ktaletsk / CuCalc

CUDA + CoCalc Docker container
34 stars 11 forks source link

Kubernetes Cluster support #6

Open gabriel-milan opened 4 years ago

gabriel-milan commented 4 years ago

Hi! I'm just wondering if there's any image of CoCalc with both CUDA and Kubernetes Cluster support. I'm aware of cocalc-kubernetes and CuCalc projects, but unfortunately I couldn't find an image that attends to both. Is that a thing?

ktaletsk commented 4 years ago

Hi! Thanks for you interest. I have to admit, I missed the moment when CoCalc-kubernetes was introduced. To enable CUDA you will need to update this image: https://github.com/sagemathinc/cocalc-kubernetes/blob/master/project/image/Dockerfile to include CUDA. There are multiple ways to do that: one is to change the base image in the above Dockerfile to something that includes CUDA (i.e. https://github.com/ktaletsk/gpu_images); another is to build new image using the above image as a base (CuCalc approach). Plus, you would need to make sure that your Kubernetes nodes are supporting CUDA images (I assume you know how to do that, but feel free to ask me for help).

If you are interested in creating such an image and experimenting with CUDA support in cocalc-kubernetes, we can collaborate on that.

gabriel-milan commented 4 years ago

I'm pretty newbie on Kubernetes, but very interested on the subject.

I found out about CoCalc this week and it got me interested as I'm part of the staff of a running Rancher cluster (it has some GPU machines). I intended to implement this in order to allow users to run small tests before deploying tasks on the cluster (mostly for training ML models).

Honestly I was kinda dropping it because people in the staff were struggling with the idea, but it would be nice to collaborate on this, it would be a great experience for acquiring knowledge.