How-To

The How-To pocket is a set of task recipes: short, lookup-friendly procedures for the operational jobs that come up once your fleet is modeled and live. Each page is structured the same way — what you're doing → the steps → the result — so you can scan to the recipe you need and copy the commands.

These are procedures, not concepts. If you want the mental model first, read Learn; if you want the field-by-field schema, read the Catalog Schema Reference. This pocket sits between them: it assumes you know what an entity, a cell, a relationship, and an Action are, and shows you how to do a thing with them.

:::info Two rules that shape every recipe here

  • Terrantula never runs terraform apply. Actions open pull requests against your repo (or dispatch runs to your runner); your CI applies. Re-imports and rescans are observational — they update the graph to match your source; they never provision.
  • The graph is a read-only projection of your TF-derived state. You don't fix a discrepancy by editing entities in the UI — you fix the source (re-import for observed state) or open an Action's PR (for intended changes). :::

Recipes

RecipeUse it when…
Cross-source driftThe graph and the external source it was imported from have diverged, and you want to detect and reconcile it.
Deprovision a tenantYou need to retire a tenant cleanly — the cascade order, the optional grace period, and the PR your CI applies.
Import sourcesYou import from more than one Terraform state or Atmos stack set into one project and need each source's entities addressable without colliding.
Multi-stack golden pathYou're onboarding a tenant that spans several Terraform stacks and want one command to open the right PRs in the right order.
Deploy on KubernetesYou're self-hosting the backend and want a cluster deployment.
ON THIS PAGE