Replication from Core to Scale - I don't get it

I hate to start yet another thread on this, but I can’t find an answer on the old forums or the new. It seems a hundred people have had this issue and no one has the answers. Either the thread dies or their issue was different than mine.

I have a TrueNAS Core system that I built back in 2020, so it has GELI encryption on the pool. It’s on the latest TrueNAS-13.0-U6.7 build. I just built a new TrueNAS Scale (Electric eel) and I’m trying to Pull replication for some (not all) datasets from the Core system to the Scale system. I’ve tried going through the tutorial put out by Tom Lawrence, I’ve tried doing the “Pseudo-root dataset” trick to rectify the error that the destination dataset must be encrypted as I’m trying to replicate an encrypted dataset (even though it is encrypted at the pool level and is inheriting, and I’ve even disinherited encryption and set up passphrase encryption for the sub-dataset), I’ve ensured I’m manually typing the destination, etc., etc., and no matter what I do, I can’t get it to replicate. It fails immediately, telling me the ‘Dataset has no snapshots’. First of all, on the source, the dataset has several snapshots. Of course, on the destination, there are none because it’s new and the replication is, according to the documentation, supposed to create a new snapshot upon first run. But, it won’t do that. I’ve ensured that I’m able to ssh into the Core system from the Scale system with the credentials provided. I’ve tried going the other way, ensuring I created a new user on the Scale side, specifically for this purpose with its own SSH key, but that didn’t work either. From what I’ve read, the best thing to do is to pull it from the Scale side, so I’ve moved back to trying that, but I get stuck at this snapshot issue. I even tried setting up a new snapshot task on the Scale machine to snapshot the Core machine that way, but that doesn’t seem possible, either. I’ve spent three days on this. I’m at a loss. Anyone have any ideas?

But you do have a snapshot task and snapshots on the CORE side? The snapshots to replicate must exist before replication. Replication tasks don’t create snapshots.

1 Like

Yes, on the CORE side there are tons of snapshots for all datasets in question.

And, while I understand replication tasks, themselves, don’t create snapshots, according to the documentation, the replication wizard will create the necessary snapshot task, automatically scheduled to match the replication task, on first run of replication job created by the wizard.

I’ve got this setup working fine minus the GELI encryption (not sure if that’s the deal breaker here).

Pulling datasets from CORE to SCALE. I setup the SSH Keypairs, then the SSH Connection but used Manual instead of Semi-automatic.

Then setup the replication task making sure I matched the snapshot naming schema I wanted pulling from the CORE end.

I’m beginning to think it’s due to the GELI encryption, because I ended up doing the same thing you did, setting up the SSH connection manually, instead of semi-automatic just to try anything as a possibility and I still get the same error about the snapshots.

If that’s going to be the case, it would be nice if it were mentioned in the documentation so people didn’t waste days trying to figure out why this “easy” replication wizard doesn’t work and also doesn’t give an error that mentions that the encryption is the reason, instead of an error that makes no sense about a lack of snapshots when there are literally dozens of them.

Indeed it’s an odd one. IF (no promises) I get sometime I will try and recreate your scenario and see if I get the same results and report back.

That would be amazing. Thank you. If you need any further information about my setup to help, let me know.

Had to go all the way back to FreeNAS 11 to get GELI. God I miss the old FreeNAS logo.

Yeah sorry getting the same issues as you.

Have you tried PUSH from CORE to SCALE?

Yeah, that’s exactly what I’m getting. I tried pushing from CORE to SCALE, too. Can’t remember the exact error, now, but it didn’t like something. I’ll try again and see what it is.

Try dropping this into the Auxillary Paramters of your SSH config under ‘Services’ on the PULL machine.

PubkeyAcceptedAlgorithms +ssh-rsa

You know, I never thought about that. That’s a damn good idea. I will try that. So many network hardware pieces need that before you can SSH in, I never even give it a thought that true NAS would.

Nope, still nothing. I’ll have to try some other way. Maybe RSync, which is not ideal, but it’s better than nothing.

Very odd.

The only way I could re-create your situation was to roll an old FreeNAS version 11.3-U5 that still supported GELI pool creation. The issue I had then was old FreeNAS didn’t want to replicate to new TrueNAS due to outdated ssh library hence why I needed to add that parameter.

It seems like you have upgraded to the latest version of CORE so it would have made no sense if that did work but I honestly can’t think of anything else.

I’ll upgrade my FreeNAS all the way to the latest version of CORE and see if replication works.

Just upgraded from 11 to 12 to 13 and PULL replication is working great.

1 Like

hmm…not for me for some reason. Wish I knew. I’ll have to keep playing with it, I guess.