Databricks provides the dbutils module as a built-in set of utilities to interact with Databricks services directly from notebooks.
These utilities allow you to:

• Manage files in DBFS
• Interact with jobs
• Handle secrets securely
• Control notebook execution flow
• Work with widgets for parameterization

1. Hierarchy Overview

The main structure of dbutils is:

dbutils
│
├── credentials (DatabricksCredentialUtils)
├── data (DataUtils)
├── fs (DbfsUtils)
├── jobs (JobsUtils)
├── library (LibraryUtils)
├── meta (MetaUtils)
├── notebook (NotebookUtils)
├── preview (Preview)
├── secrets (SecretUtils)
└── widgets (WidgetsUtils)

2. Utilities in Detail with Examples

2.1 credentials — DatabricksCredentialUtils

Purpose: Manage and interact with credentials in notebooks.
Example:

# View current credential configuration
creds = dbutils.credentials.showCurrentCredentials()
print(creds)

Use Case: Debugging credential configurations when using cloud storage or API connections.

2.2 data — DataUtils (EXPERIMENTAL)

Purpose: Interact with datasets (currently experimental).
Example:

# Not widely documented; used for experimental dataset operations
dbutils.data.summarize("dbfs:/mnt/raw/data.csv")

2.3 fs — DbfsUtils

Purpose: Manipulate Databricks File System (DBFS) like a local file system.
Key Functions:

• ls(path) — List files
• cp(src, dst, recurse) — Copy files
• mv(src, dst, recurse) — Move files
• rm(path, recurse) — Remove files
• put(path, contents, overwrite) — Create file with content

Example:

# List DBFS directory
display(dbutils.fs.ls("/mnt/raw"))

# Create a file
dbutils.fs.put("/mnt/raw/sample.txt", "Hello Databricks!", overwrite=True)

# Read file
display(dbutils.fs.head("/mnt/raw/sample.txt"))

2.4 jobs — JobsUtils

Purpose: Manage and interact with job features inside notebooks.
Example:

# Exit a job with status
dbutils.jobs.taskValues.set(key="status", value="completed")

# Get task values
value = dbutils.jobs.taskValues.get(taskKey="MyTask", key="status")

Use Case: Pass values between tasks in a Databricks multi-task job.

2.5 library — LibraryUtils

Purpose: Manage session-isolated libraries.
Example:

# Install a Python package
dbutils.library.installPyPI("pandas")

# Restart Python process to apply
dbutils.library.restartPython()

Note: Library installation through dbutils is less common now; most packages are added via cluster init scripts or UI.

2.6 meta — MetaUtils (EXPERIMENTAL)

Purpose: Hook into Databricks compiler internals.
Example:

dbutils.meta.displayMeta("notebook")

Use Case: Rarely used in production — more for internal/expert debugging.

2.7 notebook — NotebookUtils

Purpose: Control flow between notebooks.
Key Functions:

• run(path, timeout_seconds, arguments) — Run another notebook
• exit(value) — Exit a notebook with a return value

Example:

# Run another notebook
result = dbutils.notebook.run("/Shared/ETL_Notebook", 60, {"date": "2025-08-10"})
print(result)

# Exit current notebook
dbutils.notebook.exit("Notebook Finished Successfully!")

2.8 preview — Preview

Purpose: Experimental/preview utilities.
Note: These may change or be deprecated.

2.9 secrets — SecretUtils

Purpose: Securely store and retrieve secrets (keys, tokens, passwords).
Example:

# Get a secret value
token = dbutils.secrets.get(scope="my-scope", key="api-token")

# List all scopes
scopes = dbutils.secrets.listScopes()
print(scopes)

Use Case: Prevent storing plain text passwords in notebooks.

2.10 widgets — WidgetsUtils

Purpose: Create and get values of input widgets for parameterized notebooks.
Key Functions:

• text(name, defaultValue, label)
• dropdown(name, defaultValue, choices, label)
• combobox(...)
• multiselect(...)
• get(name)
• remove(name)

Example:

# Create widgets
dbutils.widgets.text("param1", "default_value", "Enter parameter")
dbutils.widgets.dropdown("country", "US", ["US", "UK", "IN"], "Select Country")

# Get widget value
country_value = dbutils.widgets.get("country")
print("Country Selected:", country_value)

Use Case: Pass parameters dynamically to notebooks when running from a job.

3. Practical Workflow Example

Imagine an ETL pipeline:

1. widgets: Collects parameters for date range.
2. secrets: Retrieves API token securely.
3. fs: Reads raw files from DBFS mount.
4. notebook: Runs transformation notebook.
5. jobs: Passes processed data status to next task.

4. Best Practices

• Use secrets instead of hardcoding credentials.
• Avoid experimental utilities (meta, data) in production unless stable.
• Use widgets for parameterized jobs — improves reusability.
• Use notebook.run() for modular code separation.
• Avoid storing sensitive data in DBFS without encryption.

Mohammad Gufran Jahangir