Skip to main content

DB Test and Preview step

Last updated on

The DB Test and Preview step validates a database changeset generated by the Author DB Change feature before the changeset is committed to your Git repository. It runs the candidate migration against the selected DB Instance in a dry-run mode and surfaces the output, errors, and a preview of the SQL that would be applied, allowing you to review and approve the change before it is merged.

note

The DB Test and Preview step is exclusively used as part of the Author DB Change workflow. It is not a general-purpose apply or validate step. Go to Configure Author DB Change to set up the full LLM authoring pipeline that uses this step.

Before you begin​

  • Harness AI enabled: Author DB Change requires Harness AI to be enabled at the account level. Go to Configure Author DB Change to enable Harness AI and configure the LLM authoring pipeline identifier.
  • LLM authoring pipeline: This step must be part of the pipeline referenced in Account Settings > Default Settings > Database DevOps > LLM Authoring Test and Commit Pipeline. The pipeline identifier must match exactly.

Add the DB Test and Preview step​

The DB Test and Preview step is added to a Custom Stage pipeline that serves as the LLM authoring pipeline.

  1. In Harness, go to Database DevOps and select your project.
  2. Create or open the pipeline designated as the LLM authoring pipeline (the identifier must match the one configured in account default settings).
  3. Add a Custom Stage with a step group.
  4. Inside the step group, select Add Step and choose DB Test and Preview.
  5. Configure the step parameters described below.
  6. Select Apply Changes and save the pipeline.

Step parameters​

DBDevOps Test and Preview Step

  1. Name - Display name for the step. Default is DBTestAndPreview_1.
  2. Timeout - Maximum duration before the step times out. Default is 20m, which accommodates the time needed to spin up a container, run the migration tool, and return output.
  3. Container Registry - Select the container registry connector that Harness uses to pull the migration container. (Use the Docker Registry)
  4. Leave the following fields as runtime inputs (<+input>). The Author DB Change workflow populates them automatically when it invokes the pipeline.
  • Select DB Schema - The DB Schema against which the generated changeset will be previewed. Set as <+input>.
  • Select DB Instance - The DB Instance that will receive the preview run. Set as <+input>.
  • Changeset - The changeset content generated by the LLM. Set as <+input>.

How it fits in the Author DB Change pipeline​

A typical LLM authoring pipeline uses two steps:

  1. DB Test and Preview — validates the generated changeset against the target DB Instance and writes the changeset files to the dbops/DBTestAndPreview_1 directory in the runner workspace.
  2. Run step — checks out the generated files, pushes them to a new branch in your Git repository, and opens a Pull Request for human review and CI validation.

Go to Configure Author DB Change for the full pipeline setup, including the Run step script for GitHub, Bitbucket, and Harness Code.

Next steps​