Repository navigation
fix(recovery): retain the imported Cloudflare state package - #53
Conversation
Evaluated the real retained recovery package: authenticated both encrypted files with the independently held matching identity, verified all manifest bindings, and repeated authenticated recovery from a clean archive of the committed source. The imported state and complete desired configuration match the confirmed plan producer, run 41368905763 at source 0925d15. The saved binary plan was independently rendered with the pinned native OpenTofu version and matched its exported JSON exactly: one create, no updates or deletes. No provider write was performed. The full repository checks passed locally; current-head hosted checks also pass. This change retains the initial imported generation and updates its custody instructions. It does not activate a backend, prove remote locking or enable provider applies. Those retain their named migration gates. |
@coderabbitai review |
Action performed
|
|
No actionable comments were generated in the recent review.
|
Ready at The real package has also passed authenticated recovery from a fresh fetch/archive of this exact published head. Both retained ciphertext files match the verified producer. The custody instructions are documentation; their named state/configuration authentication path was exercised. No provider resources, credential privileges or remote backend have changed. |
Why
Cloudflare recovery currently retains the old DNS snapshot but lacks the imported infrastructure state needed to resume declarative management after an incident.
What
Retain the encrypted imported state together with its matching desired configuration and recovery instructions. This provides an independently recoverable starting point before the first infrastructure change.
Part of #35