/container_jupyter_gcp

Primary LanguagePython

Container for launching custom docker inside google cloud vertex ai notebook. These are the steps for container registry:

    1. build docker and tag it: docker build . -t TAG_NAME
    1. Link it with GCP Project: docker tag TAG_NAME eu.gcr.io/GCP_PROJECT_ID/TAG_NAME
    1. Push it, upload docker into container registry: docker push eu.gcr.io/GCP_PROJECT_ID/TAG_NAME
    1. create a custom notebook instance in vertex ai specifying the custom docker image just uploaded.

The following are the steps for artifact registry, more secure solution:

    1. create repo in artifact with REPO_NAME
    1. authentication: gcloud auth configure-docker europe-west1-docker.pkg.dev
    1. build: docker build . -t LOCAL_NAME
    1. tag: docker tag LOCAL_NAME europe-west1-docker.pkg.dev/PROJECT_ID/REPO_NAME/LOCAL_NAME
    1. push: docker push europe-west1-docker.pkg.dev/PROJECT_ID/REPO_NAME/LOCAL_NAME

Before everything its' necessary to manually create (or from cloud shell) the repository.

Useful links: