An administrator may notice that when performing a Live Failover, or a Failover Test, VMs start up in a different manner from a Move operation. For example, a Windows VM may show an error due to an “Unexpected Shutdown”. This article explains the reasons for this behavior.
When configuring a Live Failover or a Failover Test operation to be executed, a specific checkpoint must be specified to which to recover the virtual machines. These checkpoints are normally created when the protected VMs are running, and as such, the recovered VMs will tend to come online in a state consistent with an unexpected shutdown.
This can result in the following types of behavior:
- Event logs indicate an Unexpected Shutdown
- The VM comes online in a System Recovery Mode state
- The VM comes online and runs chkdsk
These behaviors are to be expected, however should not be of concern. That said, if it is desired to avoid these behaviors when running a Live Failover or a Failover Test, it is recommended to proceed as follows:
- Perform a shutdown of the guest OS for the relevant VMs.
- Manually insert a checkpoint for the relevant VPG.
- Proceed with the operation, selecting the manually inserted checkpoint to which to recover.
These steps avoid potential data-loss, as the VMs are shutdown and the new checkpoint is added after all I/Os have been written to disk. This functionality is similar to that of the Move operation, which automatically shuts down the protected VMs, creates a new checkpoint, and then proceeds to bring the application online on the recovery site.