Data Engineer Professional · Domain 8: Debugging and Deploying · 10% of the exam
Task 8.4: Configure Git-based CI/CD workflows using Databricks Git folders (formerly Repos) to deploy notebooks and code.
Production Git folders updated by automation, jobs that run from a Git reference, service principal Git credentials, and branch workflows.
Study it
Git folders in CI/CD
Lesson coming
Sample questions
Try each one before opening the answer. Every option is explained, with the Databricks documentation page that proves it.
Question 1 · choose 1
A production job runs six notebook tasks from a production Git folder that automation updates after every merge to main. A merge during a run left the first tasks on old code and the later ones on new code. Every task in a run must use the same code version, and auditors want to see which commit each run used. What should the team change?
ARestrict updates of the production Git folder to admins
BHave the automation update the Git folder only once an hour
CPoint the tasks at the remote Git repository's main branch
DCopy the notebooks to a shared folder before each release
Show the answer and why
ARestrict updates of the production Git folder to admins
Incorrect
Restricting who updates the folder does not stop the automated update during a run.
BHave the automation update the Git folder only once an hour
Incorrect
A run can still span an update, and the commit used is not recorded per run.
CPoint the tasks at the remote Git repository's main branch
Correct
At run start, Databricks snapshots the branch or commit, all tasks use that snapshot, and the run details show the commit SHA.
DCopy the notebooks to a shared folder before each release
Incorrect
Manual copies drift from source control and record no commit.
Tasks can check out code directly from a remote Git repository. Databricks takes a snapshot of the branch or commit when the run begins, so every task uses the same version, and Databricks recommends this for staging and production jobs rather than workspace paths in Git folders.
A notebook task that checks out code from a remote Git repository fails when it writes a small CSV to a workspace folder. Where should it write?
ATo the Git repository itself
BTo a Unity Catalog volume or a table
CTo /Workspace/Shared
DTo the user's home folder in the workspace
Show the answer and why
ATo the Git repository itself
Incorrect
Tasks do not commit output to the repository.
BTo a Unity Catalog volume or a table
Correct
Git-sourced tasks cannot write to workspace files; persistent output goes to a volume or table.
CTo /Workspace/Shared
Incorrect
That is still a workspace file location.
DTo the user's home folder in the workspace
Incorrect
Home folders are workspace files as well.
Tasks configured with a remote Git repository cannot write to workspace files. They write temporary data to the driver's ephemeral storage and persistent data to a volume or table.
A team sets up a production Git folder that jobs run from. Which setup follows Databricks guidance?
AEach developer keeps it under their own /Workspace/Users folder
BDevelopers edit and push changes directly from the production folder
CAn admin creates it outside user folders; automation updates it
DAll users get Can Manage on it so that anyone can fix it quickly
Show the answer and why
AEach developer keeps it under their own /Workspace/Users folder
Incorrect
User folders are for development checkouts.
BDevelopers edit and push changes directly from the production folder
Incorrect
Production folders should not be edited by hand.
CAn admin creates it outside user folders; automation updates it
Correct
Production Git folders sit outside user folders and are updated only by automation.
DAll users get Can Manage on it so that anyone can fix it quickly
Incorrect
Broad edit rights invite manual changes.
Admins create production Git folders outside /Workspace/Users, give users and service principals Can Run, and keep them in sync with the deployment branch through automation.
A team must deploy jobs, pipelines and code from Git through test and prod. Which approach does Databricks recommend?
ADeclarative Automation Bundles
BA production Git folder for everything
CExporting notebooks to DBC archives
DRecreating jobs by hand in each workspace
Show the answer and why
ADeclarative Automation Bundles
Correct
Databricks recommends bundles to package and deploy CI/CD workflows.
BA production Git folder for everything
Incorrect
A production Git folder fits when only code is deployed.
CExporting notebooks to DBC archives
Incorrect
Archives are not a CI/CD deployment mechanism.
DRecreating jobs by hand in each workspace
Incorrect
Manual setup drifts between environments.
Databricks recommends Declarative Automation Bundles to package and deploy CI/CD workflows; a production Git folder is an option when you deploy only code into the workspace.