Hey everyone, I am in the process of trying to set up NVIDIA DIGITS in our local Kubernetes Cluster. Basically I want this running on our cluster so that I can access the Web UI and run some models. From the documentation I am reading it looks like they describe how to run the container on a localhost. Has anyone had any luck setting up DIGITS in a Kubernetes environment. I am fairly new to Kubernetes configurations. I assumed I would just just create a Pod yaml file and point to the DIGITS container but I still cannot access the UI from our cluster. In assistance is greatly appreciated.
For anyone who may run across this issue in the future, I finally figured it out. All I had to do was issue the following commands after setting up an NGINX Load Balancer:
kubectl run --namespace digits --image=nvcr.io/nvidia/digits:20.02-tensorflow-py3 digits --port=8888
kubectl expose deployment digits -n digits --port=8888 --target-port=5000 --name=digits --type=LoadBalancer