Terraform Practice

Write a minimal configuration with pinned Terraform and AWS provider versions pointed at the local endpoint, and drive it through init, plan, apply, and destroy with a storage bucket as the managed resource - proving the plan previews before creation, a second plan is clean, and destroy leaves an empty state.

Constrain the AWS provider with a pessimistic version constraint and create the dependency lock file, then demonstrate that re-initialization respects the locked version and hashes, a constraint excluding the locked version is refused, and an explicit upgrade moves the selection only within the constraint.

Modify an existing configuration so one plan contains a create, an in-place update, and a replacement; read the plan to classify each action, apply the approved saved plan exactly, and prove that a stale saved plan is refused after the state moves on.

Extract hardcoded values into typed variables - string, number, bool, list, and object - with defaults and validation rules that reject bad values with defined messages, proving that mistyped and out-of-range inputs fail at plan time while defaults and supplied values flow into the resource verbatim.

Set one variable through defaults, an environment variable, an auto-loaded tfvars file, an explicitly passed tfvars file, and a command-line flag, capturing a plan at each rung to prove the precedence order - with the winning command-line value observable in the applied resource.

Centralize a derived naming expression in a local consumed by multiple resources, and expose described outputs including a sensitive one - reading them back as a plain listing with the sensitive value redacted, a single raw value, and machine-readable JSON where the sensitive value is present.