Postgres Logical Replication Failover Runbook Builder

You are a senior database reliability engineer. Build a failover runbook for a Postgres logical replication setup.

Environment: [Postgres version, cloud provider or self-hosted, managed service constraints]
Topology: [publisher, subscribers, publications, subscriptions, replication slots, downstream consumers]
Data scope: [tables, schemas, sequences, large objects, partitioned tables, excluded tables]
Failure scenario: [publisher outage, subscriber lag, slot bloat, schema drift, network partition, failed migration]
RPO/RTO target: [acceptable data loss, recovery time target]
Monitoring available: [pg_stat_subscription, slot lag, WAL disk alerts, application metrics, logs]
Access level: [superuser, managed DB admin, limited role, read-only observer]
Traffic cutover method: [DNS, app config, feature flag, maintenance window, read-only mode]
Validation needs: [row counts, checksums, sequence values, application smoke tests, audit queries]
Rollback constraints: [old primary availability, bidirectional writes risk, backups, restore point]

Produce:
1. Preconditions and assumptions to verify before any failover.
2. Detection checklist and severity classification.
3. Step-by-step failover plan with commands or pseudo-commands.
4. Replication slot, WAL retention, sequence, and schema checks.
5. Data validation queries before and after cutover.
6. Application cutover and communication checklist.
7. Rollback or roll-forward decision tree.
8. Post-incident cleanup tasks and prevention work.

Call out where exact SQL depends on the managed provider or schema. Do not invent credentials or run destructive SQL without an explicit confirmation step.