Replication Task "Finished", but "No snapshots sent yet." Daily replication is a full day behind.

[alexmarkley]

I'm running TrueNAS-12.0-U4.1 on a TRUENAS-MINI-3.0-XL+. I have two identical systems running (almost) identical configurations. (We'll call them A and B.)

In both cases I have a small zpool which I want to replicate (backup) to another local zpool on a daily basis. (So the replication is A -> A and B -> B. No cross-system or SSH based replication.)

I have a snapshot task set up on B like this:

Snapshots are running as expected:

My replication task is set up like so:

Additional details:

This was all set up yesterday. As of midnight, however, the replication did not actually copy any data, and here's the current state (again on system B):

In addition, when I click RUN NOW it asks for confirmation and then goes directly to FINISHED with "No snapshots sent yet".

Now, interestingly, on the other system (on system A, which has effectively the same configuration) I have a few more daily snapshots (because it was set up earlier) and it is showing a different (but possibly related) issue:

Notice the Last Snapshot in this case is actually yesterday's snapshot and not the snapshot from today. Even if I manually hit RUN NOW it doesn't successfully pick up today's snapshot. So the replication job is a full day behind.

There is definitely a snapshot from today on system A:

Not sure what I'm doing wrong here, but it almost seems like the replication task is simply looking for the wrong YYYY-MM-DD...? Anybody have any thoughts here?

Thanks for reading, and thanks in advance for any thoughts!

 

[alexmarkley]

Well I'm very confused. The problem seems to have "cleared up" by itself. I would really like to understand what happened there, so any input would be appreciated!

 

[winnielinnie]

I'm glad it cleared up by itself, but is tank meant to serve as a destination for backups? If so, it looks like you have Periodic Snapshot Tasks defined for uncaged, as well as tank. Why would that be neccessary if you're replicating the same automatic snapshots from uncaged to tank?

For context, I replicate to a "local" pool named backups, but I have no automatic snapshots defined for backups, only for my main pool.

Consider also the option "Save Pending Snapshots", so that snapshots which haven't yet been sent to the destination are not destroyed on the source (which can leave you without a "base" snapshot for incremental replications.)

 

[alexmarkley]

@winnielinnie this is a good catch. Tank is primarily used for video editing projects, so I would like to make sure its datasets have periodic snapshots in case somebody accidentally deletes something.

Tank is serving as a backup destination for uncaged simply because tank has redundancy, whereas uncaged has no redundancy.

It did occur to me that the periodic snapshot tasks for tank might conflict with the replication task for uncaged, so I updated the tank periodic snapshot tasks, adding an exclusion like so:

I wonder if the conflicting periodic snapshot configuration may have caused the weird behavior I observed before?

Also sure I will go ahead and enable "Save Pending Snapshots" in the replication task.