The db_healthy signal (#131) is unit-tested (badge + field) but never exercised end-to-end: nothing makes the storage layer actually fail and confirms the dashboard reports it.
Add: a --fault-injection step that makes the dashboard's data volume read-only, waits for .db_healthy == false via a readiness predicate, then restores write access and confirms recovery.
The db_healthy signal (#131) is unit-tested (badge + field) but never exercised end-to-end: nothing makes the storage layer actually fail and confirms the dashboard reports it.
Add: a --fault-injection step that makes the dashboard's data volume read-only, waits for .db_healthy == false via a readiness predicate, then restores write access and confirms recovery.