I finally got around to setting up backups on my home server. Got borgmatic saving to a raspberry pi I have at my parents house with a big USB HDD attached.
Took me weeks of chipping away at it trying to get my head its yaml config, cron jobs and SSH keys but feels good once its all working!
I've actually been thinking more and more recently around setting up a local backup (Nas?) But haven't really done much research and the more I think about it, I wonder what I have to really backup (other than my photos which are stored on an external hard drive and online).
Just before I read this, I actually saw a nice setup using a rpi5 and some SSD's. Is using a rpi a good way to go about something like this?
Yeah, it probably is - in terms of low overheads and electricity. Not great for long term, as SSDs can fail spectacularly, but if it's a staging post for cloud syncing or another device elsewhere rather than the ultimate endpoint, that's probably good.
HDDs can fail too. I'd probably use a Raspberry Pi + SSD if putting it at someone else's house, since they will both be very quiet. I have a 14TB HDD at my house for backups, and it's pretty noisy.
If you really want redundancy you can use a mirrored array but for one backup of multiple I probably wouldn't, especially if there's a cloud backup in there.
I literally have no knowledge in this stuff. Once you have something like this setup, how does it work? Once you configure it, does it automatically backup files in a certain location on your computer?, or do you manually move files onto it that you want to backup?
Oh man I'm gonna show my lack of knowledge too 🙂. I'll try anyway, but take everything I say with a grain of salt.
There are different types of RAID, but I believe RAID 1 would be a mirrored array. This Lemmy instance is on a ZFS mirrored array, but that's handled by the host so I don't have to understand it 🙃. I am not sure if the ZFS mirror is considered RAID but it's basically the same thing as RAID 1 (perhaps one is hardware one is software based?)
I see this setup as more of a redundancy for a live system so it can continue with one of the drives after a failure without downtime, and I probably wouldn't consider the mirrored drive a backup. But I'm no expert!
Is failure in SSD's more common than HDD's? Ideally I'd want to use it as my "offline" storage option, using it instead of my current external HDD that I use.
Hmm that's pretty interesting. I'd always assumed (I guess they were talking about reliability alot when SSD's became widely distributed) that they were a lot more robust.
Was that Jeff Geerlings video on YouTube?
That was very cool! But using ssd's are not very cost effective...
I'm only using a pi4 as I already had it sitting around. If I was to buy something id get one that can take hdd's natively without using USB or a random adapter from some unknown company online
Yeah - I'd only seen a post on his Instagram at the time, but starting watching the Youtube video today whilst prepping for dinner lol. Havent yet finished the video though.
Do you have any recommendations as to what you might look at?
Its an Ubuntu server with a bunch of stuff in docker containers. The main thing I wanted to backup is my next cloud data which stated as just a hobby but now has the whole family using, and a Lot of audio books all hosted through audiobookshelf.
Ah yes, nextcloud would be important to back up. Audiobooks you can probably re-aquire but if people are storing personal stuff on Nextcloud you will want to make sure it's backed up. A common strategy is the 3 2 1 backup strategy. Have three copies of your data, two on different media, and one offsite.
It sounds like you have your original copy and your offsite copy now set up. I'd recommend having an extra backup, a backup of the nextcloud data on a different hard drive on the same machine (or another machine - but this copy should be on a different hard drive to the original copy). Personally I don't do anything fancy, I just run a cron job to bring down nextcloud (so nothing changes), and to run a script to zip up the nextcloud data (I use bind mounts for volumes so everything sites nicely in a nextcloud folder, including docker-compose.yml file, database, and nextcloud data), then I copy that zipped file to my backup drive with the date in the file name so I have multiple copies. I run this each night scheduled in the middle of the night, then have another script that trims the backups to keep 14 days of backups then a monthly backup indefinitely (always first of the month, for simplicity).
That sounds like a lot of disk space....
My next cloud is approaching 1tb so having 14 copy's is not going to fit... But yes I need to sort out a third copy some how.
Audio books have all been riped from family members audiable accounts, so I want to keep them safe. I can see Amazon patching the service to prevent this one day.
My nextcloud is not 1TB, as interestingly I don't really use if for files but far more so for the nextcloud apps and as a sync server for various other things.
The 14 copies probably are not needed. I do this because I have space for it (I think nextcloud backups are about 2GB each for me). In your case aiming for three is probably fine, certainly better than two. I'd likely only keep the three copies if mine were that large. You could also investigate backup options that only backup changes and not every file every day.
My main concern is if I accidentally deleted a bunch of files, then that deletion got synced to the other copies. That's why I zip up daily copies and keep them for a couple of weeks, just extra protection. Finding a good incremental backup tool (that let you roll back changes) would probably provide the same protection anyway, I haven't looked into this myself though.