Verify a backfill: decrypt a sample of rows and compare them to the plaintext column they were encrypted from.
Run this between backfill and cut over. Cutting over is the irreversible step of the migration, so it should follow a clean verify — after it, the plaintext column is no longer written and the only copy of the data is the ciphertext.
This task writes nothing and mints no key material beyond what decryption needs.
Usage
mix ash_vault.verify MyApp.Accounts.User email --tenant acme
mix ash_vault.verify MyApp.Accounts.User email --from legacy_email --sample 0
mix ash_vault.verify MyApp.Accounts.User email --all-tenants MyApp.Accounts.list_tenant_ids/0Options
--from ATTR plaintext column to compare against (default: `backfill_from`)
--tenant TENANT required when the key scope is `:tenant`
--all-tenants MFA zero-arity `Module.function/0` listing tenants
--domain MODULE Ash domain (inferred from `:ash_domains` when omitted)
--batch-size N rows per read (default 500)
--sample N rows to decrypt; `0` checks every row. Default: 10% of the
table or 1000 rows, whichever is smaller, always including the
first and the last row.Exit status
Non-zero on the first tenant with any mismatch — a row whose ciphertext is NULL, whose plaintext differs, or which fails to decrypt at all. The summary names the offending primary keys.
ash_vault.verify=result resource=MyApp.Accounts.User field=email rows=2000 checked=200 mismatches=0
Verified 200 of 2000 row(s) of MyApp.Accounts.User.email: no mismatches.