I have a NAS which gets backed up to the cloud every night. I have an MSDN subscription through work and I used the backup facility in that (as it’s free). At some stage I’m intending to move to a paid one as I don’t want to lose all my backups if I quit my job.
Oh yeah, if I had a beard it would be well and truly grey if the remaining hair on my head is anything to go by 
I still don’t think any personal Linux user these days is going to be backing up to tape 
Pretty sure floppies can’t get the job done. 
It’s only 3.5" anyway.
However, it is double density.
I get that writing to a non-native file system would not be as quick. If I were to try to send files from the Windows computer over to the ext4-formatted Linux partitions, I might get the same results. I just wasn’t expecting to see the same thing happen across the network. Nautilus is using the SMB protocol to do the transfer, but it should be Windows that actually writes the file onto the hard drive and updates the file allocation table, and that shouldn’t have the excessive delays I saw.
Are you hitting The he was word state where some shells try to do multiple copies in parallel and they’re all interrupting each other?
Our business machines are all VMs running under one cloud provider - which means they may or may not be "here’ or “there” and the VM infrastructure doesn’t care.
The data backups are in a different part of the same cloud provider, lessening the chance a crash of the machine will also take down the backups.
We also take backups of the machine images, too, and move them to another geographic region.
So, a theoretical meteor would take us down for hours but not permanently.
Personal backups are real-time by Carbonite. It wants to back up more than I do, but I stretch my gigabytes that way - I back up data, not the OS, and not executables.
I think a lot of Enterprise stuff is still local (as in, “in the same building”) with off-site deep storage.
Personally:
I have all the ‘important’ house machines back up to my NAS. (They’re mostly Macs, so have OK built-in… I don’t think the laptop I rebuilt for my wife gets used enough to matter). I try to set them to exclude apps and system folders: The key for me is data, and while rebuilding a box would be annoying, it’s not critical.
The NAS is redundant, and I have had it upload to the cloud, although that may be broken right now. Secure back-ups should be off-site, though, to be safer.
That’s just my 2 cents, of course, and there’s better solutions. For example with your specifics I’d almost suggest something like NextCloud if you cared (which I don’t think you do): It’s basically a mix of stuff akin to Dropbox and Google Cloud you can self-host on AWS or whatever so it’s more private but accessible while traveling. Scheduling, files, contact sync, etc. If you needed that, which again I don’t think you actually do.
I think everything of lasting importance is on my Google Drive and/or Dropbox. That said, it would be nice to have access to things I don’t bother to keep there when I’m overseas, for example, but it’s only a year or so. If it doesn’t fit on an external drive to come with me, it’s probably not critical.
We’re told the wi-fi is free and actually quite good, but compromised by foreign intelligence so using a VPN is strongly recommended. So that’s nice.
To save everyone else frustration, do not use the separate Nautilus program to transfer files to a Windows computer. The built in File Manager (Thunar) can handle it. You just put smb: in front of the network path.
I’m having to do this in pieces because of the slowdown I mentioned before, which is now also happening on the Windows computer when copying between partitions, even if they’re on separate physical drives. It must be something with the NTFS file system, regardless of which OS the drive is plugged into.
When I check the properties of several folders via Nautilus, it’s reporting a different amount of items (it doesn’t count folders) and a different total file size. In one case, it said the files were taking up 60GB more than they actually were. I don’t know how many days I spent re-copying files and comparing source to target trying to figure out why it wasn’t matching up.
Gah! It’s just Nautilus being stupid.
Yesterday I put in the Things that Suck thread that I had a hard drive on the Linux computer look like it died. The issue may have been the power supply because while I was doing further troubleshooting, the fan on the power supply started making a lot of noise and wasn’t turning very well, and then the second hard drive started acting up.
My thinking is that the voltage going to the drives went flakey at some point (two hard drives and an SD on the same cable) while that drive was writing data. During the troubleshooting, I had the drive removed from the case and picked it up with my hand. Sometimes I couldn’t tell if it was running.
