09-08-2021 05:43 PM
09-09-2021 06:54 AM
I found the answer to my question for how to restrict library installation both on clusters and notebooks and posting it here for others reference: we can restrict outbound access to the public pypi at the workspace level using these instructions https://docs.databricks.com/administration-guide/cloud-configurations/aws/customer-managed-vpc.html#...
09-09-2021 06:28 AM
you can assign permissions to (existing) clusters.
There are 4 kinds of permissions:
So basically if you assign Attach or Restart permissions, the user is not allowed to install libraries.
That is only allowed with the Manage permission.
09-09-2021 06:54 AM
I found the answer to my question for how to restrict library installation both on clusters and notebooks and posting it here for others reference: we can restrict outbound access to the public pypi at the workspace level using these instructions https://docs.databricks.com/administration-guide/cloud-configurations/aws/customer-managed-vpc.html#...
09-13-2021 10:28 AM
With it being restricted at the cluster level as well how are you installing libraries that need to be added adhoc? Did you block the entire pypi domain?
09-15-2021 03:07 PM
Admin can set up a connection to Azure Artifactory/Jfrog Artifactory or other artifactories of required libraries for the workloads.
09-09-2021 10:28 AM
@Mojgan Mazouchi - I'm so glad you found the answer and shared it with us. Thank you!
09-13-2021 11:34 AM
one way to manage is make the cluster permission only to can restart and then use an init script to install libraries on start up so that users wont install libraries on the fly.