SOLVED Error when replicating between 2 FreeNAS boxes

[adrianwi]

Finally got my main FreeNAS box running just how I wanted, so thought I'd get around to sorting out my backup properly. I'd been using rsync to copy some folders across, but I was only picking up the ones containing my personal data (files, pictures, music, movies, etc.) and thought setting up a full replication would be a better solution.

I set-up one periodic snapshot of the zpool (/mnt/APE_pool) and marked as recursive. This created the ~25 snapshots. I then set-up the replication task and checked the keys were set-up correctly, which I think they are.

Task appears to have started, but I'm getting the following error:

Code:

Feb 23 01:01:07 freenas1 autorepl.py: [tools.autorepl:432] Replication of APE_pool@auto-20150222.1719-2w failed with cannot unmount '/var/db/system': Device busy
Error 33 : Write error : cannot write compressed block

Any ideas what might be wrong?

 

[adrianwi]

Don't really understand what the problem was, but I thought I'd remove the replication task, snapshots and periodic snapshot and start again.

Looks like it's up and running now although might take some time to copy across ~9TB!

What happens to the other periodic snapshots whilst the 1st replication is underway, and will it just sort itself out?

 

[cyberjock]

It should, yes.

 

[adrianwi]

That's great, thanks!

 

[adrianwi]

Can I ask another noddy question?

The replication is almost finished so I have the exact same structure on freenas2 as I have on freenas1. As it's just a backup, of what it essentially backups from other systems (on freenas1) I built freenas2 without redundancy so have more storage available than on freenas1.

Can I add a dataset and some data on freenas2 (that's not stored on 1) without screwing up the replication from freenas1?


Sent from my iPhone using Tapatalk

 

[adrianwi]

The answer to this noddy question, if anyone is interested, is yes, but it will get removed when the next replication task runs!

A better solution is to point the replication task at it's own dataset, and then you can do as you like with the root :D

Oh, and the snapshots taken when the first (long) replication task is running are processed once it's finished, as CJ stated.

 

[cyberjock]

Sorry, I'm catching up on forum threads tonight.

You should *always* replicate to a separate dataset. Replicating from zpool -> zpool creates lots of problems (some will be fixed in some future patches). Even if you are replicating with intent of using the destination as your new server and new storage, you should replicate from zpool -> zpool/dataset and then move the datasets to their proper location afterwards.