checksum_charlie@checksum_charlie3-2-1 or it isn't a backup. Long, unasked-for, and written in the blood of my old photo library
I lost about four years of photos once because 'backup' meant a second folder on the same drive. When the drive died, both copies died together, holding hands. So here's the rule I evangelise, and the reasoning, because the rule without the reasoning doesn't stick. 3-2-1: three copies of your data, on two different types of media, with one copy off-site. Three copies means a single failure never leaves you at zero. Two media types means a failure mode that kills one kind of storage — a bad drive batch, a controller frying everything on the same bus — doesn't take all your copies at once. One off-site copy means the fire, the flood, or the theft that takes your whole homelab doesn't also take your only backup. The part people skip: a backup you have never restored is not a backup, it's a hope. Do a test restore. Actually pull a file back from the off-site copy and open it. I've seen backups that ran green every night for a year and were writing corrupt, unrestorable garbage the whole time. Untested backups fail exactly when you need them, because that's the first time anyone checks. RAID is not backup either, before someone says it. RAID is uptime. It protects against a drive dying; it does nothing against 'rm -rf the wrong directory' or ransomware or your own fat fingers, all of which replicate instantly to every mirror. Different job.
Join the conversation
Facet is free to read. To reply you need an account: one private root identity, and up to ten public personas that can never be linked to each other or to you.
Create an account
read_only_friday@read_only_friday· 7/6/2026, 5:09:44 AMThe 'RAID is not backup' line should be tattooed on every homelabber's forearm at initiation. I've watched someone confidently delete a share off a beautiful redundant array and then discover in real time that all six drives faithfully agreed the file should be gone. Redundancy and recoverability are different words for a reason.
greybeard_junior@greybeard_junior· 7/6/2026, 5:38:02 AMThe untested-restore point is the one that actually saves people. My rule is a scheduled restore drill — once a quarter I pick a random file and one whole service and rebuild them from the backups onto scratch storage. Half the value is confirming the backup works; the other half is that when the real disaster comes, I've already practised the recovery instead of learning it at 3 a.m. with my heart rate at 180.
checksum_charlie@checksum_charlie· 7/6/2026, 9:29:33 AMthe practice-run angle is underrated. the worst time to read your own restore docs for the first time is during the actual outage. muscle memory beats panic.
packet_loss@packet_loss· 7/6/2026, 7:35:44 AMadding the boring off-site option people overlook: a couple of rotating external drives, one always at a friend's place or a desk drawer at work. it's not sexy, it's not automated, but it's a genuine air-gapped off-site copy that ransomware physically cannot reach because it's unplugged in another building. sneakernet still counts.
kernel_of_truth@kernel_of_truth· 7/9/2026, 2:57:00 AMreviving this for the newcomers: the cheapest way to satisfy the 'one off-site' part of 3-2-1 without buying anything is an encrypted backup pushed to a remote you control or rent. encrypt it before it leaves your network so the remote host only ever holds ciphertext, and you've got a real off-site copy that survives your house burning down. do the encryption locally, always.