Troubleshooting

Cause: Lighthouse couldn't reach your warehouse with the credentials provided.

Fix: Double-check the credentials — especially account identifier format for Snowflake and service account JSON for BigQuery. Make sure the database user exists and has the correct permissions. If your warehouse requires IP-based allowlisting, contact support@lighthouseai.co for current IP ranges.

Cause: The values you entered don't match the session defaults Snowflake returns for that user.

Fix: Run SELECT CURRENT_WAREHOUSE(), CURRENT_DATABASE(), CURRENT_SCHEMA() as the Snowflake user you're connecting with. Make sure the fields in Lighthouse exactly match these values. If the user's default role doesn't have the right defaults, specify the role field.

Cause: The service account doesn't have permission to access the specified dataset.

Fix: The service account needs three roles: BigQuery Data Viewer and BigQuery Job User at the project level, plus BigQuery Metadata Viewer on the specific dataset (dataset-level, not just project-level). Also confirm the dataset name is spelled correctly — BigQuery dataset names are case-sensitive.

Cause: You tried to save a data source without running the connection test first.

Fix: Click Test Connection and wait for a green success state before saving.

Cause: One or more required sections in the metric configuration are incomplete or have validation errors.

Fix: Look for red highlights in the metric builder — each incomplete section will be flagged. Common issues: no aggregation column selected, time window without a datetime column, or threshold conditions that are incomplete.

Cause: The metric's schedule is set to run more frequently than your plan allows.

Fix: Change the metric schedule to at least the minimum interval for your plan (4 hours on Free, 1 hour on Professional, 5 minutes on Enterprise). Or upgrade your plan.

Cause: You've reached the maximum number of active metrics for your plan.

Fix: Pause or delete metrics you're no longer actively using. Or upgrade to a plan with a higher limit.

Cause: A column referenced in the metric definition was removed from the dataset, or its role was changed so it's no longer valid.

Fix: Open the metric builder and re-select a valid column. The invalid column will be flagged with a warning. Update the dataset column roles if needed.

Cause: The metric query returned the wrong number of rows or is missing expected columns.

Fix: For SQL metrics: verify your query returns exactly one row per segment combination. For Business metrics: check that the time window and filters are producing the expected result by running a preview query.

Cause: The metric query returned duplicate segment keys — two or more rows with the same combination of segment column values.

Fix: For SQL metrics: ensure your GROUP BY clause correctly produces unique segment combinations. For Business metrics: check that the segment columns you selected have unique values per time window.

Cause: The warehouse returned an error when Lighthouse ran the metric query.

Fix: Check your warehouse query logs for the specific SQL error. Common causes: table or column was renamed or dropped, the warehouse user lost permissions, or the warehouse was paused (Snowflake auto-suspend).

Cause: A metric tried to run but its parent data source or dataset is paused.

Fix: Resume the data source or dataset from the respective settings page. The metric will resume on its next scheduled run.

Cause: The Slack channel selected for this metric no longer exists or the Lighthouse bot was removed from it.

Fix: Re-select a valid channel in the metric's notification settings. If using a private channel, ensure the Lighthouse bot is still a member of that channel (invite it in Slack with /invite @Lighthouse).

Cause: The Slack OAuth token for your workspace has expired or been revoked.

Fix: Go to communication sources settings and reconnect your Slack workspace via OAuth.