Cautionary tale about editing replication tasks

It’s my own fault.

I replicate two large data sets (2TB each, video) from my main server to a remote server on a different continet. When I first set it up a few years ago, I was a little overeager, setting up daily snapshots and keeping them 6 months. The main server is on the US typically crippled cable upload of 20mbps, so replicating from scratch would take many moons. So I set up a local pool, replicated to it, detached it, carried it to the remote location, imported it, replicated to the new location, and after that, replication took minutes at most. Working fine for years.

Deciding I was overeager, I changed the snapshot schedule to monthly, and renamed it monthly-auto etc. When it next replicated, it started replicating from scratch. And of course the first step was to delete everything on the remote location. I stopped, reset everything back to the original, but it was still all gone.

I can’t see it in the documentation, or in the screens that make it so easy to modify snapshots. But is there a way to safely modify an existing snapshot/replication schedule?

It sounds like you have enabled the “Replication from scratch” checkbox in the replication settings. Uncheck that option and replication shouldn’t automatically start from scratch.

Changing snapshot and/or replication schedule shouldn’t cause any issues, at least I don’t see how. I think your issue is related to changing the naming schema. I would suggest keeping the naming schema always the same, that way it won’t interfere with replication.