Upgraded the OS on my #HomeLab #KVM #hypervisor and its #Sanoid + #Syncoid #ZFS backup target #server today.

It never ceases to amaze me that upgrading from one #Ubuntu LTS version to the next is way faster than installing the monthly updates on our #Windows servers at work.

I finally have a compelling reason to start copying a #ZFS file system from one host to another. I am using #sanoid on #FreeBSD so, naturally, it seems I should use #syncoid (since they both come from the same package). The primary purpose of this replication is backup; this is a #bhyve instance I want to keep running.

Are you using synoid already for replication between hosts? Please share your recipe.

@sebsauvage #ZFS mirrors et/ou des datasets avec copies=2 et/ou avec #sanoid+#syncoid pour les sauvegardes distantes.

Tests de résilience à la corruption : https://www.openoid.net/testing-the-resiliency-of-zfs-set-copiesn/
> The TL;DR here is that copies=n is better than nothing... but not by a long shot, and you do give up a lot of performance for it. Conclusion: play with it if you like, but limit it only to extremely important data, and don't make the mistake of thinking it's any substitute for device redundancy, much less backups.

@somenxavier If you have the freedom to choose the file-system, and the time investment needed to learn it, #ZFS match all your criterias: you can either use directly the zfs subcommands like snapshot/delete/send/receive or use a higher level tool like #sanoid and #syncoid.

Tip: go to the hidden .zfs/ directory where a ZFS dataset is mounted to browse its snapshots.

If and only if you're using it on a single disk setup, check also #Btrfs for the same features.

@carlchenet
- #ZFS pour les fichiers ; avec un peu de #Perl pour aider : #sanoid (création et expiration des snapshots) et #syncoid (envoi incrémental des snapshots chiffrés via #ssh)
- #rsync vers #ZFS pour les fichiers hors #ZFS
- #NixOS+#Git pour les confs de / et de ~
- #ZFS pour chaque BDD #PostreSQL (aucun dump requis)

À noter que #ZFS ne requiert pas la clé de déchiffrement pour la vérification régulière de l'arborescence des sommes de contrôle des données (zpool scrub)

@ntnsndr clearly more demanding than other solutions (eg. #rclone, #rsync, #BorgBackup) but I'm using #OpenZFS on #Linux for backups, with the help of #sanoid (for taking and pruning snapshots) and #syncoid (for sending them incrementally and permanently encrypted over #ssh)
https://en.wikipedia.org/wiki/ZFS
https://github.com/jimsalterjrs/sanoid
FYI, their author co-hosts a weekly podcast about #ZFS sysadmin https://2.5admins.com/
Also I'm aware of at least one commercial cloud provider https://zfs.rent

@Ninjatrappeur to limit that stress, even with #ZFS which resilvers only the data used and not the whole disk, I still stick with mirroring (it's like RAID1) and avoid the more sophisticated raidz/dRAID. And since using #ZFS, I also keep many external backups up to date, because it's easy even without decrypting datasets (FYI, I'm using #sanoid & #syncoid, either with the help of their #NixOS modules or with standalone #bash scripts)

@luc il y a aussi le couple #sanoid (pour prendre et supprimer automatiquement les snapshots) et #syncoid (pour transférer les snapshots par #ssh) en #Perl. Je l'utilise désormais pour mes #sauvegardes #ZFS.
https://github.com/jimsalterjrs/sanoid

#syncoid supporte bien les fonctionnalités permises par #ZFS : incrémentale, résumable, via #mbuffer, avec ou sans déchiffrement / décompression.

L'intégration à #NixOS nécessite encore un correctif que j'essaye de pousser ici : https://github.com/NixOS/nixpkgs/pull/83904