Hello, so, I have been self-hosting some basic stuff recently, including data storage so i don’t have to rely on external services like google drive.
It’s working fine, but I wondered what would be the best backup solutions in case something unexpected and unfortunate happens (accidentally wipe out everything, drives dying, electrical issues, house burning down, that sort of thing).
I was wondering if more experienced self-hosters had recommendations about that ?
Maybe storing a physical drive in an especially sturdy box ? Perhaps using distant cold storage solutions ? Or even something I have never heard of ?
In addition to the recommendations you’re getting already, I recommend learning about Borg Backup. It has great documentation.
Borgmatic is a project that makes automatic backups easier with Borg.
https://torsion.org/borgmatic/
If you want a GUI for your personal machines there is Vorta.
I’ve used Borg for years now and test my restores every year. Which reminds me… I should do that more often. 😅
while borgbackup is superior I’d like to hear your thoughts on that whole restic thing, e.g. backrest.
i have one hard drive in an old pc in my basement, my backup is my willful ignorance
fr tho i plan on just setting up a sync with my friends server, the data isnt important at all lol. will be getting more drives soon dw :)
Depends on your paranoia/fault tolerance level. In general some form of the 3-2-1 backup rule. Personally I use arqbackup:
- Local live copy
- Hourly differential backups to another network share
- Hourly differential backups to an s3 compatible bucket with object lock and versioning so malware can’t wipe the latest 3 months or so.
Edit: I also forgot, 2 and 3 don’t exist unless you’ve verified that you can restore from them recently.
Your situation sounds like a two server solution for local. So one server for hypervisor/vms and then snapshots and backups go to a separate box like a NAS. As for “house burning down”, a solution for that is off-site backups. I’m guessing building a small TrueNAS server and installing it at a friend’s house or your parents or whatever and then find a backup solution to sync(syncthing may be an answer here for you?).
I don’t care about my homelab much, but I do care about my family photos. For that I follow my own 3-2-1 where:
3 copies of my data
2 copies are local
1 copy is off-site
I have a NAS at my house and another NAS at my parents house. They are both linked with syncthing and I do a one-way backup to the other NAS. Now, my parents are a 10 minutes away by car, so I consider that NAS “local”.
And then I backup my NAS to backblaze for my off-site backup.
So, setting up an entire server might be overkill for the amount of data I truly want to never lose (passwords, some pictures, some important documents, maybe some music if there is space left) , but asking a relative i see every once in a while to just keep a drive or even a USB stick in a desk somewhere is probably the easiest and reliable option.
Thanks for the suggestion !
Remember you aren’t just backing up the data but also backing up the hours of effort it takes to rebuild and get it how you want. If frequently backing up just the service OS you can store heaps for $2 on offerings like OVH cold archive.
Remove yourself from the backup you’ll forget or it will be inconvenient the week it blows up, so automate it, check the automation monthly. Don’t care that the 2nd cold backup takes ages if you have a quicker main backup.
Fireproof safes aren’t melt heatproof. Don’t rely on a local house backups for fireproofing.
I’m a self hoster, and hate subscription services but I believe cloud storage for use with a compressed encrypted backup makes sense.
Backup media and other stuff separately to avoid one large slow monolithic backup.
- Hourly snapshots using btrbk
- Daily local backup to a NAS, also with btrbk (note: requires btrfs on both sender and receiver systems)
- I’m currently setting up a remote backup solution using borg to the NAS of a relative
I’d consider paper (physical) backups for essential passwords and keys, but be careful about security.
3 methods for me…
- (daily) live nas backup (I have 2 of these mirrored for giggles)
- (90 days) cold, manual copy of nas data to drive stored in pelican case
- (live) cheap cloud storage of essential data in backblaze encrypted vault (duplicati)
I wouldn’t trust the disconnected drives. They fail more often when offline than on, in my experience.
Granted it’s your 3rd backup, so it’s a smaller risk.
i spose…
my anecdote;
ive been runnin drives like this for 30 years. ive literally never had a cold storage drive go bad out of the case… mostly because their hourly use rate is kept so incredibly low. even back in the day of very unreliable drives…
they even retain their data better than modern ssds
It’s a bit difficult. I don’t have the money for an entire 2nd server on my network and $500 in HDDs just for a backup solution as part of 3/2/1.
I have 3TB of fault-tolerant-ish data in a ZFS mirror then 12TB in a third, single drive full of stuff that I don’t care a ton if I lost (media and stuff mostly)
Maybe I could back up the more needed data to Hetzner or something for cheaper, but it still adds up.
Veeam VBR (+NFR license)
Backs my
Windows PC
Several ProxMox VMs
Linux serversto my NAS (NFS TrueNAS)
and on occassion an external HDD I keep at work.
1 of which is at home, the other at my work desk.I like the oldschool way of tar, rsync and cron
Depends on what you’re using as a platform. You should use something that integrates with it, like proxmox PVE and PBS
I have 2 powered RAID enclosures from icybox with 2 multi TB HDD in each one. The RAID is set to mirror the drives.
They are connected via usb 3 to a raspberry pi which runs borgbackup.
One is in my own place, just next to the main server.
The other is at my parents place in another city.
All my desktops, laptops and servers have borgmatic installed with the two pi’s as two targets. So when I create a backup it gets send to both locations. On my PCs I manually do a borg backup like once a month when I feel like it. The server computers are all on a daily schedule.
Borg has extremely efficient compression and defuplication. So having 20 historical snapshots of the whole file storage of each device takes about 30% less space than the original size on disk.
For example my desktop currently uses ~800GB but the borgbackup of said desktop takes only ~500GB.
The only disadvantage I find is that there is no cross system deduplication.
The super useful advantage is that I was able to just take the HDD enclosure, plug it into my Desktop and restore whatever files I want. I did an rsync to a blank fs once and it restored everything properly. And it’s pretty cheap. like 150$ total per backup location without any significant monthly costs.
I used hetzners storagebox for a while for borgbackups but restoring from it was SO SLOW. And my internet connection is not stable enough to do that without interruptions for multiple days. Never again, except for using it as an extra last resort “cloud” backup.
The only disadvantage I find is that there is no cross system deduplication.
You could achieve this by having all machines write to a single Borg repository, where everything would get deduplicated. But downsides include: 1. You lose everything if something goes wrong with that one repo, and 2. You’d have to schedule backups across all systems so as not to run at the same time, because the single repo can only have a single writer at once.
I tried that once and it takes way longer to run a backup. I forgot ehy exactly, something with running the comparisons against everything.
It makes a certain amount of sense. More deduplication means more CPU (and IO) spent on that work.
My server takes weekly backups via proxmox that are pushed to an NAS. The NAS backups and some additional files are copied up to filen.io for cloud storage. Probably not as professional as many of the setups you will see but it works for me.
Anyone try Peergos ?
