[BETA] [HPE] HPE-VME Single Worker per Backup Job?

[phil_nz]

We have a Veeam Worker for every node in the VME cluster.

However, if we have a backup job that includes multiple instances across multiple hosts then it appears the job only selects a single worker for the whole job.

This means that VMs on the same host as the worker get backed up 'hot-add' style and the rest are 'NBD'. This generates the 'Performance may be affected' warning.

Is it possible to use multiple workers per job, to get the best performance and possibly run the streams in parallel?

[EvgenyBaev]

Hello phil_nz

Thank you for your insightful question. You have identified a bug in our worker selection logic.
Unfortunately, there is currently no feasible workaround without rebuilding the plugin. However, we will address and resolve this behavior in the upcoming 1.0 release, ensuring it works as expected.

[Sloan]

I was wondering about this same behavior in my testing. Good to hear that it should be addressed in the official release.

[patrickds]

I've recently added a second worker on a 2 host cluster, and see similar behaviour.
Both workers are used, but not always for VMs on their own host.
So some VMs get the notice about NBD being used.
Since this is a Simplivity cluster, fully 10Gb connected, I doubt there is an actual performance impact.
Just updated from B&R 13.02 to 13.1, but that hasn't changed anything.

[cody.ault]

Hey, could you login to the Morpheus console and go to the VMs and validate you have a worker on each host? If you didn't use the setting for host affinity, it's possible that both workers are on the same host: https://helpcenter.veeam.com/docs/vbr/u ... tml?ver=13
Hotadd only works when the worker is on the same node as the VM being protected. If you have a worker on each node and you're using affinity, then hotadd should work.

[patrickds]

Hi,

affinity is set on the worker configuration, and the workers are indeed on their assigned hosts.
I had the placement policy set to fixed in the VME manager, but when checking it now, it has been reset to failover.
Probably by Veeam B&R when doing the prepararation on starting a job.

[cody.ault]

If you have host affinity configured for the workers in the worker settings in VBR, then it might be helpful to create a support case. Please feel free to post the ticket number after you've uploaded the logs.