Shiny app loading locally, disconnects from server when deployed

Question

My app runs fine locally. I can interact with it, etc. But when deployed globally it immediately throws the "Disconnected from Server" error. I've been googling for days and have tried everything I know how to try.

First things first:

I tried uninstalling then reinstalling a bunch of packages
I updated all the packages I am using
There is no information at all in the shiny logs. "This application currently has no logs."
My data (csv) is only 17MB (3 columns, 1M-ish rows)
when I reduced it to only about 200 rows I continued having the same issue
I went through the settings, changed the startup timeout to 300 and systematically changed each of the other settings, one by one.
I've read the browser's JavaScript logs end to end. They didn't make a ton of sense (probably because I don't know JavaScript), but nothing obvious stood out.
I am not loading my data with the entire file path, just the end part

My "Deploy" log in RStudio indicates successful deployment:

Preparing to deploy application...DONE
Uploading bundle for application: 193997...DONE
Deploying bundle: 994423 for application: 193997 ...
Waiting for task: 489405538
  building: Building image: 992534
  building: Fetching packages
  building: Installing packages
  building: Installing files
  building: Pushing image: 992534
  deploying: Starting instances
  rollforward: Activating new instances
  terminating: Stopping old instances
Application successfully deployed to https://jesstme.shinyapps.io/shinynames/
Deployment completed: https://jesstme.shinyapps.io/shinynames/

Link to the app itself: https://jesstme.shinyapps.io/shinynames/

Server code:

#set wd & environment----
setwd("/Users/OldJess/Dropbox/R Stuff (Home)/ShinyNames")

#load packages------
library(datasets)
library(ggplot2)
library(viridis)
library(ggthemes)
library(gridExtra)
library(dplyr)
library(rdrop2)
library(shiny)
library(devtools)

#base <- read.csv("data/NationalNamesBrief.csv", stringsAsFactors = FALSE,     row.names = NULL, na.strings = c("NA","","#MULTIVALUE"))

#temporary df for demonstration purposes
base <- structure(list(Name = c("Ellie", "Ellie", "Ellie", "Ellie", "Ellie", 
                            "Ellie"), 
                   Year = c(1880L, 1881L, 1882L, 1883L, 1883L, 1884L), 
                   Gender = c("F", "F", "F", "F", "M", "F"), 
                   Count = c(17L, 27L, 37L, 24L, 7L, 28L)), 
              .Names = c("Name", "Year", "Gender", "Count"), 
              row.names = c(NA, 6L), class = "data.frame")

#clean data----
base$name <- tolower(base$Name)
base$MF <- as.factor(base$Gender)

#add ranking data by Year
base <- base %>%
  group_by(Year) %>%
  arrange(Year, desc(Count)) %>%
  mutate(Rank = row_number())

#add ranking data by Year AND Gender
base <- base %>%
  group_by(Year, Gender) %>%
  arrange(Year, desc(Count)) %>%
  mutate(GenderRank = row_number())

#create functions----
#function to create line & heat charts
lineHeatCharts <- function(pickaname){
  pickanameLower <- tolower(pickaname)
  subDf <- subset(base[base$name == pickanameLower,])
  heat <- ggplot(subDf, aes(x = Year, y = MF, fill = Count)) +
    scale_fill_viridis(name = "",
                   option = "B", 
                   limits = c(0, max(subDf$Count))) +
    geom_tile(color = "white", size = 0) +
    theme_tufte() +
    theme(axis.text.x = element_text(angle = 90, vjust = 1, hjust = 1),
      axis.ticks.x = element_blank()) +
    scale_x_continuous(breaks = seq(min(subDf$Year), 
                                max(subDf$Year), by = 5)) +
    labs(x = "Year", y = "")
  line <- ggplot(subDf, aes(x = Year, y = Count, fill = MF)) +
geom_line(aes(colour = factor(subDf$Gender)), size = 1.5) +
theme_tufte() +
theme(axis.text.x = element_blank(),
      axis.ticks.x = element_blank()) +
scale_x_continuous(breaks = seq(min(subDf$Year), 
                                max(subDf$Year), by = 5)) +
labs(x = "", y = "", color = "")
  return(grid.arrange(line, heat, 
                  ncol = 1, nrow = 2,
                  heights = c(5, 2), top = max(subDf$Name)))
} 

# Define server logic 
function(input, output) {

  output$view <- renderPlot({
    lineHeatCharts(input$list)
  })
}

UI code:

library(shiny)
library(shinythemes)

# Define UI for dataset viewer application
fluidPage(theme = shinytheme("flatly"),

  # Application title
  titlePanel("First Names on U.S. Social Security Applications, 1880 - 2014"),

  sidebarLayout(
    sidebarPanel(
      textInput(inputId = "list", label = "Enter a name:", value = "Ellie"),

      helpText("Note: This page will take about 30 seconds to load the first     time you open it. Data are from US Social Security applications via data.gov. For privacy, only names with at least 5 babies per year are included. Errors in Social Security form submission, like incorrect sex, are not corrected. Names with special characters and spaces are not included."),

  submitButton("Refresh View")
),

mainPanel(
  h4(""),
  plotOutput("view")
    )
  )
)

jesstme · Accepted Answer

As it turns out there were two problems: 1) I needed to remove the setwd() 2nd line of my code 2) Shiny logs weren't working.

I posted on Google's Shiny forum, and a RStudio person fixed the issue with the logs. Once the logs were working I saw that the error pointed to my attempts to setwd. Deleted that and the problem was fixed. Keeping this question up because I'm sure this issue will plague someone else at some point.

Shiny app loading locally, disconnects from server when deployed

Answers (2)

Related Questions