VM with many RDM needs to be backed up

figure out a way to exclude them. Avamar excludes them for me automatically. These devices need to stay as RDM, physical compatability mode.

i don't have experience with Networker, you might want to ask in Networker forum.

gatekeepers typically are 3mb devices, you don't need to put a file system on them to be used by symcli.

I assume there are no actual vital files on these devices, do they even have a filesystem?

You can just go ahead and configure an image backup as per normal procedure with save set *FULL*. NetWorker will automatically skip them since it is just as you say not possible to back them up with the VADP interface, assuming you use VMware as your hypervisor. You won't even get any error or warnings, but if you look in the log you will probably see information messages that the devices were skipped.

Hey cmorrall,

Thanks for your reply.

Could you please let me know if this is documented in any whitepaper or KB?

Need to convince the customer with the same.

Regards

Harshal

Hi Cmorrall,

This is the error that is received when tried to take Full save set from Networker.

Is there a manual way to exclude these RDMs?

Thanks

Hi Dynamox,

This may sound silly, but I wanted know, whether there is really any need to backup a server that has these RDM devices connected. As the devices are directly on VMAX and will be protected on array level.

My second question is, Unisphere is just a interface to the storage array, or does the server on which its installed holds some kind of database that will hold the changes performed such as configuration changes or LUN creation/deletion etc.

The customer need to be very much sure before excluding this server from backup.

This server is solely used to access VMAX through Unisphere.

Thanks

Harshal

There could be customer files on that server (device groups for example) that are not located on the storage array but local to that server. So my recommendation is to backup it up.

Hi Dynamox,

Please let me know, if the OS corruption on this server will have any effect on the configuration changes performed through unisphere (like configuration changes or LUN creation/deletion etc.)

I just want to be sure that image backup may be skipped for this server and normal backup can be applied.

Customer is concerned whether OS corruption will have any effect on configuration changes?

Thanks for your help..

If OS is down, Unisphere is down so you can't manage your storage array.

bottom line, box should be physical, not virtual to avoid creating dependency on ESX for SAN management.

that's your choice, i have been using VMs as my management boxes since ESX 4.0  (for symcli, ECC, SMAS, UVMAX). I place my management VMs on different storage arrays so if one array is having issues i still have multiple management hosts that can be used.

harshgw2k3,

if the only thing your customer uses to manage VMAX is Unisphere,  have their CE enable Unisphere for VMAX on the VMAX storage processor ..then you don't have to run on your own server at all. You just need to understand that you can't login to SP and import device groups and things like that, but for general use that could be used as a backup option to your "installed" instance or the only solution period.

RDM backup is not possible in VMware Protection Environment (VBA , VADP) . Even if you succeed in backing up those RDMs (or for some reason the console shows backup as successful), you will not be able to recover them.

RDM disks are not supported for VADP/VBA backups, because VM snapshots cannot be applied to such VMs. During NetWorker backup of a VM, no RDM related information is backed up, and no RDM disks/data are restored upon VM recovery. If RDM disks are required, they must be reattached after the recovery.

Note If reattaching RDM disks after recovery, make note of all LUNs that are zoned to the protected VMs.

Isn't it that this depends on type of RDM?  vRDMs should work just fine with snapshots (pRDMs not).