Databricks Apps Guide¶
This page is the practical guide for running BrickflowUI inside Databricks Apps.
It is written for the actual questions people ask:
- How do I write
app.yaml? - What goes in
requirements.txt? - Why do I only see "Connecting to runtime..."?
- How do I install from GitHub instead of PyPI?
- How do I make sure the frontend bundle is included?
Minimum working setup¶
For a basic Databricks App, you usually need:
app.pyrequirements.txtapp.yaml
Correct app.yaml¶
For BrickflowUI, use Python directly.
command:
- python
- app.py
env:
- name: BRICKFLOWUI_ENV
value: production
Important:
- do not use
streamlit run app.py - do not hardcode
DATABRICKS_TOKENdirectly inapp.yaml - keep the first deployment minimal until the runtime is proven healthy
Correct requirements.txt¶
Install from PyPI:
brickflowui>=0.1.6
Install directly from GitHub:
brickflowui @ git+https://github.com/AjayAJ2000/brickflowUI.git@v0.1.6
Install a branch from GitHub:
brickflowui @ git+https://github.com/AjayAJ2000/brickflowUI.git@main
If the app only shows "Connecting to runtime..."¶
This usually means one of these:
- the frontend bundle is missing from the installed package
- the WebSocket did not connect
- the app process did not start correctly
- the browser is serving an old cached build
What to check first¶
Open browser devtools and check:
- the Network tab
- whether
/events?...gets a101WebSocket upgrade - whether the frontend asset files load successfully
Verify the installed package contains the frontend¶
import pathlib
import brickflowui
pkg = pathlib.Path(brickflowui.__file__).parent
print("index exists:", (pkg / "frontend" / "dist" / "index.html").exists())
print("assets dir exists:", (pkg / "frontend" / "dist" / "assets").exists())
Expected:
True
True