Issue creating a custom pipeline in Clara Deploy

I am trying to create a custom pipeline in clara to use a custom operator. I created the operator which uses pytorch and then created a docker image that takes an input from the dicom-reader and outputs an mhd as an input to the dicom-writer. The pipeline file is very similar to the ct-recon example provided and it was created successfully. The issue comes when I trigger the pipeline when it reaches my custom operator it gives me the following error:

ImagePullBackOff: Back-off pulling image “my-docker-image”

The docker image is loaded successfully and I can run a container from it successfully as well.

Do you know what might be the issue?

I solved the problem, apparently you need to have a tag for the image as latest does not work for some reason.

Hi

Sorry for not answering in time. I am glad your figured it out. you are correct you need to tag your image as in k8s, if you use the ‘latest’ tag, it will always try to pull the container (in order to get the latest). See the documentation here https://kubernetes.io/docs/concepts/containers/images/

Thanks