How to fail back?

[tom11011]

Hi Group.

We are running veeam 6.5 in a vmware environment.

Currently, we have 2 geographically diverse sites whereby SiteA replicates to SiteB. It's time to test DR failover.

Although I can power up my replicated vm's at siteB, how do I do a failback to site A?

When I goto Restore --> Restore from Replica --> Failback to Production

No replicas are found in the list at all, nothing from siteA or B, even after clicking the populate button.

The goal here would be to failback any changes from the replia vm at the DR siteB back to production vm at SiteA.

Thanks for any help.

[Vitaliy S.]

Hi Tom,

Can you please tell me how do you power on VMs at SiteB? Do you do that via Veeam backup console?

Thanks!

[tom11011]

No, all of our tests have been simply shutting off SiteA, changing DNS, and activating siteB without failback. Once our failover test is completed, we shutoff siteB, reactivate siteA, and just let Production overwrite any changes to DR at siteB.

This would be the first time we will attempt to fail back.

[Vitaliy S.]

Failback option is only available if you run the failover process using Veeam backup console, see our Evaluation Guide (page 72) for further details.

Right now, you can create replication jobs, and map VM replicas to production VMs to transfer all VM changes back.

[tom11011]

Ok, what exactly happens when you put all your vm's into failover mode? How are the changes tracked in order to failback? I have this feeling you are going to tell me snapshot? Or is it more like a vrb file.

If so, what's the word on operating say 100 vm's for say 24 hours at the DR site? Those snapshots might become very large??

Also, what if the production site is down and you do not have access to place a vm into failover mode?

I guess what I'm looking for is a way to replicate only the changes from the DR site back to the production site.

also, I wasn't sure what you meant by "Right now, you can create replication jobs, and map VM replicas to production VMs to transfer all VM changes back." is that something different than what the guide described putting the vm's into failover?

[Vitaliy S.]

Ok, what exactly happens when you put all your vm's into failover mode? How are the changes tracked in order to failback? I have this feeling you are going to tell me snapshot?

You're right. All the changes are written to a snapshot file and then replicated back to production VMs (if required).

If so, what's the word on operating say 100 vm's for say 24 hours at the DR site? Those snapshots might become very large??

Yes, they can, that is why you should have at least 10-20% of free space on the destination datastore. Depending on the disaster situation you can choose one of these ways to proceed with:

1. Perform permanent failover
2. Perform failback
3. Undo failover

Also, what if the production site is down and you do not have access to place a vm into failover mode?

I guess there is a confusion with "failover mode", you do not place source VMs into this state, you do it with target VMs. So when you choose to perform failback operation, Veeam B&R checks what VM replicas you have failed over to and then suggests to start the failback operation for these VMs. In other words, if you failover manually, Veeam B&R doesn't "know" that VM replicas are in a failover state.

I guess what I'm looking for is a way to replicate only the changes from the DR site back to the production site. also, I wasn't sure what you meant by "Right now, you can create replication jobs, and map VM replicas to production VMs to transfer all VM changes back." is that something different than what the guide described putting the vm's into failover?

Yes, it is. Since you've started your VMs manually, in order to transfer all the changes back you need to create regular replication jobs and map them to production VMs.

Hope this helps!

[tom11011]

Ok, I think I understand what you are saying in the last paragraph.

If I don't use veeam to activate failover, then at the end of the disaster, I create new replication jobs to go from siteB to siteA. Will these be full replications overwriting the original virtual machines or will they only send the changes?

[foggy]

They will only send the changes if you map them to the existing VMs in site A.

[tom11011]

thanks for the help. I'm going to setup a test vm and give it a try.