A company has implemented a stretch cluster, active/passive configuration, between two sites that are 100km apart in conjunction with synchronous data replication. They want to perform an upgrade to t

Failing over the cluster alias before the application is ready on the secondary site could lead to service disruption or errors, and failover storage is usually handled by the cluster's replication.

To ensure no data loss during a planned failover in a synchronously replicated active/passive stretch cluster, first stopping services on the local (primary) host ensures all pending transactions are flushed and no new data is written. Next, failing over the application through the cluster management software brings the application online at the secondary site using the already synchronized data. Finally, failing over the cluster alias (e.g., virtual IP or hostname) ensures clients seamlessly redirect to the application now running on the secondary site without manual intervention.

Failing over storage might be an internal replication step, but it's not the primary user-facing action for application failover, and failing over the cluster alias before the application is active could cause issues.

Failing over the cluster alias before the application is prepared and active on the secondary site means clients would be directed to an unresponsive or partially ready application, leading to disruption.