Warning that this doesn't detect errors properly. Reporting that it successfully backed up when it definitely did not.
> command.run(job)
File "/Library/Python/2.7/site-packages/tarsnapper-0.2.1-py2.7.egg/tarsnapper/script.py", line 347, in run
> OSError: [Errno 2] No such file or directory
Has anybody ever used Tarsnap to back up Time Machine sparseimages? I've wanted to try it out but I'm not sure if Tarsnap will handle the de-dup on all of the Time Machine snapshots properly. Will I end up with just the original plus deltas or hundreds of copies of the original?
Short answer: Tarsnap can back up a sparse image without messing up any de-dup properties of the contents of the sparse image.
I haven't tested Tarsnap with my Time Machine backups yet, but I've used non-Mac file systems as file servers to Mac OS X clients for more than 10 years now.
The only special bit of Time Machine backups is the hard-links to directories. This requires HFS+.
A sparseimage that is an image of an HFS+ file system can use that file system to hold Time Machine backups.
Such a sparse image can be copied to a non-HFS+ file system, without affecting the Time Machine backups on the sparse image.
Using a "sparsebundle" instead of a single-file sparse image will let you use non-Mac file systems such as FAT32 that limit the file size to less than 2 GB.
Tarsnap is really great service. It's obviously targeted at tech people, but if you are one, just check it out. I do not feel the need for any wrappers as the CLI is quite nice, but your mileage may vary.
What I'm backing up are a few true crypt containers. Not out of paranoia (well...), but because I use TC anyway to separate different kinds of context (by usage and importance). The nice thing is that incremental backups feature still works great. Compression is not really useful though. Currently I have ~70gb of data, ~20gb being unique.
The service is also priced quite reasonably. Highly recommended.
As a newcomer to the tarsnap family (started using a month ago)- I completely agree - I use Dropbox, Super Duper, Arq, Crashplan for various types of backup scenarios - but the tools that I use to snapshot my ultra critical (all Customer content - code, scripts, visio diagrams, network configs, site/RF surveys) 677 MB "customers" folder is tarsnap. It really is as simple as:
Oh, that's neat. I was on the lookout for a Arq-like client for Tarsnap, this looks promising and I'll definitely give it a shot. At first I thought that's the one being posted on the Tarsnap mailing list [0] lately but it's not and looks easier to install than the other one.
I fought the urge to want a GUI for Tarsnap and hand-wrote cronjob scripts that I'm very happy with now. It was part of the fun. Definitely non-ideal for everyone but I recommend it to any hackers to attempt the same.
One thing I learned as a result is becoming intimately with the unix/linux folder structure and being able to backup an entire OS deployment, reinstall the OS and just pull an old backup to recreate most of the important configs.
There might be a reason why there are a lack of good scripts, even on Github.
Most tarsnap wrappers do not add any special functionality. You can find hundred of tarsnap cron scripts or bash wrappers, or python wrappers... But truth is that they don't add nothing special. Most people using tarsnap can roll their own personalized solution easily if they require anything more than the binary offers.
> command.run(job) File "/Library/Python/2.7/site-packages/tarsnapper-0.2.1-py2.7.egg/tarsnapper/script.py", line 347, in run > OSError: [Errno 2] No such file or directory