Skip to main content

Head Metadata

You should always manage the lifecycle of database objects, like tables and views, from the Coalesce application. Coalesce needs to have full control over the objects it creates and manages in your data platform.

When you drop a table directly in your data platform, you can cause deployment failures and metadata mismatches.

What Happens When You Drop a Table?

If you drop a table in data platform that is managed by a Node in Coalesce, you may see the following issues.

Deployment Failures

During a deployment, Coalesce may try to access a managed object that it expects to exist. If you dropped that object outside of Coalesce, the deployment will fail because it can't find it.

Automatic Table Recreation

When Coalesce runs a deployment, it compares its metadata with the actual state of your data platform environment. If it detects that a managed table is missing, it will try to recreate it based on the Node definition.

Nodes Remain in the Coalesce UI

Even if you drop the underlying table in the data platform, the Node will still appear in your Workspace graph and search results. This can cause confusion because the UI shows a Node that doesn't point to a real table.

Metadata Mismatches

Dropping tables outside of Coalesce creates a mismatch between the Coalesce metadata and your data platform environment. This can lead to unexpected errors and behavior during builds and deployments.

Best Practice: Deleting Nodes in Coalesce

The correct way to drop a table is to delete the corresponding Node from your graph in Coalesce.

  • Open your Workspace.
  • Find the Node that represents the table you want to drop.
  • Right-click the Node and select Delete Node.
  • Commit and deploy this change.

This process ensures that both the table in the data platform and the metadata in Coalesce are removed correctly and stay in sync.