26 ms·

Ask HN: How do you do backups for personal/home server?

24 points by e3bc54b2 3y ago

jqpabc123 3y ago
I have an external RAID 1 hard drive that I backup everything to every 6 months or so. Before someone points it out, the individual drives in the array have build dates that differ by at least a year. In addition; for really critical stuff only (accounting data, deeds, business contracts, access codes, software source code, etc.) I use micro SD cards for more frequent backups as needed. I have a cheap digital watch with a SD card holder that I 3D printed and built into the strap. The SD drive is encrypted with Bitlocker and it travels with me wherever I go. Basically, this really critical data is as safe as I am. I always make 2 copies of the SD drive and keep one in a fire proof safe. The key to the safe is revealed in my will which my lawyer has. Inside the safe is the decryt key to the SD card. A trusted relative also has the decrypt key.
Zetobal 3y ago
I have a media nas at my parents location that are mirrored over night.
stamps 3y ago
I tried a bunch of different ways but ultimately settled on Duplicacy [0]. It runs inside a Docker container and backs up both my data as well as configurations like my docker compose file and smb.conf. Off site storage was Backblaze B2, but I moved to Hetzner. Likely will move back just because B2 is cheaper and a bit faster for my region. Another layer of backup I do is use Duplicacy to backup to a portable hard drive occasionally that I keep off site. [0] https://duplicacy.com/ https://duplicacy.com/
- browningstreet 3y ago
  I recently tried to cancel Backblaze and wanted to download some of the stuff I’d archived there. The download limits were a bit of a blocker. I re-upped for a year and decided I’ll deal with it over the coarse of a year but something to think about. I could upgrade to increase download limits but didn’t have the mental bandwidth (haha) to deal with it at the time.
jonas-w 3y ago
1. Raid1 (ZFS), if a drive fails 2. Borgbackup [0] with Borgmatic [1], daily backups to another server which also has Raid1 [0] https://www.borgbackup.org/ https://www.borgbackup.org/ [1] https://torsion.org/borgmatic/ https://torsion.org/borgmatic/
shawabawa3 3y ago
I use tarsnap and a cronjob I loaded up $10 credit like 6 years ago and it's still going somehow
techhazard 3y ago
Desktop, Laptop, NAS and Server all run on ZFS and push snapshots to an extra backup pool on the Server (colocated in a DC). Those backups are also pushed to an offsite location (with a hosting provider). Laptop and NAS run on a single disk rootfs, Desktop and Server run on mirrored nvme drives.
- egberts1 3y ago
  Am still intimidated by ZFS not being introduced into the Linux Kernel tree.
  - bombcar 3y ago
    It’s not a big deal if you have a distribution that helps, but then again I run Gentoo.
    - egberts1 3y ago
      good to know. Been running Gentoo since 2004. Last of the split-usr advocate due to embedded engineering and IoT.
      - bombcar 3y ago
        I run Gentoo because I prefer init.rd to what I like to call Systemd/Linux.
  - techhazard 3y ago
    I now run NixOS, so that’s a non-issue for me. But even so, it’s not hard, you have to install the ZFS package and make sure it’s included in your initramfs (if you use ZFS for / ). On arch I used the linux-lts kernel with ZFS and never had any issues in three years.
Zizizizz 3y ago
Restic + backblaze
ndreas 3y ago
I use borg to backup all my computers to my home server, including the server. It runs raid 1+0. I then use rclone to sync the borg repos to backblaze and dropbox. I encrypt both in the borg repo and in the rclone transfer.
FeelingGood 3y ago
Use a raspberry pi 4B with some external HDDs attached as my "NAS". The Pi is accessible everywhere i go using Tailscale as VPN for all my devices. For my PC, laptop, phone and such i use Resilio Sync to sync files between devices while also syncing to the Pi. The Pi creates an incremental backup using restic to Backblaze B2 every night.
rubatuga 3y ago
ZFS + rsync.net
bombcar 3y ago
I separate “important” data from “cache” data - the second being things like Linux ISOs and personal DVD rips that I could recover if needed. This reduces my backup set from tens of terabytes to less than one terabyte. And makes it easier to keep many more copies of the very special things (photographs, documents) on multiple devices and cloud services.
yuppie_scum 3y ago
Apple Time Capsules still work on 2022 M1 Macs. (Slow but they work) I got one on FB marketplace for cheap, i think I will put a faster spinning drive in there and should be good until apple kills support.
LinuxBender 3y ago
On individual nodes I keep data in specific directories away from the OS. I rsync those directories to a machine that acts like a NAS. That machine uses rsnapshot to have multiple file level diffs uses hardlinks to save space on dupe files. I back that up to multiple external SSD USB drives. I have one drive in my vehicle. Encryption is just plain dm-crypt with non default cipher/hash, no LUKS headers.
TheLoafOfBread 3y ago
Syncthing all my computers to one and then robocopy on external NAS for cold storage. Repeat once a month.
fdw 3y ago
I do backups to a local, external HD (that I need to manually plug and unplug, unfortunately), and to Backblaze. Depending on the importance of the data, it's being pushed daily to B2 and weekly to HD, or more seldom, or only to HD. Technology-wise, I've been on restic for some years now, because of its support for cloud backends. (Otherwise, Borg is a fine tool, too.) Until last week, I've used crestic, but I just migrated to resticprofile, as that can do web hooks and pre-/post-commands, which I intend to use for SQL server dumps. restic: https://github.com/restic/restic/ https://github.com/restic/restic/ crestic: https://github.com/nils-werner/crestic https://github.com/nils-werner/crestic resticprofile: https://github.com/creativeprojects/resticprofile/ https://github.com/creativeprojects/resticprofile/
OnlyMortal 3y ago
Apple TimeMachine to S3. That’s right… S3. Edit: to be clearer it’s something I wrote myself from scratch.
znpy 3y ago
My home server is a small machine with two disks, one for the os and one for data. The disk for data is using zfs, and i take hourly snapshots between 7am and 11pm. I keep two weeks worth of snapshots for each dataset. I have another small machine (a nuc) with a similar setup at in law’s house, connected to my vpn, and I replicate snapshots over there through the vpn. Worst case scenario i can go visit the in law’s and get my snapshots. Zfs is really awesome. Edit: everything is scripted and ran via cronjobs, of course.
electroagenda 3y ago
I use a NAS. It is synchronized with Google Drive. So I can access my files with the peace in mind there is a back up.

