Time Machine for OSX Lion 10.7 using Debian 6.0 Squeeze

My computer is an iMac and my server is an ubuntu server with a separate 2TB HDD in it for Timemachine.

Edited for spelling

There was a Time Capsule firmware + OS X update duo that fixed reliability issues somewhere in the Snow Leopard timeline. IIRC it added a bunch of "damage control/prevention" methods to AFP and HFS+ (backups being stored in a remotely accessed sparsebundle), aiming to bulletproof against flaky network situations and interruptions like sleep.

Hopefully OS X appears to degrade nicely in absence of those features, although you might not get all the safety net features.

I built an Debian armel netatalk package myself for fun about one year ago. It took "quite a while" to compile on my NSLU2, but AFP shares ended up working like a charm.

That said, the netatalk (and other vendors such as Drobo) have been pretty good at staying on top of the changes. You can safely ignore this as an issue.

Example: Time Machine uses a special file format called a "sparse bundle" that requires special HFS+ filesystem support. When the backup drive runs out of disk space, it starts removing "bands" from the bundle to free up disk space.

However, it at least used to be the case that this did not work on remote network volumes due to differences in the file sharing protocol or filesystem, and the result was that it would just sit there deleting bands over and over again until you had no more backup.

https://discussions.apple.com/thread/1221879?start=0&tst...

Check out that post, and scroll down to where it says: """Everything went fine until today, where finally the disk filled up... and I got this in the console""", and the epic journey of discovering he no longer has any backups.

This specific issue may be fixed. There was a comment made by someone later in the thread that this behavior was changed somewhat in Leopard in a way that made this specific issue no longer be the problem.

http://blog.cohen-rose.org/2008/04/time-machine-update-fixes...

However, do you really know yet? No. Do I honestly trust these people to have looked into the issue well enough to know what else might burn me? No. Could they demonstrate this trust? Actually: yes.

What I'd want to hear that they reverse engineered Time Machine sufficiently to determine how the check was implemented, and then went through the logic of the storage system to determine how it was storing things, proving to themselves that this specific alternative setup provided that, despite the check.

However, what we see here is just yet another article telling us nothing newer than we knew in 2007/8: that there is a setting you can turn off to allow Time Machine to store to drives it otherwise believes are unsupported... nothing new to see here.