Issues with ShinyProxy 3.1.1 Docker Image on Docker Swarm Backend

[ziyunxiao]

Description:

I discovered two issues with the ShinyProxy 3.1.1 Docker image on a Docker Swarm backend.

Issue 1: Port Access Error

During startup, some processes attempt to open port 80 on localhost but encounter a permission error due to running as a non-root user (shinyproxy). This is because the normal user cannot open ports <1024. A temporary workaround is to build a custom image that runs as root. However, a proper fix is required for security reasons.

Issue 2: Redis Connection Error

After bypassing the port access error, I encountered a connection error to Redis due to authentication failure. Despite providing the correct password, the same configuration works correctly on Kubernetes. This suggests that the issue is specific to Docker Swarm.

The workaround is to remove setting of store-mode: redis

Steps to Reproduce:

1. Follow standard deployment procedures for ShinyProxy 3.1.1.
2. Set container-backend: docker-swarm
3. Set backend & Redis

spring:
  application:
    name: Data Science Applications
  session:
    store-type: redis
  redis:
    host: redis-service
    password: xxxxxxxxxx
proxy:
  container-backend: docker-swarm
  store-mode: redis
  # ... other settings

4. startup Shinyproxy and watch the shinyproxy service logs

example of docker-compose.yml

services:
  shiny:
    image: openanalytics/shinyproxy:3.1.1
    volumes:
      - /var/run/docker.sock:/var/run/docker.sock
      - ./conf/example.yml:/opt/shinyproxy/application.yml
    networks:
      - shiny-net

Best Regards,

Robin

[LEDfan]

Hi

I tested the procedure described here https://github.com/openanalytics/shinyproxy-config-examples/tree/master/07-containerized-docker-swarm and I don't get any error for a process trying to use port 80. ShinyProxy doesn't use port 80 so I'm not sure where this originates from.

Regarding redis: I don't experience this issue. Did you maybe try to change the redis password of an existing deployment? This sometimes does not work and then it's better to start a fresh redis deployment.