I put in a spare PSU that I have. The drive’s behaving now mechanically. It’s mounting consistently where it wasn’t before, but it comes back with a kind of access or input/output error if I try to go into look at the files. Strangely, if I go into the /mnt folder and click on the folder with the drive name to get the Information (aka Properties), it will do a file count and show how much of the drive’s used and free. Those seem about right.
So, the files could still be there. I just can’t get at them. I’ve got gparted/gpart trying to repair the partition table or whatever’s making it grumpy. If that doesn’t work, there’s a couple of drive recovery programs I can try like SafeCopy.
Don’t go with the partition tool as your first port of call.
Try doing an ‘fsck’ (File System ChecK) to verify the FS integrity first.
When you do go to try and pull the files off, mount the drive read-only if possible.
So far, this isn’t going well. Fsck immediately warns that the boot drive is still mounted and file system damage will occur. I think I need that mounted so I can actually run it. Not risking that right now.
Redo Backup and Recovery sat there for about 6-8 hours with a spinning circle after I clicked on the backup option. I gave up. It seems to only be able to create .img or .iso disk images anyway.
I’m trying to figure out the command line options for Safecopy. That might work if it will do a file copy instead of making disk images.
DDRescue-GUI spent all of last night trying to create an image of the drive. It aborted because it ran out of room on the target drive, stopping at 2998GB on a 3000GB drive. I guess it doesn’t do any compression or skip sectors that never had any data on them. It also did something to the target disk that required reformatting to get it to work again.
As dumb as this sounds, I’m going to try recovering the Linux files with Windows. Restorer2000 Pro did well with recovering deleted files back on my Windows XP computer. The new Restorer Ultimate Professional supports Linux partitions. I needed a new version so I just bought it and I’m going to see what it can do. (Minor update: getting Windows to even detect the drive is the next challenge. It’s not showing up right now.)
I think I’ve recovered as much as I can. There’s a program called R-Undelete (aka R-Linux) that has a GUI and works the same as Restorer2000/Ultimate. The only thing that’s missing is browsing to select where you want the files recovered. You have to type in the path, such as /media/drive/folder.
I lost a lot of info. One or two folders were completely gone that had the most files. Maybe some of them are on an external drive or flash drive, but not all. Finding them will take a while.
The good news is I recovered probably the most important file and it’s a version from November. I may be able to recreate what’s missing since it’s fairly fresh in my mind.
Time to start looking at building a NAS.
bookmark post
thanks for this thread, will check it once in a while 
Ouch. Data loss always sucks.
Thanks, Viking. A majority of what’s gone is media files I can live without or would take a lot of work to fill back in what wasn’t recovered. But there’s some that are irreplaceable unless I happen to find a backup of them. The data’s been migrated through a few hard drives as I got newer computers.
I ran through the data recovery one more time. It didn’t find anything new but I had it try anyway. About two-thirds of the way through, I heard a kind of chirping noise every so often when I walked back by. I picked the drive up and I felt it vibrate for a second, stop, then the chirp happened again shortly thereafter.
That’s apparently what Seagate uses to indicate the motor’s not turning. Pulse, pause, chirp. It’s been off for about twelve hours now and might spin back up if I tried again. No need to try any more, though. Two passes with the recovery software tells me I’ve got all I can.
After moving the mouse to a different USB port, I began saving my work so I could do the updates. Maybe at that point, the motor had stopped turning or slowed way down and either trying to update the file scrambled something in the allocation table, or it’s something like Linux locks out a folder when it’s going to change info and it couldn’t figure out how to release it once the drive came back online.
As for the power supply, I’m retiring that, too. Replacing the fan’s easy, but it’s a 430 watt PSU. Just a little under-powered for my comfort. The kicker is there’s nothing on it to show who makes it. I had to look it up by the LPW2-23 model number, which is made by Linkworld Electronic Co. They’ve been around for over a decade and they’ve got a US branch that makes the newer/fancier AZZA PSUs, but come on, guys. No company name or brand name on what you sold just 3 years ago? That’s not a good way to build customer confidence in your products.