SAN -> Virtual Appliance -> Net backup

[JohnG]

Let's say I have a single job that backs up all my VMs. The VMs are all in one vSphere cluster and this vSphere cluster has both iSCSI and NFS datastores. Both types of datastores have running VMs. Veeam backup server v6 is running in a VM. Veeam proxy (same as Veaam backup server) is set to 'Automatic' with network failback.

When the job runs, all VMs that are in iSCSI datastores are backed up through SAN transport mode. This works great, no problem.
When the job encounters a VM that is on an NFS datastore, the VM is backed up in NBD (Network transport mode). This is no good.

I've toyed with adding another proxy and making this proxy run only in virtual appliance mode. But it seems that Veeam picks a proxy at the beginning of the job, not on a per-VM basis so Veeam picks the 'Automatic' proxy and so fails back to NBD when it encounters an NFS VM.

I am assuming this is working as designed. If not, how can I make Veeam try SAN backup mode, then Virtual Appliance and as a last resort, Network mode? It would be great if I could specify the order in which the different transport modes are used. Can Veeam make this a configurable option instead of an either/or proposition?

No, we do not want to use Virtual Appliance mode exclusively.

Am I going to have to setup different jobs for iSCSI datastores and for NFS datastores?

[foggy]

Hello! Each VM in the job can be processed by different proxy depending on proxy availability and load. In your case, it seems that hottadd (Virtual Appliance) mode is not supported for any of them. Remember, that the main requirement for hotadd is that the host running backup proxy server VM must have all datastores where protected VMs' disks reside connected to it. You can also find some other limitations of using hotadd mode in this KB article). Thanks.

[JohnG]

Hi, thanks for the reply.

But, Hotadd does work if I create a job and pick a proxy that is only configured with Virtual Appliance as the transport (not automatic). Let me try to explain better...

Here's what I have observed: for the same group of 50 VMs (let's say half are in iSCSI luns and half in NFS):

case a) If I select a proxy for that job and set the transport mode to automatic, the proxy selects SAN mode for VMs in iSCSI and NBD for VMs in NFS.

case b) If I select a proxy for that job and set the transport mode to 'Virtual Appliance', then the job runs in Vritual appliance for all VMs (iSCSI and NFS).

case c) If, for the same job of 50 VMs, I create two proxies, one automatic and one forced to Virtual Appliance, then the job only seems to use the 'automatic' proxy. So when the job backs up a VM in iSCSI lun, it uses SAN mode and when it reaches a VM in NFS, it uses NBD.

Hence my assumption that the job picks the proxy at the beginning of the job. If not, how can we explain that the job with two proxies does not use the Virtual Appliance proxy for VMs that are hosted in NFS?

Is this working as intended or not? How do I get around this? So that if I run one job, all VMs in iSCSI luns use SAN and VMs in NFS use Virtual Appliance.

Thanks.

[foggy]

If multiple transport modes are supported for the same proxy, Veeam B&R selects the mode using the following precedence: Direct SAN > Virtual Appliance > Network. However, the list of available proxies is detected at the beginning of a job, indeed. In your case, most likely Direct SAN was considered available for both datastores at the beginning of the job and the single proxy was selected for the whole job (and this proxy cannot use Virtual Appliance mode). But for some reason at the stage of processing it failed over to network (it obviously could not failover to another proxy (the one using Virtual Appliance) as it was not selected as available at the beginning of the job). To confirm or disprove this, full logs are required, so I suggest to open a support case to let them investigate.

As a workaround, you can create two jobs - one per each datastore - and specify the appropriate proxies to them.

Thanks.

[JohnG]

Thanks. I'm going to double check all my settings and submit a ticket.