in any case, the docker image must have access to the (distribution-specific) Hadoop JARs and the cluster configuration files (typically from /etc/hadoop/conf/) to be able to talk to the Hadoop cluster.
There are several ways to achieve that, not sure which is simpler, depending on your setup. Maybe you could build a docker image derived from our base DSS image by adding the relevant Hadoop packages (the base image is Debian-7 based) and grab the configuration files by mounting the underlying system's /etc/hadoop directory?
For more precise / personalized help, we'd need to know more about your setup. I suggest you go to support.dataiku.com and open a ticket there so you can add relevant the information