One Chrome node of many hangs when running parallel tests

[Pickford]

I've setup a Selenium Grid using the latest docker-hub 3.3.1 on Ubuntu 16 and use Jenkins to manage my test runs. 

I spin up the latest chrome-node using docker-compose scaling to 12 nodes.

When I run tests in parallel one node always hangs at the getDriver stage of the test.

I have adjusted my docker-compose.yml file to try and accommodate browser / grid timeouts but no combination of settings I've used to date have rectified the hanging node.

CHROME_DRIVER_VERSION=2.29

Chrome version: google-chrome-stable=google-chrome-stable_57.0.2987.133-1

Docker: Server Version: 17.04.0-ce

Operating System: Ubuntu 16.04.2 LTS

OSType: linux

Architecture: x86_64

CPUs: 8

Total Memory: 31.42GiB

Docker compose yml file:

seleniumhub:

  image: selenium/hub

  ports:

    - 4444:4444

  privileged: true

  environment:

    - GRID_BROWSER_TIMEOUT=100000

    - GRID_TIMEOUT=90000

chromenode:

  image: selenium/node-chrome

  ports:

    - 5900

  links:

    - seleniumhub:hub

  environment:

   SCREEN_WIDTH: 1920

   SCREEN_HEIGHT: 1200

  volumes:

    - /dev/shm:/dev/shm

Any suggestions? Or questions that might lead to a solution?