Use your own data
Run it on one computer with your own semantic layer instead of the demo, and sign in as yourself. It starts the same way as Try the demo, with docker compose.
Read your own Cube model
Footnote reads metric and dimension names from your semantic layer, and never writes to it. For a Cube model, name its folder:
CUBE_MODEL_DIR=/path/to/your/cube/model docker compose upWith your own model it starts empty, with no demo data. The sign-in page offers the sample people of the local people file. Sign in as Ravi Kapoor, the admin, and follow the setup checklist in the Inbox: add your domains, then publish the names each read makes, on one card. Your context applies to those names.
On a Mac the folder must be one Docker shares, or the service refuses to start and says so. Docker Desktop shares only the folders under Settings, Resources, File sharing, and Colima shares your home folder unless told otherwise. The Cube connector says what the model needs.
Read dbt on this computer
For a dbt project on this computer, run dbt parse in it, then name the folder that holds the semantic_manifest.json it writes, most often the project’s target:
DBT_MANIFEST_DIR=/path/to/your/dbt/project/target docker compose upThe folder is shared into the container read only, as a Cube model’s is, and the service reads it as its own dbt source, with no demo data. After the next dbt parse, use Read again on the source’s page, in Settings, Integrations.
A semantic model takes its domain from its meta, as dbt MetricFlow shows. On a Mac the folder needs the same file sharing as a Cube model’s.
Read dbt or another tool
For dbt, the service reads the semantic manifest that dbt parse writes, from a link, object storage, a Git repository or a folder: see dbt MetricFlow. An admin connects it in Settings, Integrations, as Integrations says.
Connectors lists every tool it reads. Tokens and signed links stay in a secret store in Docker’s footnote_keys volume.
Sign in as yourself
Add your name and email to the top of the local people file with roles: [administrator], then start with docker compose up --build, so the image takes the file in.
This is still the people picker of a computer of your own: anyone who reaches it can sign in as anyone. To let real people sign in, see Run in production.
Private profiles
A private profile is how you try it on your own company’s data on your own computer: a folder with the install’s settings, its keys and anything you bring in, beside a clean checkout. Keep the folder outside every git checkout, this one’s and any other, so no git add can reach it.
Start every file in it with a line that holds the marker:
FOOTNOTE-PRIVATE-PROFILE <the profile's name>In a file with comments, put the marker in a comment; in JSON Lines, in a field of the first line. The repository’s secret scan fails every pull request and every push whose commits carry the marker, so a profile file, or something made from one, cannot be merged into the repository or reach the demo.
Run the profile on its own database, with its own settings file:
FOOTNOTE_PROFILE_LABELnames it in the header and on the sign-in page, so you always see which install you are in.- Leave
SEED_DEMOoff andFOOTNOTE_AUTHatdev: the sign-in page then says nothing about the demo, which it does only forFOOTNOTE_AUTH=demoor with the demo loaded.
Next
- Choose a model provider, to answer Ask on a real model.
- Back up, restore and upgrade, before the data matters.