Migrating Data to New Server

[MidasMagoo]

I've inherited a project mid-stream and could use some help.

We are temporarily using our backup storage server as the primary server while the primary server is being upgraded. The backup server runs FreeNAS as will the primary server. It would seem that replicating a snapshot is the most efficient way to transfer the data back to the primary server.

1) Presumably I have to clone the snapshot once it is on the Primary Server to make it read-write. Can I run this clone forever? Is there a way to designate the clone as the new primary dataset?

2) Future replication tasks will go the opposite direction--from Primary to Backup. Will that first snapshot from the Primary recognize that it came from the Backup? Or will it be treated as all brand-new data that needs to be seeded to the backup?

3) The Backup Server intentionally has about 2.5x the storage space of the Primary. The idea was to keep a minimal amount of snapshots on the Primary Server, but many more on the Backup Server. What I've read so far on snapshots seems to indicate that the number of snapshots kept on the Backup is dictated by what is on the Primary. Can I setup a separate snapshot schedule on the backup to retain old data longer?

Thank you.

 

[Chris Moore]

you can take a snapshot on the machine where the data is coming from and then create a pool on the destination machine and do a ZFS send and receive between the two machines you don't have to send the snapshot over as a snapshot

 

[MidasMagoo]

you can take a snapshot on the machine where the data is coming from and then create a pool on the destination machine and do a ZFS send and receive between the two machines you don't have to send the snapshot over as a snapshot

Thank you, Chris. Clearly I have some more reading to do, as I was left with the impression that the snapshot STAYED a snapshot, would have to be mounted and would be read-only unless cloned. You seem to be suggesting that I could:

1) Return the data from a Backup Server snapshot to the Primary Server via ZFS send and receive in a form where the data is ready to use at the Primary Server,

2) Send a snapshot once per day from Primary to Backup, such that the Backup is a usable mirror (as of that snapshot) of the Primary, and

3) Create a snapshot routine on the Backup Server that is independent of the routine on the Primary Server, wherein I incorporate however far back in time I want to be able to restore deleted/changed data--presumably by mounting that older snapshot and copying/pasting any data I want back to the Primary Server.

I understand that such an architecture would mean that the older snapshots are not backed up--much like using Apple's Time Machine--in the event of failure of our Backup Server.