A recovery plan runs a series of steps that must be performed in a specific order for a given workflow such as a planned migration or reprotect. You cannot change the order or purpose of the steps, but you can insert your own steps that display messages and run commands.

Site Recovery Manager runs different recovery plan steps in different ways.

  • Some steps run during all recoveries.
  • Some steps run only during test recoveries.
  • Some steps are always skipped during test recoveries.
  • Some steps run only with stretched storage.

Understanding recovery steps, their order, and the context in which they run is important when you customize a recovery plan.

Recovery Order

When you run a recovery plan, Site Recovery Manager performs the following operations:

  1. Site Recovery Manager powers off virtual machines according to the priority that you set, with high-priority virtual machines powering off last. Site Recovery Manager skips this step when you test a recovery plan.

  2. Site Recovery Manager powers on groups of virtual machines on the recovery site according to the priority that you set. Before a priority group starts, all the virtual machines in the next-higher priority group must recover or fail to recover.

During recovery, dependencies between virtual machines within different priority groups are ignored. If dependencies exist between virtual machines in the same priority group, Site Recovery Manager first powers on the virtual machines on which other virtual machines depend.

If Site Recovery Manager can meet the virtual machine dependencies, Site Recovery Manager attempts to power on as many virtual machines in parallel as vCenter Server supports.

Recovery Plan Timeouts and Pauses

Several types of timeouts can occur during the running of recovery plan steps. Timeouts cause the plan to pause for a specified interval to allow the step time to finish.

Message steps force the plan to pause until the user acknowledges the message. Before you add a message step to a recovery plan, make sure that it is necessary. Before you test or run a recovery plan that contains message steps, make sure that a user can monitor the progress of the plan and respond to the messages as needed.

Recovery Steps for Stretched Storage

The recovery plan wizard provides an option to use cross vSphere vMotion to perform failover for all protected, powered-on virtual machines residing on stretched storage at the protected site. When this option is selected, two additional steps occur during recovery immediately before powering off the protected site virtual machines.

  • Preparing storage for VM migration. Site Recovery Manager changes the preference to the recovery site for each consistency group.
  • Migrating VMs. If the production virtual machine is not powered on, the step fails. If the production virtual machine is powered on, Site Recovery Manager initiates vSphere vMotion to migrate the virtual machine to the recovery site.
Caution: Virtual machines that are eligible for migration are not migrated if they are lower priority than non-eligible VMs, or if they have dependencies on non-eligible VMs.