Hello @siva_pusarla ,

As per official docs, the “absolute path” examples for running/importing notebooks are like:

• /Users/<user>@<org>/folder/notebook (workspace notebooks)
• and for repos, /Repos/... or /Workspace/Repos/... (those two are equivalent for Repos)

So /Workspace/YourFolder/NotebookA is often not a valid notebook object path, unless that exact path exists as a notebook path in your workspace.

Also, when the job is configured to run from a remote Git repository, the notebook being executed is an ephemeral checkout, and paths behave differently between Git folders and workspace folders. Databricks even calls out that Git-folder path behaviour can differ from workspace-folder behaviour.

Please refer to doc: https://kb.databricks.com/libraries/paths-behave-differently-on-git-folders-and-workspace-folders

So in the job run, Databricks tries to resolve /Workspace/YourFolder/NotebookA as a notebook path and can’t find it and hence throws the pathNotFoundError

You need to refer to this doc for best practices to use Notebooks: https://docs.databricks.com/aws/en/notebooks/notebook-workflows

Hi Anudeep,

thanks for your response..

notebookA is not the part of repo code.. it is a constant file in workspace.

I would like to keep some files local to the workspace(for env setup) irrespective of the repos.

| git_folder (GIT)
| -- module
| ---- app.py

...
| Workspace_folder
| -- Common_Utils
| ---- env_setup.py

env_setup is local to different workspaces - dev,test and prod - hence cannot checkin to repo. 

Similar to the above set up i want to run dbutils.notebook.run(/Workspace/Common_Utils/env_setup) from app.py while executing the app through workflow/job

com.databricks.WorkflowException: com.databricks.NotebookExecutionException: FAILED: Unable to access the notebook "Workspace/Common_Utils/env_setup". Either it does not exist, or the identity used to run this job, xxx, lacks the required permissions.

But both the notebook and the permission to the notebook exists and works fine when run outside the job

@siva_pusarla , 

Try to convert env_setup into repo-based code and control behavior via environment

Instead of a workspace notebook, use a Python module in the repo and drive environment differences using:

Job parameters

Branches (dev / test / prod)

Secrets (workspace-specific)

Example repo structure

repo/

├── common_utils/

 |       └── env_setup.py

└── app.py
Example:
env_setup.py

from databricks.sdk.runtime import dbutils

def load_env(env):
if env == "dev":
return {
"catalog": "dev_catalog",
"password": dbutils.secrets.get("app-secrets", "db-password")
}
if env == "test":
return {
"catalog": "test_catalog",
"password": dbutils.secrets.get("app-secrets", "db-password")
}
if env == "prod":
return {
"catalog": "prod_catalog",
"password": dbutils.secrets.get("app-secrets", "db-password")
}

app.py

from common_utils.env_setup import load_env
from databricks.sdk.runtime import dbutils

env = dbutils.widgets.get("env")
config = load_env(env)

print(f"Running in {env}, catalog = {config['catalog']}")

I agree with you approach, but I would like to keep my env set-up away from repo. Limiting the repo only to the application code and env-set up close to the environment where the job runs.

In future, I should be able to run the application in various different environments and every time a change in the env should not update the app code in the repo.

@siva_pusarla: We use the following pattern and it works,

1) Calling notebook - constant location used by Job.

            + src/framework
                   + notebook_executor.py

2) Callee notebooks - dynamic

            + src/app/notebooks
                   + notebook1.py
                   + notebook2.py
                   ....
                   + notebookn.py

Job will send the notebook names as parameter to notebook_executor.py and it will execute notebooks using the below pattern,

notebook_path = f"../app/notebooks/{notebook_name}"
dbutils.notebook.run(notebook_path, TIMEOUT, input_params)

The important learning for us is the usage of relative path for notebook execution. Hope it helps

Hi Siva, thanks for the response.

Yes we use the relative path for calling the notebooks within the repo.

Here I'm trying to call a notebook which is in the workspace from a job that is being executed from repo.

I have ensured all the required permissions but it says "either the notebook doesnot exists or user does not have permissions to run"

I see this behavior starting from last couple of months in 2025.