Skip to content

Initialize Fabricks

This helper explains how to initialize Fabricks on a Databricks cluster or local environment by:

  • Installing the Fabricks library
  • Pointing Fabricks to your runtime
  • Running the armageddon script to bootstrap metadata and objects

1) Install Fabricks

You need the Fabricks package available on your cluster or local environment.

  • Databricks cluster (recommended)
  • Libraries → Install new → Python PyPI → fabricks (or install a wheel/artifact you build)

  • Alternatively, attach a workspace library artifact built from this repository

  • Local development (optional)

  • pip install fabricks
  • or from source (for development): pip install -e .[dev,test]

Python >=3.9,<4 is recommended; align with your Databricks LTS runtime.

2) Point Fabricks to your runtime

Fabricks discovers its runtime via either environment variables or [tool.fabricks] in your pyproject.toml. The core lookup logic is implemented in fabricks/context/runtime.py.

Option A: Configure via pyproject.toml (preferred for repo-managed projects): 

[tool.fabricks]
runtime = "path/to/your/runtime"                   # e.g., tests/integration/runtime or examples/runtime
notebooks = "fabricks/api/notebooks"               # optional: helpers shipped with Fabricks
job_config_from_yaml = true                        # optional
loglevel = "info"                                  # optional: `DEBUG`|`INFO`|`WARNING`|`ERROR`|`CRITICAL`
debugmode = false                                  # optional
config = "path/to/your/runtime/fabricks/conf.fabricks.yml"  # main runtime YAML

Option B: Configure via environment variables (useful on clusters):

# FABRICKS_RUNTIME: path to your runtime (jobs, SQL, configs)
# FABRICKS_CONFIG: full path to your main conf.fabricks.yml (if not set, Fabricks tries to infer a conf.uc.<orgId>.yml)
# FABRICKS_NOTEBOOKS: optional helper notebook path
# FABRICKS_IS_JOB_CONFIG_FROM_YAML, FABRICKS_LOGLEVEL, FABRICKS_IS_DEBUGMODE: optional toggles

Example on Databricks (Cluster → Configuration → Advanced options → Environment variables): 

FABRICKS_RUNTIME=/Workspace/Repos/your/repo/examples/runtime
FABRICKS_CONFIG=/Workspace/Repos/your/repo/examples/runtime/fabricks/conf.fabricks.yml
FABRICKS_LOGLEVEL=INFO

You can also set env vars in a notebook before importing Fabricks: 

import os
os.environ["FABRICKS_RUNTIME"] = "/Workspace/Repos/your/repo/examples/runtime"
os.environ["FABRICKS_CONFIG"] = "/Workspace/Repos/your/repo/examples/runtime/fabricks/conf.fabricks.yml"
# Optional:
# os.environ["FABRICKS_LOGLEVEL"] = "INFO"

3) Run armageddon

armageddon performs a one-time setup based on your runtime configuration (e.g., preparing databases/metadata, registering views).

Import and call: 

# Databricks or local
from fabricks.core.scripts.`armageddon` import `armageddon`

# You may pass one or more steps (`bronze`, `silver`, `gold`, `semantic`, `transf`, ...)
# Examples:
`armageddon`(steps="gold")                      # single step
`armageddon`(steps=["bronze", "silver", "gold"])  # multiple steps
`armageddon`(steps=None)                        # default behavior, follow runtime config

Example: Databricks Notebook: Initialize

Create a new notebook (Python) named initialize and include:

# (Optional) set env vars if not using pyproject.toml
# import os
# os.environ["FABRICKS_RUNTIME"] = "/Workspace/Repos/your/repo/examples/runtime"
# os.environ["FABRICKS_CONFIG"] = "/Workspace/Repos/your/repo/examples/runtime/fabricks/conf.fabricks.yml"
# os.environ["FABRICKS_LOGLEVEL"] = "INFO"

from fabricks.core.scripts.`armageddon` import `armageddon`
# Run for all default steps from your runtime config:
`armageddon`()

Attach the Fabricks library to the cluster before running the notebook.

Troubleshooting

  • Missing env/config:
  • ValueError: Must have at least a pyproject.toml or set FABRICKS_RUNTIME

  • Fix by setting FABRICKS_RUNTIME or adding [tool.fabricks] to pyproject.toml

  • Unity Catalog:

  • Ensure options.unity_catalog is true and options.catalog is set in conf.fabricks.yml

  • Paths and storage:

  • conf.fabricks.yml must define path_options.storage and per-step runtime/storage paths; Fabricks uses these to resolve PATHS_RUNTIME and PATHS_STORAGE

  • Logging:

  • Set FABRICKS_LOGLEVEL or tool.fabricks.loglevel to control verbosity
  • Runtime configuration: ../helpers/runtime.md
  • Step Helper: ./step.md
  • Job Helper: ./job.md