logs (see excerpt below) indicate that the file is included and uploaded properly:

ℹ Bundling 2 files: app.R and df_full.rds
ℹ Capturing R dependencies
✔ Found 123 dependencies
✔ Created 374,209b bundle
ℹ Uploading bundle...
✔ Uploaded bundle with id 10079136
4

If the app you deployed actually includes the last few lines you show (invoking functions from rsconnect), try removing those (although I'm not sure why it would work without data loading if you included those lines). You don't want to try deploying an app from within itself.

5

That was indeed the problem. I can't believe I didn't realize this - thank you so much for your help!