How can I safely roll back a FaunaDB cluster after a failed schema migration upgrade?

After applying a FaunaDB schema migration as part of a cluster upgrade, the migration job reported a failure and stopped. Some nodes appear to have received the new schema version while others remain at the previous version, leaving the cluster in a mixed state. The goal is to determine whether the cluster can be brought back to a consistent condition without losing data, and what verification steps are appropriate before deciding on a recovery path.

What checks can I run to confirm that all nodes agree on the current schema version? Is it safe to re‑run the migration after correcting the underlying issue, or should I first restore the cluster from the latest snapshot? If a snapshot restore is required, what precautions ensure that no committed transactions are lost?