Deploy changes to production

Deploy local models or refresh your project from its connected Git repository

lightdash deploy pushes your local dbt or native Lightdash YAML models to Lightdash.

Use it to publish local changes, or use lightdash refresh to compile the files already committed to your connected repository. For native models, follow the Lightdash YAML deployment steps; the profiles and targets in the CLI examples below apply to dbt.

Before you deploy: A quick word of caution

lightdash deploy pushes directly to your production project.

For dbt projects, this command uses your local dbt profile (from your profiles.yml). If your default target points to a dev or staging environment, lightdash deploy will overwrite your production semantic layer with that configuration, breaking dashboards for all users.

If you’re working locally, use lightdash preview instead. It creates a temporary, isolated project that won’t affect production.

For a safer, scalable approach to deploying and previewing changes, use CI/CD. This allows you to:

• Spin up Lightdash preview environments for each PR

• Deploy production changes consistently using shared, secure credentials

• Avoid accidentally pointing to the wrong target in your local profile

Check out our guide to setting up CI/CD workflow

Option 1: Deploy via the CLI

Basic usage:

git checkout main # checkout main or master - or whatever your production branch name is where you've merged your changes
git pull
lightdash deploy # --target prod. If you use developer profiles in your dbt project, you might need this flag. See below.

By default, this will:

• Use your local dbt profile for the connection

• Push the current state of your local dbt project files (including uncommitted changes or code from your active branch) to the authenticated Lightdash project

• Trigger a re-compile and refresh of your Lightdash project

If your dbt project uses different profiles or targets for dev and prod, you can specify which profile to use:

lightdash deploy --profile prod

Be careful. If your local prod profile isn’t properly pointing to your production database, this can result in incorrect connections (e.g. accidentally deploying to a dev dataset).

Use strict compilation

Lightdash uses partial compilation by default. Field, join, and set-validation failures that can be isolated are reported as warnings so the rest of the explore stays usable; failures that cannot be isolated can still produce errors.

Add --no-partial-compilation to report the failures that partial compilation would otherwise isolate as errors instead. It doesn’t promote unrelated, non-blocking warnings — other generic compilation warnings stay non-blocking.

If your project uses physical column references such as ${TABLE}.order_id, add --validate-warehouse-columns to check supported unquoted references with warehouse queries that return no rows. It’s opt-in because it sends those queries; whether your warehouse scans or bills for them is warehouse-dependent, and the check needs warehouse credentials and the warehouse catalog. Warehouse-column errors are reported whether partial compilation is on or off.

See the CLI compilation options for the supported reference syntax and every skip condition.

For the strictest deploy, combine both flags:

lightdash deploy --no-partial-compilation --validate-warehouse-columns

Do not combine --ignore-errors with this strict CI pattern — --ignore-errors lets a deployment continue after compilation errors, which defeats the check.

Option 2: Safely refresh metadata using lightdash refresh

To pull committed changes from the configured branch of your project’s Git repository, run:

lightdash refresh

Unlike deploy, this:

• Uses the connection saved in Lightdash

• Compiles the repository using the project’s selected semantic layer format

• Leaves your local files and dbt profile out of the refresh

You can also do this directly from the Lightdash UI:

Click Refresh dbt for a dbt project or Refresh models for a native YAML project. This gives the same result as running lightdash refresh.