I’ve been sifting through a lot of backup solutions but they’re all incomplete or not clear in what they can do.
Would prefer free or open source ones like UrBackup but it’s not encrypted, or Duplicati but it doesn’t support making (mountable) sector-by-sector Disk Images.
What I have right now is a dumb NAS box to which I can push backups 24/7 over the internet, and some windows PC on that same network that can access that and should consolidate them. (because I won’t be uploading 1TB full images over the internet)
Is that a thing I could make work?
I could also buy a NAS Raspberry Pi (NextcloudPi) if linux helps, but I can’t yet afford the Threadripper 32GB ram 32TB ZFS FreeNas of my dreams .
Checking this thread on the urBackup forums, as well as the reddit link on duplicati, it seems there are a lot of people who can’t find a solution and these 2 are the best candidates they found.
It’s so sad I mean how do admins even do remote backups then? There are those per-GB paid proprietary cloud services sure but what the heck do all the Freenas and Owncloud people do for their off-site backups?
I don’t use freeNAS or Owncloud, but I use Duplicati to take file-level backups. If you do image-based backups you lose the benefits of incrementals.
I also backup my linux containers and those are effectively images, but of course not quiesced before backups so databases will need to do crash recovery and such.
Actually no. I’ve been using Acronis for a while for example and it can 100% do incremental or differential backups of whole drives, + encrypt them. But it doesn’t support your own server (only their clouds) and also it costs a ton of money
From what I read, you can rsync an image across and it’ll only transfer the changed bytes, but I don’t know how well that works. If rsync does it Duplicati would too, it’s a block-level incremental.
Hmm I would have to find a way to create a full image of my C drive while I’m running the OS, and have Duplicati always treat that image file as the same file so incremental backups work. Could be interesting.
Downsides:
-I’ll have to make a full image of C every time I backup (but send over the internet just the incremental differences)
-I won’t be able to store a copy of the incremental backups for C on my local machine
In the enterprise we have a $120K storage server from Dell EMC to store critical backups too, and a second one off site for the remote backups.
All end user work stations are not considered critical and are not backed up. Any important data that they need to store/share is hosted on a local file server; which is backed up weekly.
Company is also making its first shy attempts at using MS OneDrive for a few things.
If any new user needs a new pc for what ever reason, we just image a new one with SCCM and plop registry and environment variables on it.
That sounds like the worst you know, as opposed to the open source philosophy. It’s like hiring Cisco instead of just using PFSense for free - except I can’t find the PFSense equivalent in client/server backups yet.
Haha well this is a Fortune 500 company, $Pharma, which is happy to pay the bills. Also, its an AD infrastructure. Most of our networking shit is Cisco too.
Not to mention each Data center burns about $1M in power each month, and the dev team burns about $2M annually on Azure containers.
I tried to use UrBackup but it’s dumb: it won’t do backups locally first and then sync them. You can only set it up to have one server, and if that server is online it’s dumb because you need to run a different software / instance to do the same backup jobs again for your local onsite backups…
Duplicati can’t do system images and I heard it’s sometimes unreliable with huge files (need more investigation)
So the question remains: how the heck do people back up their 500 GB sector-by-sector drive images online to their own servers???
I’m in the same situation. Did you ever find a solution? I’ve been looking hard at UrBackup but I’ve read several reports about failed backups and the Windows client is NOT opensource.
Hey jeffshead. I tried many things. Freenas, OwnCloud, Openmediavault, looked into unraid, and all the backup software they support and the wikipedia feature tables. Unfortunately the least infuriating, least break-y, least clownshoes way for me was to use:
Just Syncthing on plain desktop Ubuntu or Fedora on a small “server”, to receive:
Paragon backup and Syncthing (and FreeFileSync) on windows.
Syncthing on phones.
As you can see it’s really terrible. And it works without any issue and the ux is: it just works on its own, or I press one button.
Worked for years, no problems, no changes.
no limitations: versioning, incremental backups, filtering, splitting, encryption, sector-by-sector, boot, self cleanup etc.
syncthing finds peers & ‘just works’ over any network config and vpn, no bs.
I don’t have to recompile any kernels to get past setup step 1
I don’t have to get a PhD in nebulous manuals just to get past step 2
I don’t have to spend weeks on terminals solving some niche of a niche of a dependency that’s ruining the whole stack.
I don’t restart the server one day and the OS gets f’d irreparably because of some bad pending update / routine
I have all the redundancy and validation I need and I can use NTFS everywhere which is natively supported by literally anything. No reiserfs, xfs, btrfs, whatever, which I’m sure are totally the second coming of christ, but I don’t want to have a brick of a HDD if the server stops booting.
This is a terrible and cynical response, and I truly hope I can be shown the light. I’m a programmer but I have a life and dayjob. No time for clownshoes or wizard hats. So I need something for backup & maintenance that is at least half way towards any actual existent ux and doesn’t funnel me into a dysfunctional feature-corner.
Did anyone here ever hear about restic? Planning to use it, can’t vouch on it, but people have been praising it all over the place (not on this forum though).
My backup server is an Odroid HC4 running NixOS. You can run just plain debian, but not sure about the ZFS DKMS support on it. If you want to run just a plain md raid1 or btrfs mirror, then debian is fine on it. And on top of Debian you can install OpenMediaVault, for easier management.
A backup server doesn’t need to be expensive. All it needs to do is be able to share NFS and / or SMB and you’re done. Restic takes care of the rest on the client-side. Deduplication and encryption happens on the client. Everything is only stored once on the backup server.
There are guides online, like from intermit.tech, on proxmox + restic + minio (s3-compatible object storage), but you can adapt it for just restic and a s3 storage (idk, DIY with minio or if you’re insane, ceph, or pick your poison among the providers, I think linode and digitalocean both offer s3 buckets). Or just run a home backup server, like I do on my hc4.
Yeah I got ubuntu (easy, most flexible, less dependable) and also OMV (harder, less flexible, more dependable but harder to fix) running on my Odroid.
Pretty nice overall for making shares, ran into some minor limitations of the os vs hardware architecture support. And (because SMB is M$ trash) it can be tricky to get SMB right in linux if you’re not lucky (e.g. user/pass, workgroups, talking to windows vs the smb linux integrations).
You still need Syncthing to do backup-over-internet (and not worry about how to find peers, auto-resume, working via your vpn of choice and ports etc). AFAIK/unfortunately there is no other online “backup” solution that can (re)connect like syncthing.
And, now I remember! I looked at Restic a while back (and now) and this was my experience:
visit website
see nothing but a command line and 0 (zero) concrete FEATURES listed.
– sector-by-sector? exclusions? encryption? incremental? self cleanup? what does the backup format look like? what happens if one shard/version gets corrupted? splitting? backup config-file profiles at least?
if I can’t see what it does, means sorry but HELL NO I ain’t putting those time-sink anti-ux clownshoes on! (I’ve been hurt too many times )
.
you know, as opposed to the open source philosophy. It’s like hiring Cisco instead of just using PFSense for free - except I can’t find the PFSense equivalent in client/server backups yet.
and doesn’t funnel me into a dysfunctional feature-corner.
)