Initial seed

[ratease]

Hi,

How do you guys handle the initial seeding of data?

Our situation is this:

We have our primary SAN and ESXi-infrastructure at site A. Also at site A we have a spare SAN. We want to copy (seed) all the important VMs to the spare SAN and afterworths drive it to site B for ongoing replication.

My question is, how do we get the initial seed data on the spare SAN the easiest way? I've tried copying, but that requires the source VMs to be powered off, which is a no-go. If I do a clone, I will manually have to remove the extra copy (clone) from the servers afterworths (but not deleting the data) and even then, I'm not sure a clone keeps the same disk UUID.

So it seems to me however I do this, it involves a lot of manual work. Can anyone think of a better way? The manual is not very specific on this subject.

Thanks,

Rasmus

[a_p_]

Is there any reason why you don't setup replication while stil on the same site to let it do its job? Another option would be to restore a backup (in case you are using image based backup).

André

[ratease]

I don't use image based backup unfortunately.

I thought about setting vReplication up on-site and then move the whole SAN afterworths. But how will the vReplication handle this? I'm mean when the destination for replicas suddenly "disappears" because it gets a new IP address/location?

Can I set up vReplication on-site and then unconfigure vReplication without deleting the replicated data? In that case I can see how it could work...

Message was edited by: ratease

[mvalkanov · Accepted solution]

Hi,

You can:

1. Set up vSphere Replication on-site.

2. Wait for the initial sync to complete.

3. Perform failover using the latest available data (without powering off the source VMs) and without powering on the recovered VMs.

4. Unregister from vCenter Server the recovered VMs, but preserve the disk files.

5. Unconfigure the replications (already recovered - VRMS won't attempt to delete the replica .vmdk files).

6. Move the SAN to the other site.

7. Configure replication to the new site and use the initial copies as seeds.

Regards,

Martin

[ratease]

Hi Martin,

Just tested it - and it works! Thanks a lot.

Got one question though - and it's without a doubt just me not fully understanding the way vReplication works:

Step 3. Why does the recover create a new VM? I mean the source VM is still in the source system, so how come it recovers to a complete new VM instead of just using the original source VM? The reason I ask this, is because I'm a little worried about doing this operation on our production VM's, just in case the source should get overwritten or something... 😕

/Rasmus

[mvalkanov]

Hi Rasmus,

When configuring replication you choose the target datastore location for the replica files.

vSphere Replication will warn you, if you by mistake select the datastore folder of the source VM disks.

The recover operation always registers a new VM. You need to select a VM & templates folder in vCenter Server that does not have VM with the same name.

Be careful if/when removing the original source VM - vSphere Replication automatically cleans up any replica files, when the source VM is removed from the source vCenter Server inventory, unless recovery has been successfully performed on the replication.

Regards,

Martin