In order to support rollback in case of a failed PingAM upgrade in a two...

Upgrading PingAM 8.0.2 in an environment with multiple PingDS (Directory Server) instances requires a careful strategy to ensure data integrity and to allow for an immediate Rollback if the upgrade fails or introduces instability.6

According to the PingAM "Plan for Rollback" and "Upgrade in a Replicated Environment" documentation, the recommended safety measure involves isolating one of the data store nodes to act as a "time-capsule" of the pre-upgrade state. The process is as follows:

Stop Replication: Before starting the upgrade, administrators should disable or stop replication between the PingDS nodes. This prevents any schema changes or data modifications performed during the AM upgrade from being propagated to the backup node.

Isolate a Node: One PingDS instance is shut down or taken out of the load balancer rotation. This instance remains in its original, healthy state.

Perform the Upgrade: The PingAM upgrade is performed against the remaining active PingDS instance. If the upgrade involves schema updates (which is common when moving to version 8.0.2), only the active node's data is modified.

Rollback Path: If the upgrade fails, the administrator can simply shut down the "corrupted" upgraded environment and restart the isolated PingDS instance along with the original PingAM WAR file. Because replication was stopped, the isolated node still contains the original configuration and user data.

Option D is the only answer that correctly identifies this "safety-net" architecture. Option A is risky because a failed upgrade would corrupt both instances. Option C is incomplete because simply "starting up" the shutdown instance after an upgrade doesn't account for how you handle the discrepancy between the upgraded and non-upgraded nodes. The goal is to keep the shutdown instance as a valid, un-touched recovery point.