Looking for advice on setting up disaster recovery/DR testing for a small company.

[AdminKMC]

I have a three server cluster that currently runs our primary application server and will be running additional servers over time. All servers are on the same rack/switch on 10Gbps ports + dedicated Corosync network via dedicated and isolated switch We have HA setup, Veeam running nightly backups, and PBS running hourly backups during business hours. So far so good, but I don't have the ability to test restore without actually restoring the production server.

Potential complication: Our application server is running from shared storage in the form of NvME media from a NAS also 10Gbps port.

Currently, I'm considering setting up a single node PVE outside the cluster to restore to via both Veeam and PBS, but I'm not sure if I can restore the full server to a single node without the shared storage. Looking for advice on a setup that would allow me to restore and test the application server backups on a system completely separate from our production cluster.

 

[HawHoo]

We have a similar Issue. We are a MSP and need to do RD Testing for our clients.
With other backup systems we are able to spin up the VM from backup ( in a sandbox) and test the backup of a VM.
We like PBS but it does not seem to offer a straight forward DR test option short of having a separate PVE node dedicated for DR Testing.
That is cost prohibitive. I hope I am missing something.....

 

[AdminKMC]

We have a similar Issue. We are a MSP and need to do RD Testing for our clients.
With other backup systems we are able to spin up the VM from backup ( in a sandbox) and test the backup of a VM.
We like PBS but it does not seem to offer a straight forward DR test option short of having a separate PVE node dedicated for DR Testing.
That is cost prohibitive. I hope I am missing something.....

That's how I understand it to be. My situation is further complicated by shared storage. Short of the hardware to mirror the production setup for a test environment, our only option is to schedule a time to restore the actual server in place. We have both Veeam and PBS, but it'll still be some high tension during the restore process until we see the server come back up. If I come across any other options, I'll reply back to this thread.

 

[HawHoo]

We had this option with StorageCraft as well as Synology's active backup for business.
As far as PBS is concerned, I believe it allows you to restore to a different VM ID so you could still use the same shared storage.

 

[deakus]

Create some isolated networking bridges, perhaps create a new zone in SDN called DR. in SDN and create a DR version of your production vlans/subnets *isolated* from your production bridges. You can then restore your backups in isolation ( DO NOT TICK BOOT ON RESTORE).

Once you have performed your restoration, you can modify the networking settings in your VM's to your DR Zoned networks. You could restore to the same storage as production, but if you are willing to create a whole new PVE host, perhaps a separate NFS share on your NAS suffice for testing.

 

[alexskysilk]

but I'm not sure if I can restore the full server to a single node without the shared storage.

you can, for either. the backup product only cares that there is an api entry point, not the overall structure of your hardware. As long as there is a node with virtual resources for the restored vm configuration it will restore just fine.

We like PBS but it does not seem to offer a straight forward DR test option short of having a separate PVE node dedicated for DR Testing.

define straightforward. Going further, define a DR test, especially since for actual DR there would be components (network, primarily) that PBS cannot handle in any case. In my experience, DR protocols involve more then one product/api, and a DR test can/should be for the entirety of your environment.

More to the point- restoring to the same hardware you run your production isnt a DR test at all, its a restore test.