ShinyProxy container fails to respond despite app listening on 0.0.0.0:3838 inside docker

ShinyProxy
1

I am trying to host my shinyproxy applications in AWS’s Amazon Linux 2023. The instance can listen on port 80 only. Hence I have configured application.yml to as follows:

proxy:
  title: SMART Tool
  container-wait-attempts: 15
  hide-navbar: false
  logo-url: REDACTED
  landing-page: /
  heartbeat-rate: 30000
  heartbeat-timeout: 120000
  port: 80
  docker:
    url: unix:///var/run/docker.sock
    internal-networking: true
    network: shinyproxy-net
    port-range-start: 20000
    skip-image-pull: true
  template-groups:
    - id: adot
      properties:
        display-name: Arizona Department of Transportation
  specs:
  - id: inrixspeedmodule
    template-group: adot
    display-name: Inrix Speed Module
    description: Perform before and after median speed comparisons statewide using INRIX as the data source
    container-image: inrixspeedmodule
    port: 3838
    access-groups: [REDACTED,REDACTED]

I launch shinyproxy using:

sudo java -jar /opt/shinyproxy.jar --spring.config.location=/home/ec2-user/test_smart_tool/shinyproxy/application.yml

The user on an external browser can access the shinyproxy running on AWS through https://shinyanalytics-lb.colo-prod-aws.arizona.edu/ , which displays the list of available applications. Clicking on one of the applications (inrixspeedmodule) from the list, I am redirected to that application’s page. In there, it tries to launch the app but ends up saying "Failed to start app Inrix Speed Module" (after 55 attempts to start the app) after which it deletes the container from docker as well

In the app.R file of inrixspeedmodule, I have the following loc:

shiny::shinyApp(ui = ui, server = server, options =  list(host = "0.0.0.0", port = 3838))

The shinyproxy logs keep saying

2025-03-26 11:30:23.707  WARN 41220 --- [pool-2-thread-1] e.o.shinyproxy.ShinyProxyTestStrategy    : [user=kirthanaramesh@arizona.edu proxyId=62f3b924-9b6e-4251-a8f8-af58fd4128f2 specId=inrixspeedmodule] Container unresponsive, trying again (54/55): http://cf4ccd32f24c:3838
2025-03-26 11:30:24.138  WARN 41220 --- [pool-2-thread-1] e.o.containerproxy.service.ProxyService  : [user=kirthanaramesh@arizona.edu proxyId=62f3b924-9b6e-4251-a8f8-af58fd4128f2 specId=inrixspeedmodule] Container did not respond in time

[ec2-user@shinyanalytics ~]$ docker container ls
CONTAINER ID   IMAGE              COMMAND                  CREATED          STATUS          PORTS      NAMES
444abee96cd0   inrixspeedmodule   "R -e 'shiny::runApp…"   12 seconds ago   Up 11 seconds   3838/tcp   admiring_hermann

I checked the docker logs. There the container starts up without any issues and says Listening on http://0.0.0.0:3838

How to make ShinyProxy detect the container as “responsive” when the app runs fine inside Docker?

2

Hi, if you are running ShinyProxy from the jar (and not in docker), you cannot use internal-networking: true. If you remove this line, ShinyProxy will contact the docker container using a port mapping and it should work.

FYI you configuration file contains a few settings that do no exist in ShinyProxy: proxy.container-wait-attempts, proxy.docker.network, proxy.docker.skip-image-pull. Out of interest, where did you find these settings?

3

What worked in my case is I had to set
docker:
url: unix:///var/run/docker.sock
internal-networking: true
specs:

  • id: inrixspeedmodule
    display-name: Inrix Speed Module
    description: “”
    container-network: host
    container-image: inrixspeedmodule
4

Thanks for pointing that out. I didn’t add those settings myself. They were already present in when I started working on it. I believe they were added by a senior who initially set up the system.