You must log in or register to comment.
I have a RPi running nextcloud and a second RPi rsyncing all the files weekly from it, nothing off-site, but at least it’s two separate drives on different machines. Anything that I can improve here(cheaply)?
You might consider what you would do if your source has an issue that syncs the to your off-site copy. If it isn’t a lot of data, you might want to keep another copy or two in either location that is created at a less frequent schedule but would give you a fall back.
As an example, if your files got ransomware encrypted and then sync’d to the off-site location, how would you recover your data?
looks at my raid of HDDs backing up my raid of SSDs
also before everyone slaps me with “why aren’t you using zfs?” it’s because I keep swapping out drives and testing various file systems whenever they get new features, which mdadm will accept even with the most insane unreccomended setups.
Also using it in my mad pursuit to figure out what Stratis is supposed to be useful for.
I feel like we could get rid of #6 and let the others carry the message for it
Who is still using raid? It’s not 2006.
I have something resembling RAID5 in my NAS. 4 drives, 1 drive failure tolerance.
👋
raidz2 here. Still learning. What alternative should I look into?
Unless you want to go to large scale cluster storage systems, nothing.
Raid is the best way to get disk-level redundancy for a disk volume.
I’m mainly using RAID 6, but I’m still using a lot of SATA drives. I’ll probably need to go with one of the software raids, like z2 when I move away from SATA.
Raid is no longer viable as a performance component, but it is completely viable for redundancy. Large scale cluster storage like Ceph is the way forward for anything larger than what can fit in a single chassis, or a single disk controller. Basically if you have or need more than one 45 drive chassis for storage, look into Ceph.
For everything less, RAID, and if you don’t need redundancy and just want performance, just get a high end NVMe drive and do backups.
My house burned down right after building my first raid array. It hadn’t even been put into use. The plan was to move all the data from assorted servers, desktops and laptops in my house to the array THEN backup that volume to something offsite. /sigh
- RAID SHADOW LEGENDS
Also: don’t wait until you got the perfect setup. A bad/incomplete backup is better than no backup.
This is something I should be doing.
I need quite a lot of space for backups and I don’t have enough space for them. I should at least start with partial backups with whatever I can fit in the storage I have.
My weak point right now is off-site, and homelab. My homelab isn’t backed up at all, and my personal data is only backed up on-site.
It’s better than nothing, but I should be doing better. I work in IT afterall. I think this is in the same vein as the mechanics car…
THIS! RIGHT HERE!
When I was young and naive about digital things, I had NO BACKUP
One day I got a new laptop. Yay me. Transfer all the data from my old hard drive using some jank-ass local network setup because young and dumb about tech still.
Six months go by, and my new laptop shit itself. Still no idea what happened, but it BSODd and a factory reset got it working again.
I still had my old laptop, so after about a week of searching on forums and reading everything I could find about how to build a pc, how laptop internals compare, data transfers, and literally anything I could so I could pull the old hard drive out without damaging anything and get at least some of my data without issue…
I lost 6 months of new stuff on a much more capable laptop, but it’s better than losing EVERYTHING.
Experienced that recently. Every part of the backup was fine, except for the one with all the 2fa codes. Fun times.
Ooooooooooooooof
That stings
I like to annoy my IT friends by saying my backup strategy is chucking what little important data I have in my free Dropbox account.
It’s not even that important; I don’t care!
If you want to annoy them, tell them you just take snapshots as your backup strategy.
I mean sure you could be sunc’ing your snapshots elsewhere as a real backup strategy but you don’t need to tell them that.
Why does that annoy them? That’s an off-site backup.
For one it’s not a full, real backup strategy. That’s supposed to include multiple tiers.
Also it’s instantly synced, so if I bonk my stuff locally, it could be bonked over there and history might not be able to save me depending on the situation.
And I guess if Dropbox dies my data dies.
Some people take their data seriously enough to worry about that kind of stuff. I don’t.
As someone working in IT, I don’t care about your data either. Just don’t come crawling to me for help when it all goes wrong.
That being said, one drive/drop box/Google drive/whatever cloud storage… IMO, that’s fine for personal files.
I don’t personally like the “full disk” backups for personal stuff. It seems like massive overkill. Like, you’re backing up Windows and applications that are probably out of date, and stuff… Why? Plus restoring a full disk image to a bare-metal system is a massive pain in the backside. Unless it’s a server that needs to get up and running ASAP after a failure, just back up the important/unique files (generally the user folder), and if the worst happens, reinstall everything and restore the important files.
The only point on my IT approved backup list that you don’t meet the criteria for is incremental/historical snapshots or restore points. Bluntly, if you’re okay not having those and accepting the liability that if the files get deleted by accident using a legitimate method to delete them, then that’s your risk to accept.
None of what you’re doing ruffles my jimmies. As long as you’re making an informed decision, and accepting the risks, then the rest is on you. If you don’t care, then I certainly don’t care either.
least it’s better than using chess
“Son, the time has finally come. Today I’m going to teach you TNO (Trust No One) security.”
my two-year-old stares blankly at me
They’ll be ready to learn about Cryptography, Chains Of Trust and Two Channel Authentication by the age of 3!
Testing restore time is a key part of being a miracle worker. That way you can tell them it’ll take three times as long.
Need to factor in the buffer time
The amount of times we define a DRP procedure, testing times and so on, just to be told “it’s not THAT important” is staggering to say the least.
At that point I figure backups are the least of my worries. These people like to live on the EDGE
my backup is once in a while manually borging my files onto this chonk, but i have plans to improve on that
I have 8 of these chonks in a raid array. Am I backed up yet?
(No)
It INFURIATES me how many companies will spend money on backups, but not ever test that their backups restore or allow for continued functionality afterwards.
At one company, I banged this drum for years, and one day we had a situation where someone “accidentally” deleted all the media from a client website. I had to dig through several backups and rebuild from beta, which annoyed me endlessly, but I dropped the “I fucking told you so” several times, and hinted that our “restore scripts weren’t working as intended” to the client. It took me a full day to do what should have taken maybe 1-2 hours at most…
Alright. I have a confession to make…I actually DID make sure my eldest son heard ALL of these things… (And more, but wow…the WHOLE list…)
#proudNerdDadMoment
