snapshot needed by replication - run replication first

[herzkerl]

Hi everyone,

I’m running a Proxmox VE cluster (PVE 8.x) with three nodes and have a daily ZFS replication job configured. One of our VMs has a snapshot named Test (created on 2024-11-18) that I would like to delete, but I’m receiving the following error when trying to remove it:

TASK ERROR: snapshot 'Test' needed by replication job '145-0' - run replication first

Now here’s what puzzles me:
• The snapshot is more than 4 months old.
• Our replication job runs daily, and it has been running without errors since then.
• The replication target is the other two nodes in the cluster, and I assumed that snapshots are no longer needed locally once they’ve been successfully replicated multiple times.

My questions:
1. Why is this old snapshot still considered “needed” by the replication job?
2. Is there any way to verify whether this snapshot is still actively used by the replication mechanism?
3. Could this be a bug or some kind of leftover state from an earlier replication run?

If needed, I can share the output of zfs list -t snapshot or pvesr status for more context.

Thanks in advance for any insights – much appreciated!

 

[UdoB]

Yeah - been there just yesterday.

Or at least at a similar situation: an old VM had been on ZFS at the beginning, was migrated to Ceph several months ago and now back to ZFS. Configuring Replication gave me the same error message.

For me the culprit I did not realize immediately was: the problematic snapshot was on the destination, not on the source. And it was a relict from former times, definitely outdated.

I just destroyed it (be careful - there is NO "are you sure?"!) manually and Replication startet a fresh relationship from source to destination