zfs encryption is still corrupting datasets when using zfs send/receive for backup (huge win for mail datasets), would be cautious about using it in production:
Please stop using send/recv . Your backups should be based on non ZFS tech to avoid all your eggs in one basket.
Yes send/recv is fine for immediate recovery, but other than block level replication for immediate (my server is now inside the tornado) recovery this isn't advised.
Also, who cares if a single filesystem dies, that's why you have inter-server replication. Nuke the bad server and rebuild before the next 3 or 4 die.
I’ll never use ZFS in production after I was on a team that used it at petabyte scale. It’s too complex and tries to solve problems that should be solved at higher layers.
yeah, we use Cyrus replication still - it's protocol specific so it detects changes very efficiently as well, using the internal MODSEQ system also used for the JMAP /changes and IMAP CONDSTORE/QRESYNC.
Plus it has protocol consistency sanity checks built in.
https://github.com/openzfs/zfs/issues/12014