GCP Vertex AI Workbench custom image - persistence with gcs don't work

Question

While creating a workbench with a custom jupyter image I choose backup/ persistence with GCS (google cloud storage) and it doesn't work. For now, I tried to test with:

jupyter/base-notebook:python-3.8.8 https://github.com/Santhin/jupyter-images/blob/main/base-v2/Dockerfile
gcr.io/deeplearning-platform-release/base-cpu:latest https://github.com/Santhin/jupyter-images/blob/main/vertex/Dockerfile

I thought it was a problem with permissions on the folder named jupyter which is created always while creating a workbench on every jupyter image (custom / pre-built by gcp). Is there any method to accomplish persistence with gcs with a custom docker image? Screen showing option with backup on gcs

Backup seems to work with pre-built environments but not with custom containers right now. As a workaround you can sync the data manually in `SSH` instance running these commands`$cd /home/jupyter ` ```$ gsutil cp * [GCS PATH]``` , it will copy all the data to your `GCS` bucket. Maybe you can try to create a script using these commands and then schedule the script to run and sync data every hour. Refer to this [Scheduling a script to Run Daily](https://stackoverflow.com/questions/34753831/execute-a-shell-script-everyday-at-specific-time) — Sayan Bhattacharya, Jun 20 '22 at 06:38

GCP Vertex AI Workbench custom image - persistence with gcs don't work

0 Answers0