First failover

Start with a test recovery — it recovers from the last consistent replicated copy into an isolated environment without touching production or replication, and consumes no license capacity. Run a real full recovery only in an actual DR event or planned promotion.

Run a test recovery

On the recovery site's console: Configure → Protection Plans → (plan) → Actions → Test Recovery.

1. Select VMs and copy

   Pick the VMs and recover from Latest or a Point in Time checkpoint. Supply guest credentials only if scripts are configured (or for static-IP assignment on VMware targets).

2. Point at isolated networks

   In the test recovery configuration, choose sandbox networks and right-sized instances so the drill cannot touch production.

3. Tools and scripts

   Optionally install system agents and cloud packages in the recovered guests; skip plan-level recovery scripts if they mutate shared infrastructure.

4. Finish and monitor

   Jobs appear under Jobs → Recovery with the current step shown. Windows recoveries include an extra validation step.

5. Validate and clean up

   Connect via the IP on the recovery job (RDP file for Windows, SSH for Linux) and verify the application. Then remove the test instances with Actions → Cleanup Recoveries — leaving them running accrues cloud cost.

The complete procedure, including the agent matrix per platform, is in Recovery testing.

When you need the real thing

A production failover follows the same shape with two differences: it uses Actions → Recover, and it consumes license capacity. The full runbook is Failover; afterwards, failback replicates the recovered workloads back to the original source.

Related docs

• Recovery testing

• Failover

• DR drill