24 points by e3bc54b2 3y ago

I have an external RAID 1 hard drive that I backup everything to every 6 months or so. Before someone points it out, the individual drives in the array have build dates that differ by at least a year. In addition; for really critical stuff only (accounting data, deeds, business contracts, access codes, software source code, etc.) I use micro SD cards for more frequent backups as needed. I have a cheap digital watch with a SD card holder that I 3D printed and built into the strap. The SD drive is encrypted with Bitlocker and it travels with me wherever I go. Basically, this really critical data is as safe as I am. I always make 2 copies of the SD drive and keep one in a fire proof safe. The key to the safe is revealed in my will which my lawyer has. Inside the safe is the decryt key to the SD card. A trusted relative also has the decrypt key.

Zetobal 3y ago

I have a media nas at my parents location that are mirrored over night.

stamps 3y ago

I tried a bunch of different ways but ultimately settled on Duplicacy [0]. It runs inside a Docker container and backs up both my data as well as configurations like my docker compose file and smb.conf. Off site storage was Backblaze B2, but I moved to Hetzner. Likely will move back just because B2 is cheaper and a bit faster for my region. Another layer of backup I do is use Duplicacy to backup to a portable hard drive occasionally that I keep off site. [0] https://duplicacy.com/ https://duplicacy.com/

browningstreet 3y ago
I recently tried to cancel Backblaze and wanted to download some of the stuff I’d archived there. The download limits were a bit of a blocker. I re-upped for a year and decided I’ll deal with it over the coarse of a year but something to think about. I could upgrade to increase download limits but didn’t have the mental bandwidth (haha) to deal with it at the time.

jonas-w 3y ago

1. Raid1 (ZFS), if a drive fails 2. Borgbackup [0] with Borgmatic [1], daily backups to another server which also has Raid1 [0] https://www.borgbackup.org/ https://www.borgbackup.org/ [1] https://torsion.org/borgmatic/ https://torsion.org/borgmatic/

shawabawa3 3y ago

I use tarsnap and a cronjob I loaded up $10 credit like 6 years ago and it's still going somehow

techhazard 3y ago

Desktop, Laptop, NAS and Server all run on ZFS and push snapshots to an extra backup pool on the Server (colocated in a DC). Those backups are also pushed to an offsite location (with a hosting provider). Laptop and NAS run on a single disk rootfs, Desktop and Server run on mirrored nvme drives.

