Solved... adding SE_OPTS will help you to set capabilites

docker run -d -e SE_OPTS="-browser browserName=chromeku,version=56.0,maxInstances=3,platform=WINDOWS" --link selenium-hub:hub selenium/node-chrome:2.53.0

There are multiple ways of doing this and SE_OPTS is one of them, however for me it complicated what I was trying to accomplish. Using SE_OPTS forced me to set capabilities I didn't want to change, otherwise they would be reset to blank/null

I wanted to do:

SE_OPTS=-browser applicationName=Testing123

but I was forced to do:

SE_OPTS=-browser applicationName=Testing123,browserName=firefox,maxInstances=1,version=59.0.1

Another way to set capabilities is to supply your own config.json

-nodeConfig /path/config.json

You can find a default config.json

Or you can start the node container and copy the current one from it

docker cp <containerId>:/opt/selenium/config.json /host/path/target

You can also take a look at entry_point.sh, either on github or on the running container:

/opt/bin/entry_point.sh   

You can run bash on the node container via:

sudo docker exec -i -t <container> bash

This will let you see how SE_OPTS is used and how config.json is generated. Note config.json is generated only if you don't supply one.

/opt/bin/generate_config   

By examining generate_config you can see quite a few ENV vars such as: FIREFOX_VERSION, NODE_MAX_INSTANCES, NODE_APPLICATION_NAME etc.

This leads to the third way to set capabilities which is to set the environment variables being used by generate_config, in my case APPLICATION_NODE_NAME

docker run -d -e "NODE_APPLICATION_NAME=Testing123"

Finally, when using SE_OPTS be careful not to accidentally change values. Specifically, the browser version. You can see by looking at entry_point.sh the browser version is calculated.

FIREFOX_VERSION=$( firefox -version | cut -d " " -f 3 )

If you change it to something else you will not get the results you are looking for.