Transitioning from zfs-auto-snapshot to sanoid/syncoid?

Hi! I’ve been using zfs-auto-snapshot keep and prune snapshots for backup purposes for the last few years. I have a snapshots going back to offsite_daily-2022-09-21-2205. In order to get them to my offsite backup, I have to manually run a script which I almost never do. Ooops! It seems like syncoid would be the right tool for the job.

I’m wondering if there is a way to setup sanoid / syncoid on the existing dataset?

Or, if there isn’t, what would be a reasonable way to transition this setup to sanoid / syncoid?

For a peek, I have something like this setup on the main computer.

$ zfs list | tail -n20

internalpool@zfs-auto-snap_hourly-2026-10-27-0417 26.6M - 384G -
internalpool@zfs-auto-snap_hourly-2026-10-27-0517 26.6M - 384G -
internalpool@zfs-auto-snap_hourly-2026-10-27-0617 26.6M - 384G -
internalpool@zfs-auto-snap_hourly-2026-10-27-0717 26.6M - 384G -
internalpool@zfs-auto-snap_hourly-2026-10-27-0817 26.6M - 384G -
internalpool@zfs-auto-snap_hourly-2026-10-27-0917 26.5M - 384G -
internalpool@zfs-auto-snap_hourly-2026-10-27-1017 26.5M - 384G -
internalpool@zfs-auto-snap_hourly-2026-10-27-1117 15.5M - 384G -
internalpool@zfs-auto-snap_daily-2026-10-27-1138 0B - 384G -
internalpool@offsite_daily-2026-10-27-0738 0B - 384G -
internalpool@zfs-auto-snap_hourly-2026-10-27-1217 18.3M - 384G -
internalpool@zfs-auto-snap_hourly-2026-10-27-1317 26.6M - 384G -
internalpool@zfs-auto-snap_hourly-2026-10-27-1417 26.7M - 384G -
internalpool@zfs-auto-snap_hourly-2026-10-27-1517 28.2M - 384G -
internalpool@zfs-auto-snap_hourly-2026-10-27-1617 760K - 384G -
internalpool@zfs-auto-snap_frequent-2026-10-27-1630 112K - 384G -
internalpool@zfs-auto-snap_frequent-2026-10-27-1645 384K - 384G -
internalpool@zfs-auto-snap_frequent-2026-10-27-1700 7.20M - 384G -
internalpool@zfs-auto-snap_frequent-2026-10-27-1715 0B - 384G -
internalpool@zfs-auto-snap_hourly-2026-10-27-1717 0B - 384G -

If you want Sanoid to prune your existing snapshots according to policy you set there, you’ll need to rename those snapshots to match Sanoid’s naming format.

Syncoid, on the other hand, will work with any snapshots you have. Zero additional configuration is needed in order for syncoid to replicate snapshots you already have, it works with any and all snapshots and doesn’t care what they’re named or how they were taken.

I migrated away from ZAS to Sanoid as well. I pretty much started deleting all the fine grained zas snapshots because I didn’t want to spend time coming up with a better solution. I think I only kept monthlies. This is pretty lazy, I know. However I did have about 300,000 snapshots bc zas is not all that assiduous at cleanup if a system goes offline for extended periods.

I have a setup where a preexisting situation with someone else’s custom script doing a quiesce/snapshot/prune is being replicated to a backup server by Syncoid, and then I have a short little cronned pipeline with a head -n- in it that prunes all but N plus monthlies or something like that from the back-ups. No Sanoid in the picture. (And then I go in and decide what to drop when the capacity alerts, because no automatic pruning is going to obviate the need to have alerting for that. ETA: and monitoring that the newest snap on the backup server is no older than x hours).