egberts1 3y ago
Am still intimidated by ZFS not being introduced into the Linux Kernel tree.
- bombcar 3y ago
  It’s not a big deal if you have a distribution that helps, but then again I run Gentoo.
  - egberts1 3y ago
    good to know. Been running Gentoo since 2004. Last of the split-usr advocate due to embedded engineering and IoT.
    - bombcar 3y ago
      I run Gentoo because I prefer init.rd to what I like to call Systemd/Linux.
- techhazard 3y ago
  I now run NixOS, so that’s a non-issue for me. But even so, it’s not hard, you have to install the ZFS package and make sure it’s included in your initramfs (if you use ZFS for / ). On arch I used the linux-lts kernel with ZFS and never had any issues in three years.

Zizizizz 3y ago

Restic + backblaze

ndreas 3y ago

I use borg to backup all my computers to my home server, including the server. It runs raid 1+0. I then use rclone to sync the borg repos to backblaze and dropbox. I encrypt both in the borg repo and in the rclone transfer.

FeelingGood 3y ago

Use a raspberry pi 4B with some external HDDs attached as my "NAS". The Pi is accessible everywhere i go using Tailscale as VPN for all my devices. For my PC, laptop, phone and such i use Resilio Sync to sync files between devices while also syncing to the Pi. The Pi creates an incremental backup using restic to Backblaze B2 every night.

rubatuga 3y ago

ZFS + rsync.net

bombcar 3y ago

I separate “important” data from “cache” data - the second being things like Linux ISOs and personal DVD rips that I could recover if needed. This reduces my backup set from tens of terabytes to less than one terabyte. And makes it easier to keep many more copies of the very special things (photographs, documents) on multiple devices and cloud services.

yuppie_scum 3y ago

Apple Time Capsules still work on 2022 M1 Macs. (Slow but they work) I got one on FB marketplace for cheap, i think I will put a faster spinning drive in there and should be good until apple kills support.

LinuxBender 3y ago

On individual nodes I keep data in specific directories away from the OS. I rsync those directories to a machine that acts like a NAS. That machine uses rsnapshot to have multiple file level diffs uses hardlinks to save space on dupe files. I back that up to multiple external SSD USB drives. I have one drive in my vehicle. Encryption is just plain dm-crypt with non default cipher/hash, no LUKS headers.

TheLoafOfBread 3y ago

Syncthing all my computers to one and then robocopy on external NAS for cold storage. Repeat once a month.

fdw 3y ago

I do backups to a local, external HD (that I need to manually plug and unplug, unfortunately), and to Backblaze. Depending on the importance of the data, it's being pushed daily to B2 and weekly to HD, or more seldom, or only to HD. Technology-wise, I've been on restic for some years now, because of its support for cloud backends. (Otherwise, Borg is a fine tool, too.) Until last week, I've used crestic, but I just migrated to resticprofile, as that can do web hooks and pre-/post-commands, which I intend to use for SQL server dumps. restic: https://github.com/restic/restic/ https://github.com/restic/restic/ crestic: https://github.com/nils-werner/crestic https://github.com/nils-werner/crestic resticprofile: https://github.com/creativeprojects/resticprofile/ https://github.com/creativeprojects/resticprofile/

OnlyMortal 3y ago

Apple TimeMachine to S3. That’s right… S3. Edit: to be clearer it’s something I wrote myself from scratch.

znpy 3y ago

My home server is a small machine with two disks, one for the os and one for data. The disk for data is using zfs, and i take hourly snapshots between 7am and 11pm. I keep two weeks worth of snapshots for each dataset. I have another small machine (a nuc) with a similar setup at in law’s house, connected to my vpn, and I replicate snapshots over there through the vpn. Worst case scenario i can go visit the in law’s and get my snapshots. Zfs is really awesome. Edit: everything is scripted and ran via cronjobs, of course.

electroagenda 3y ago

I use a NAS. It is synchronized with Google Drive. So I can access my files with the peace in mind there is a back up.