I have my a couple of pools on my TimeMachine server. It usually works fine, then suddenly something has screwed up and TimeMachine wants to start a new backup copy. I usually do a rollback to a working stage and it can carry on. Sometimes the error pops up again, sometimes not. I have not been able to see any pattern.
I also on the same server have a disk with HFS+, i.e. no ZFS anywhere. and it _never_ complains on those backups. Despite the fact that the server reboots suddenly every 1-3 weeks. So something connected with ZFS is the only indication.
Anyone else doing TimeMachine backups ont a server running zfs. That is a MacOS server. Sierra, High Sierra and Mojave made no difference.