Changing Replica Location

[t_ahrens]

Hi,

some time ago, I have created some replicas with destination "Host1", and now I have set up a second backup host, and want to change the destination of some of the replication jobs to "Host2".

But I want to use the already existing data of the replica VMs, because the data of "Host1" and "Host2" is stored at the same NAS.

Is there a way, to unregister the replicas at "Host1" and register them at "Host2", but keep all the job settings (including restore points and so on) and the existing data of the replicas?

Regards
Thorger

[bdufour]

cant you just vmotion the replica to the other host, seems like theyre using shared storage? ive done that before.

[t_ahrens]

That might work, but in this case the hosts use "VMware vSphere Essentials" licenses, which does not include the vMotion option

[bdufour]

im pretty sure if you register the vm on another host, it will keep the restore points as they are only snapshots. it will change the moref id (i could be wrong), but im pretty sure it does if you register it on another host. in that case veeam will see it as a new vm and will have to perform a vm digest recalculation, not the end of the world - as it doesnt consume a lot of bandwidth. im sure a veeam guy can tell us exactly what happens, im interested to know myself.

by the way, are these replicas going over the wan or lan?

[skrause]

You can accomplish this pretty easily:
1) Unregister the VM from "Host1"
2) Register the VM to "Host2"
3) On your replica job, select the "Low Connection Bandwith (enable replica seeding)" on the first settings page (Name.)
4) Go to the "Seeding" page and in the lower half check the "Replica Mapping" box and then browse and select the VMs that now live on your 'Host2'

On the next job run, Veeam will have to calculate digests to compare the seed to the production system which can take some time depending upon the size of the VM (it has to read the entire disk). After the first run it will run normally.

[t_ahrens]

I'm afraid, but I can't tell you, if the way described by Steve would have worked in my scenario. We had to do some tests last weekend, and I didn't want to try around too long. So I decided to delete the replicas and created them new on the 2nd Host, that took less time than I was expecting! (We use LAN for replication)

Last weekend, we manually tested failover and failback, and my impression was, that the failback process including the calculation of digests took nearly the same time it took before, for the initial creation of the replica (something between 5 and 6 hours for a 1,5 TB VM).

I will try it the described way the next time, I'm sure it will happen again, e.g. when "host3" is necessary

[bdufour]

good deal, and may want to look at getting licensed for vmotion for the future as well.

cheers