This article aims to provide a simple and detailed instructions on configuration NMM 18 to backup Hyper-V Server 2016 VMs.
Hyper-V 2016 introduced a feature called RCT (Resilient Change Tracking). It’s a native change tracking system build in Hyper-V. It keeps track of the change blocks in a virtual hard disk. It keeps in memory bitmap and on disk bitmap. This helps keep track of changes even through power failures, when a memory only bit map would be lost.
RCT backups do not use VSS frame work. It’s more scalable and reliable than VSS. NMM 18.1 introduced support for RCT. When backing Hyper-V Server 2016 VMs with NMM 18.1, it’s recommended to use RCT. Support to backup using VSS is still available.
Refer to the Networker compatibility guide for the latest compatibility information between NMM and Hyper-V at https://elabnavigator.emc.com/eln/modernHomeAutomatedTiles?page=NetWorker
Pre-requisites for backups using RCT:
1. NMM 18.1 or higher. NMM 9.x does not support RCT.
2. Hyper-V server 2016. Hyper-V server 2012 does not support RCT.
3. VM configuration version higher than 6.2
I have used NMM 18.1 on Hyper-V Server 2016 with NetWorker server 9.2.1.5 for this article.
Below are the steps:
1. The first step is to install NetWorker client software and NMM on each of the Hyper-V nodes. No instructions are provided for this step as its a pretty simple process.
2. Configure the Client resources for the backup. Start the Client Wizard from the NMC GUI.
3. To backup Virtual Machines on a HyperV Cluster residing on CSV volume or SMB Share, use the Cluster name with the client configuration wizard. For standalone Hyper-V server use the Hyper-V server hostname or FQDN.
4. Select ‘Hyper-V Server’ under applications:
5. On the next page, keep the defaults
6. Select the Virtual Machines to backup. To exclude any Virtual Machine from the backup deselect the Virtual machine and select ‘Use Exclude Component List’ at the bottom of the screen.
7. Specify the backup options.
The account used for backups in ‘Remote Username’ should have the following group membership.
a. Should be a domain user. i.e member of the ‘Domain users’ AD group.
b. Member of the local ‘Administrators’ group on each HyperV host in the cluster
c. Have FULL access to the HyperV cluster. Run the following Powershell command to provide this access:
Grant-clusteraccess –cluster <clustername> -user <username> -FULL.
Example
Grant-ClusterAccess -cluster thor -user brontenmmadmin –FULL
To verify if the account has FULL cluster access, run the following command:
Get-clusteraccess
Note: While the above permissions are sufficient for backups, when restoring SMB VMs to the original location, additional permissions are required to perform restore to original VM location. The above permissions do not provide admin level read rights to the SMB share. There are 2 ways to workaround this:
a. When performing a restore login to the Hyper-V host with a account that has FULL rights to the SMB-share.
b. If performing the restore with the same account as backup account, first delete/rename the original VM folder or provide this user account all rights to the SMB share.
8. Review summary and click ‘Create’:
The client wizard will proceed to create client resource for the Hyper-V cluster and a client resource for every node in the cluster.
9. Once the client resources are created proceed to create a Group, Workflow in an existing policy or create a new Policy.
a. Create a group and add the cluster client to it. Note only the cluster client will be used for the backup.
b. Create a new workflow and add the Hyper-V group to this workflow
c. Create a New ‘Action’, by clicking the ‘Add’ button in the above window. Specify the schedule and backup levels (Only level ‘FULL’ and ‘Incr’ are supported)
d. Specify storage node, pool, retention as appropriate:
e. Set the ‘Retries’ to ‘0’ and the ‘Inactivity timeout’ to ‘0’
f. Review summary and click ‘Configure’ (not shown in the screen shot)
10. Add the ‘backup user’ to the Networkers ‘Operators’ group. From NMC GUI, click ‘Server’. Select ‘User Groups’ and then do properties on the ‘Operators’ group. In the ‘Users’ box add the entries similar to below:
user=nmmadmin,host=hypv2016n1
user=nmmadmin,host=hypv2016n2
There is one entry for each node in the cluster.
This completes the backup configuration for a HyperV 2016 Federated backup using RCT. Note when using RCT backups, SMB and CSV VMs can be backed up together, unlike when using VSS SMB and CSV VMs had to be backed up separately.
Also for SMB VMs, as RCT does not use VSS, there is no need to install the “File Server VSS agent service” on the File Server. Also the backup user does not need any additional rights to backup VMs on SMB share.
Save sets created for a FULL level backup
HyperVPool.003 Data Domain thor.bronte.local 1/22/2019 3:08:11 PM 15 GB 2370271269 cb full APPLICATIONS:Microsoft Hyper-VEx1E9D2B2FD-B4B7-4A4A-87DC-0DD89D89FAC6
HyperVPool.003 Data Domain thor.bronte.local 1/22/2019 3:08:12 PM 58 MB 2353494053 cb full APPLICATIONS:Microsoft Hyper-VEx1ConfigFiles
HyperVPool.003 Data Domain thor.bronte.local 1/22/2019 3:13:46 PM 8 KB 2303162745 cb full APPLICATIONS:Microsoft Hyper-VEx1
For each VM following save sets are created:
- One save set for each disk in the VM. This is represented by a long hex number in the save set name. (e.g APPLICATIONS:Microsoft Hyper-VEx1E9D2B2FD-B4B7-4A4A-87DC-0DD89D89FAC6)
- One save set for ‘ConfigFiles’
- One metadata save set. This save set ends with the VM name (e.g APPLICATIONS:Microsoft Hyper-VEx1)
For a VM with one disk, there are 3 save sets. For a VM with 2 disks there are 4 save sets.
Save sets created for a Incremental backup
HyperVPool.003 Data Domain thor.bronte.local 1/22/2019 9:04:36 PM 15 GB 2068302767 cb full APPLICATIONS:Microsoft Hyper-VEx1E9D2B2FD-B4B7-4A4A-87DC-0DD89D89FAC6
HyperVPool.003 Data Domain thor.bronte.local 1/22/2019 9:04:37 PM 58 MB 2051525551 cb full APPLICATIONS:Microsoft Hyper-VEx1ConfigFiles
HyperVPool.003 Data Domain thor.bronte.local 1/22/2019 9:05:52 PM 8 KB 2001193984 cb incr APPLICATIONS:Microsoft Hyper-VEx1
Since a synthetic FULL is performed on Data Domain, all save sets are registered at level FULL in media DB, except for the metadata save set (From the above example, its “APPLICATIONS:Microsoft Hyper-VEx1”), that reflects the actual backup level for that backup.
