PuppetDB should only create, connect, or use its HikariCP read/write pools if the schema_migrations version matches the expected schema, otherwise the data has changed out from under it.
- Hikari pools only connect using data role - read/write pools have a connectionInitSql check on the migration version - Add periodic check on the migration version, exit 3 if it doesn't match