cancel
Showing results for 
Search instead for 
Did you mean: 

Netbackup Full system restore Win 2012 VM with RDM Luns

Manumohan
Level 4

Hello Veritas,

Netbackup master : Win 2012 R2 , NBU 7.7.1

Netbackup Media: Win 2012 R2, NBU 7.7.1

Client: Win 2012 R2, NBU 7.7.1.

**********************Enquiry******************

We are planing for a DR site for the production site recovery. Backup method in production site is installed NBU Cleint agent on VMs (Becaue of RDM). Planing to do a restore to DR site where we have Virtual disk (vmdk). Clients are Microsoft clusters.Does this work for a Full System restore. pleae let know if there any challenges.

Production site info:

Server Type: VM , VMware

OS: Win 2012 R2

Disk Type: RDM, VM is living on this disk.

DR Site info:

Server Type: VM , VMware

OS: Win 2012 R2

Disk Type: vmdk (virtual Disk), this is planing does NBU restore work?

11 REPLIES 11

sdo
Moderator
Moderator
Partner    VIP    Certified

You seem to be confusing "Microsoft cluster functionality" with "VMware VADP functionality".

I don't think this work:

Production site info:  Disk Type: RDM, VM is living on this disk.

DR Site info:  Disk Type: vmdk (virtual Disk).

...not because of NetBackup, but because of Microsoft+VMware.   You haven't specified why the source cluster has RDMs, and so I assume the source cluster requires RDMs because the LUN behind the RDM is actually a cluster resource volume/disk/LUN (even if it just a quorum LUN)... and so, the use of an RDM is integral to the cluster, and therefore one or more RDMs will also be required at the DR site.

Hi Sudo,

Windows solution team is tyring to do in the cost effective way for the DR setup. Planing cluster in a box, have attached a picture.

 

For VM nodes running Windows cluster on single ESXi hosts, the supported shared storage type are:

 

  1. Virtual disks
  2. RDM with virtual compatibility mode

 

Option (a) is recommended by VMware.

Option (b) is supported but will not give the best result. (eg, potential performance issues etc ..)

**************************************************************************************************************

What if we have same RDM disk in the DR site, the Full system restore will work ? Please suggest.

Hi Sudo,

I missed to update, the production servers are VMs Clusters in physical compatibility mode. therfore it was configured using RDMs.

sdo
Moderator
Moderator
Partner    VIP    Certified

Well, I think I can see why you are asking...

...if I Google for:

https://www.google.co.uk/?gws_rd=ssl#q=vmware%20virtual%20disk%20for%20window%20cluster

...then I get this:

Microsoft Clustering on VMware vSphere: Guidelines for supported configurations (1037959)
https://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=10379...

...which links to:
Microsoft Windows Server Failover Clustering on VMware vSphere 6.x: Guidelines for supported configurations (2147661)
https://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=21476...

...and I found this:
Say Hello to vMotion-compatible Shared-Disks Windows Clustering on vSphere
http://blogs.vmware.com/apps/2015/02/say-hello-vmotion-compatible-shared-disks-windows-clustering-vs...

...

...but I'm still at a loss as to what you are aiming at?  Are you looking to convert an "RDM disk" based virtual cluster to instead become (via a restore process) a "virtual disk" based virtual cluster?  This might be possible using a manual NetBackup BMR process whereby you pre-provision the virtual disks.  But, I don't know if it is possible.  NetBackup Client for plain file system backups is essentially unaware of the underlying "LUN strategy" of VMware.  So, if you can backup as a volume drive letter, then you should be able to restore as a volume drive letter.  So, I'd envisage a multi-staged restore process of:

i) create empty VM for cluster node A

ii) BMR boot recover cluster node A - restore local disks only

iii) create empty VM for cluster node B

iv) BMR boot recover cluster node B - restore local disks only

v) re-form, re-configure cluster basic cluster with cluster resource group(s), and empty cluster volume(s) on the VMware LUN strategy of your choice.

vi) restore non-quorum (assuming the quorum volume really was an "empty" volume) cluster volumes

vii) attempt to start cluster aware applications

viii) test cluster failover and failback.

I think you would definitely need to have well configured "cluster tolerant" backup policy configuration already in place before attempting the above.

As far as I can figure out what you intend to do, you wish to place an exact copy of your HA physical server onto a virtualized environment at your DR site, correct?

IF this is the case, a possible solution would be to use Symantec System Recovery to create a vm with all disks configured as vmdk's, or use vmware's converter solution for instance?

sdo
Moderator
Moderator
Partner    VIP    Certified

Just found this tech note:

How to recover a cluster environment using Bare Metal Restore (BMR)
http://www.veritas.com/docs/000070854

Hi Sudo,

We have same infracture in production and DR site. VMs are backed up as traditionla method (Instaled NBU agent on VMs) because of RDM disk.

Is there any challenge if i do a restoration based on the below tech note. Please suggest..

https://www.veritas.com/support/en_US/article.000081333

Hi Sdo,

We have same infrastructure in production and DR site. VMs are backed up as traditional method (Instaled NBU agent on VMs) because of RDM disk. Is there any challenge if i do a Full System restoration based on the below tech note. Please suggest..

https://www.veritas.com/support/en_US/article.000081333

sdo
Moderator
Moderator
Partner    VIP    Certified

My personal opinion is.... I really really dislike that tech note:

  • Use NetBackup to perform a restore for a total and complete recovery of Windows client(s) that include C: (and other system) drive(s), Shadow Copy Component, and/or System_State from a reliable full MS-Windows backup without ...
    http://www.veritas.com/docs/000081333

...it has been re-written so many times over the years, and tweaked with this exception, and that exception, and don't forget this, and don't forget that, but don't do this if that is true, yet do do that if this is true.

So, my advice, forget you ever saw that tech note, and use BMR instead - you'll have a much better chance of success with BMR.

Hi Sdo,

Thanks for the information. I will try with NBU BMR option ( VM Convertion Clients).

Hi Sdo,

Thank you for your suggestion to go with BMR for Full system recovery of VMs. Aroud 42 servers i was able to bring UP in DR site. The BMR process is bit heavy but restore is perfect.

After recovering C and system state and other system drive the server came UP as production. The server joined to cluster. The cluster disks(data drives RDM) were restored seperatly after full system restore . This is just becaue the cluster DIKS are GPT volume which BMR will not support. After assigning original GPT ID to cluster disk the drives came online. Original GPT ID we were able to get from registry.