Skip to content
Projects

Provisioning & lifecycle

How a project is provisioned, and the operations that manage it through its life — pause, resume, retry, admin re-provision, and delete.

Creating a project kicks off provisioning against Workers for Platforms:

  1. A proj-<slug> user worker is created in the dispatch namespace from the published worker template.
  2. Its D1 database and R2 storage are created and bound.
  3. A one-time admin credential is generated and handed to the project’s own admin.
  4. The chosen template seeds an initial collection set (if any).
  5. The instance goes live at <slug>.backlex.com.

A deploy record is written for the provision, so the first deploy shows in the project’s deploy history. Most projects are live in well under a minute.

ActionEffectMin role
PauseStops traffic + billing for the instanceadmin
ResumeRestores a paused instanceadmin
Retry provisioningRe-runs provisioning for a failed projectadmin
Re-provision adminRegenerates the project admin credentialadmin
DeleteSoft-deletes (recoverable within the retention window)admin
UpdateRename, change environmentadmin

Every operation is recorded in the audit log with the actor, target, and source IP.

A project can be created from a vertical template (blog, e-commerce, SaaS, CRM, …) or blank. The cloud passes the chosen template id through to the new instance, where the project admin applies it as a ready-to-use collection set. Blank starts with only the system schema.

Pause when you want to stop the meter but keep everything in place — resume restores it instantly. Delete when you’re done; it’s a soft delete, so it can be recovered until the retention window lapses, after which the underlying resources are torn down.