-
Notifications
You must be signed in to change notification settings - Fork 0
Add an executable backup and restore proof for BReg and Casework #1640
Copy link
Copy link
Closed
Labels
1.0-blockerBlocks the 1.0 release intent.Blocks the 1.0 release intent.area:bregBReg ownership.BReg ownership.area:caseworkRegistry Casework runtime, tooling, and clientsRegistry Casework runtime, tooling, and clientscriticality:p2Priority/criticality P2.Priority/criticality P2.enhancementNew feature or requestNew feature or requesttriage:needs-designNeeds a bounded scope, product decision, or acceptance design before implementation.Needs a bounded scope, product decision, or acceptance design before implementation.
Description
Activity
Metadata
Metadata
Assignees
Labels
1.0-blockerBlocks the 1.0 release intent.Blocks the 1.0 release intent.area:bregBReg ownership.BReg ownership.area:caseworkRegistry Casework runtime, tooling, and clientsRegistry Casework runtime, tooling, and clientscriticality:p2Priority/criticality P2.Priority/criticality P2.enhancementNew feature or requestNew feature or requesttriage:needs-designNeeds a bounded scope, product decision, or acceptance design before implementation.Needs a bounded scope, product decision, or acceptance design before implementation.
The runbooks in #1415 describe backup, restore, and DR. Nothing exercises
them. After #1632 a logical restore is refused until an operator runs
bregctl instance-claim adopt --acknowledge-original-retired, and Caseworkrestore is documented as replay from BReg plus audit reconciliation, but no
test takes a dump, restores it into a fresh database, adopts it, and proves
the registry serves the same committed state with its import authorities
superseded, its outbox not re-delivering, and its audit stream continuing.
Ask: a script-level proof in the shape of
products/breg/scripts/test-adopter-workflow.sh(pg_dump, restore, adopt,verify, ready, one write), plus the Casework equivalent, run in CI, and
cited from the #1415 runbook as the procedure's evidence.