Restoring a virtual machine with Instant Recovery for VMware
You can use this procedure to do either of the following:
Copy files from a virtual machine backup.
Restore the full virtual machine.
In either case, you can restore the virtual machine to its original location or to an alternate location.
Note:
To avoid host name or IP address conflicts, shut down the current virtual machine in your production environment before you start instant recovery.
To copy files while the current virtual machine is running, use a different procedure:
Table: Basic steps for VMware instant recovery
Type of recovery | Steps |
|---|---|
Copy files or troubleshoot an issue, then delete the restored virtual machine | Basic steps are these:
See the following procedure for command details. |
Restore and keep the virtual machine | Basics steps are these:
See the following procedure for command details. |
See Requirements for Instant Recovery for VMware.
The detailed procedure follows.
To restore a virtual machine with instant recovery
- On the master server, media server, or restore host, enter the nbrestorevm command.
This command is in the following location:
UNIX, Linux: /usr/openv/netbackup/bin/
Windows: install_path\NetBackup\bin\
Enter the command as follows.
To restore the VM to its original location:
nbrestorevm -vmw -ir_activate -C virtual_machine -temp_location temporary_datastore [-vmproxy VMware_access_host] -vmpo
To restore the VM to a different location:
nbrestorevm -vmw -ir_activate -C virtual_machine -temp_location temporary_datastore [-vmserver vCenter_server] -R rename_file_path [-vmproxy VMware_access_host] -vmpo
-C virtual_machine identifies the virtual machine by the name or ID that was set in the policy's Primary VM identifier attribute for the backup. On the -C option, specify the same type of identifier that was used in the policy: VM host name, VM display name, VM BIOS UUID, VM DNS name, or VM instance UUID.
The -R option provides the path to a file that contains directives for restore to a different location.
See Restoring a virtual machine to a different location with Instant Recovery for VMware.
See Instant Recovery options on the nbrestorevm command.
The nbrestorevm command mounts the virtual machine's backup image as an NFS datastore and makes the datastore accessible to the ESX host. It also creates the VM on the ESX host. It then creates a snapshot of the virtual machine.
Note:
Storage lifecycle policies (SLPs) can use Auto Image Replication to replicate a virtual machine backup image to another NetBackup domain. To restore the virtual machine from the replicated image, you must include the -vmproxy option on the command. Use the -vmproxy option to specify the backup host (access host) that is in the domain where the virtual machine was replicated. Without the -vmproxy option, nbrestorevm defaults to the backup host in the original domain and the restore fails.
The following tasks appear in the vSphere Client interface. In this example, dbl1vm5 is the virtual machine to be restored.
Note the following:
The nbrestorevm command creates a NetBackup job of type "VM Instant Recovery."
If you cancel the instant recovery job or stop all NetBackup services, the NetBackup NFS datastore is unmounted and its media server resources are released.
Caution:
The virtual machine is deleted from the ESX host.
- In vSphere Client, turn on the virtual machine.
If you included the -vmpo option on the nbrestorevm command, the virtual machine is already turned on.
- Browse and copy the virtual machine files as needed.
To copy files while the current virtual machine is running, use a different procedure.
- If you do not want to keep the restored virtual machine, enter the following:
nbrestorevm - ir_listvm
In the output, find the VM Instant Recovery ID for the restored VM.
To remove the VM from the ESX host:
nbrestorevm - ir_deactivate instant recovery ID [-force]
where instant recovery ID is the virtual machine's numeric identifier from the -ir_listvm output. -force is optional, to suppress confirmation prompts.
The VM is removed from the ESX host. If no other VM uses the NetBackup NFS datastore, NetBackup removes that datastore and releases its resources on the media server.
The following tasks appear in the vSphere Client interface. In this example, dbl1vm5 is the virtual machine to be removed and datastore_V is the temporary datastore that it used.
This step completes the VM Instant Recovery job. Skip the rest of this procedure.
Step 5 uses Storage vMotion to move the virtual machine to a production datastore. If vMotion is already in progress for this virtual machine, you should cancel the vMotion job before you enter - ir_deactivate. Otherwise, vMotion moves the virtual machine to a production datastore where - ir_deactivate cannot remove it.
- To keep the restored virtual machine:
In vSphere Client, right-click on the restored virtual machine and select Migrate. Select the migration type and the destination.
Note: For the destination, select a permanent (production) location for the virtual machine. Do not select the temporary datastore that was used for the instant restore.
Storage vMotion transfers the virtual machine data files from the NetBackup NFS datastore to the datastore that you selected.
Note: You should migrate no more than one restored virtual machine at a time per media server.
- After the migration is complete, use vSphere Client to merge or consolidate the virtual machine's redo log (or snapshot) files manually. See your VMware documentation for details.
When the migration to the production datastore is complete, use the following steps to unmount the NFS datastore and release its resources.
- Enter the following:
nbrestorevm - ir_listvm
In the -ir_listvm output, find the VM Instant Recovery ID for the restored VM.
- When the data migration is complete, enter the following:
nbrestorevm - ir_done instant recovery ID
where instant recovery ID is the virtual machine's numeric identifier from the -ir_listvm output.
The -ir_done option completes the VM Instant Recovery job. It also removes the NetBackup NFS datastore if no other VM uses it. When the datastore is removed, its resources are released on the media server.
