Backup fails for half of VM's

[Thomas_VDB]

Hi all,

We have 2 ESXi 5.5 hosts, connected to a shared storage (NFS SAN).
We have been using Veeam to do our nightly VM backups and this has been working great.
Up till now, network transport mode (slowest) has always been used, so I wanted to optimize.

I am now able to use hot add and direct NFS, but only for the VM's on 1 of our 2 hosts. When backing up the VM's on the 2nd host, it falls back to network mode.

However I think I found the reason :
Our shared storage has 3 gigabit network interfaces : one for management via the company network, and then 2 interfaces, each one directly connected to 1 ESXi host for storage I/O.
Each network interface has a different subnet : 1.0.0.x for company network, 10.0.1.x for connection with host 1 and 10.0.2.x for connection with host 2.

Now to use hot add or direct NFS, I reused a VM on each host as a proxy (VM proxy 1 on host 1, and VM proxy 2 on host 2).

We have one backup job that contain the VM's of both ESxi hosts. Now it seems obvious that if we want to use hot add or direct NFS, VM proxy 1 should be used to backup the VM's on host 1,
and VM proxy 2 should be used to backup the VM's on host 1.

However when the backup job executes, Veeam desides that VM proxy 1 is on the same company subnet as all other VM's, so VM proxy 1 is used to backup all vm's (also the VM's on host 2).
So that's why I think that the backup falls back to network mode when it tries to backup VM's on host 2 with a proxy running on host 1.

I opened a case for this, but Veeam advises to use 2 backups jobs, each with there own defined proxy.
I understand this, but I wanted to know if this is common practice? Do we have such a 'unique' setup?

Thomas.

[PTide]

Hi,

I opened a case for this, but Veeam advises to use 2 backups jobs, each with there own defined proxy.

Can you post your case ID please?

I understand this, but I wanted to know if this is common practice? Do we have such a 'unique' setup?

If you have several backup proxies that can process VM data in the Direct NFS transport mode, VBR selects the first proxy from the list of available proxies, though it is strange that Proxy 1 cannot process VMs on the Host 2 in direct NFS mode.

Thanks

[Ben.Pierson]

Just taking a stab in the dark, but is this part relevant (from https://helpcenter.veeam.com/backup/vsp ... ccess.html)?

If NFS volumes are mounted on the ESX(i) host under names, not IP addresses, the volume names must be resolved by DNS from the backup proxy.

"Our shared storage has 3 gigabit network interfaces : one for management via the company network, and then 2 interfaces, each one directly connected to 1 ESXi host for storage I/O.
Each network interface has a different subnet : 1.0.0.x for company network, 10.0.1.x for connection with host 1 and 10.0.2.x for connection with host 2"

May be that the VMs from host 2 are reporting to connect to storage at 10.0.2.x and the proxy located on host 1 can't find that path.