What I think is happening is that the snapshot on the PUSH server are expiring and Freenas removes them.
One way you can fix that is doing replication through ssh.
I was having similar issue when testing Freenas 9.3 Beta and couldn't finalize one of the dataset because the snapshot was not on the PUSH anymore.
If you started from scratch all over, I would take a Recursive Manual snapshot of the volume.
Because you are creating a manual snapshot, I think it doesn't expire it no matter how short your snapshot life pan maybe.
You can save the snapshot list to file so you can edit it and add the zfs send and zfs receive command. You can select which snapshot to send.
To make it snappier, delete the manual snapshot you just made for large dataset. Keep the volume snapshot that do point to any dataset.
That way you can recreate a perfect replica to start with, ie replicates volume autorization and such.
Then create a new recursive manual snapshot of the entire volume and send the remaining dataset using ssh.
It is a bit of complex task but it will work. If you need more details on how to accomplish this I can explain and give you the series of commend that will help you achieve entire replication.
If you have already made manual replication in the past and they have been relicated already, then you can resume replication from there, but you may have to delete the automatic snapshots that are present on the PULL side as the PUSH will only replicate from the last snapshot (only removing automatic snapshot newer that the manual snapshot).
One way you can fix that is doing replication through ssh.
I was having similar issue when testing Freenas 9.3 Beta and couldn't finalize one of the dataset because the snapshot was not on the PUSH anymore.
If you started from scratch all over, I would take a Recursive Manual snapshot of the volume.
Because you are creating a manual snapshot, I think it doesn't expire it no matter how short your snapshot life pan maybe.
You can save the snapshot list to file so you can edit it and add the zfs send and zfs receive command. You can select which snapshot to send.
To make it snappier, delete the manual snapshot you just made for large dataset. Keep the volume snapshot that do point to any dataset.
That way you can recreate a perfect replica to start with, ie replicates volume autorization and such.
Then create a new recursive manual snapshot of the entire volume and send the remaining dataset using ssh.
It is a bit of complex task but it will work. If you need more details on how to accomplish this I can explain and give you the series of commend that will help you achieve entire replication.
If you have already made manual replication in the past and they have been relicated already, then you can resume replication from there, but you may have to delete the automatic snapshots that are present on the PULL side as the PUSH will only replicate from the last snapshot (only removing automatic snapshot newer that the manual snapshot).
Last edited: