Replication starts over and over

[eccevery]

I have a server for personal "main storage" and have set up replications to a secondary server in another building. However the replication job seems to start over and over, and I can't figure out why. There is only one snapshot in the list, and it's the same snapshot that seems to be replicated in a loop. No files have yet shown up in the target server, so maybe each "try" for replication fails? I'm not sure what log files to look in, so far I have not find anything that seems helpful.

There was more snapshots in the source server earlier. In a try to fix this I deleted all snapshots, took one snapshot and have i live for a very long time so the replication would have time to complete (it takes a few days, copying roughly 2TB over 100Mbit).

Both source sever and target uses FreeNAS 11.2. The Snapshots are taken every month and are kept for one month. I can see data going out the source server and the interface on the target server receiving the same amount of data per second, so it seems it reaches target.

Ideas?

 

[PhilipS]

You won't see any files on the target until the replication finishes. Do you have email/alerts setup? That is how I get my replication failure messages. If the connection between servers is being broken at all, then replication will restart from the beginning. There are a some ways around that - the easiest (if possible) is to perform the first replication with the machines on the same local network. Another would be to perform the zfs send manually with resume token enabled so you can restart the send if there is a failure.

 

[eccevery]

You won't see any files on the target until the replication finishes. Do you have email/alerts setup? That is how I get my replication failure messages. If the connection between servers is being broken at all, then replication will restart from the beginning. There are a some ways around that - the easiest (if possible) is to perform the first replication with the machines on the same local network. Another would be to perform the zfs send manually with resume token enabled so you can restart the send if there is a failure.

No I don't have email notifications setup. I've never really seen any benefit in this particular case to set it up, but maybe there is. I do however see files on the target now, but it still says it's replicating.

 

[eccevery]

Suppose I do this: I move the target server to the source server, plug it in the same switch and start a replication there. That means changing the IP of the target. It will be a lot faster. We assume it completes without errors. When I move it back again I will have to change the IP back as well. Will FreeNAS recognize the snapshot replicated and only replicate the diff, although the IP has changed?

 

[PhilipS]

When I move it back again I will have to change the IP back as well. Will FreeNAS recognize the snapshot replicated and only replicate the diff, although the IP has changed?

Yes.