3 ms·

While I appreciate the fun of getting this to work, overall, I really wonder whether saving $300 for a Time Capsule is worth it if it means that your backup sol

by pilif 15y ago

While I appreciate the fun of getting this to work, overall, I really wonder whether saving $300 for a Time Capsule is worth it if it means that your backup solution is now relying on you typing a command like defaults write com.apple.systempreferences TMShowUnsupportedNetworkVolumes 1 The "unsupported" would give me pause when setting up a solution I will need to be there when the worst happens. There must be a reason for this preference to be there. Granted, it could just be Apple wanting to sell their products and thus doing some public key magic for it to work without the switch. Or it could be that TM relies on some very rarely used feature in AFP which is not or at least not perfectly implemented in netatalk. Are you willing to bet your backup on this flag being needed for the former reason and not the latter?

g_lined 15y ago
You're quite write that it should give you pause! I have, however, used this flag for 2 years now and it does work very well and have had no problems at all and that's over 2 versions of OS X. My computer is an iMac and my server is an ubuntu server with a separate 2TB HDD in it for Timemachine. Edited for spelling
lloeki 15y ago
> Or it could be that TM relies on some very rarely used feature in AFP which is not or at least not perfectly implemented in netatalk. 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.
latchkey 15y ago
Apple put that in there so that they could sell more TC's. They also seem to make backwards incompatibile changes to the protocol for every OS release. 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.
- wooster 15y ago
  > Apple put that in there so that they could sell more TC's. Not only is that pure speculation on your part, it's also not true.
  - latchkey 15y ago
    Ok, cool, you worked on TM. Care to explain why Apple purposely made it difficult to setup TM using any other third party product other than a TimeCapsule? 'Because we didn't want to get support issues'?
    - wooster 15y ago
      We needed certain guarantees on the remote end in order to ensure reliable backups, that's all.
saurik 15y ago
Agree. As well, a lot of the people who report on things like this "jump the gun" by seeing a few successful runs where it "worked for a while", and then don't realize the consequences. 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&tstart=0 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-sparse-bundle.html 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.
gsiener 15y ago
Well, to be fair, I've been dealing with a fubar'd system since Monday that is all Apple -- macbook air, time capsule for backup, full disk encryption w/ Lion. Apparently there's a huge bug that prevents the PGP key from properly mounting the drive on reboot (as of 10.7.2). Restoring from time machine AND doing a fresh lion install have both failed, so I'm not convinced the "Supported" flag helps much :/

by pilif 15y ago

g_lined 15y ago

You're quite write that it should give you pause! I have, however, used this flag for 2 years now and it does work very well and have had no problems at all and that's over 2 versions of OS X. My computer is an iMac and my server is an ubuntu server with a separate 2TB HDD in it for Timemachine. Edited for spelling

lloeki 15y ago

> Or it could be that TM relies on some very rarely used feature in AFP which is not or at least not perfectly implemented in netatalk. 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.

latchkey 15y ago

Apple put that in there so that they could sell more TC's. They also seem to make backwards incompatibile changes to the protocol for every OS release. 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.

wooster 15y ago
> Apple put that in there so that they could sell more TC's. Not only is that pure speculation on your part, it's also not true.
- latchkey 15y ago
  Ok, cool, you worked on TM. Care to explain why Apple purposely made it difficult to setup TM using any other third party product other than a TimeCapsule? 'Because we didn't want to get support issues'?
  - wooster 15y ago
    We needed certain guarantees on the remote end in order to ensure reliable backups, that's all.

saurik 15y ago

Agree. As well, a lot of the people who report on things like this "jump the gun" by seeing a few successful runs where it "worked for a while", and then don't realize the consequences. 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&tstart=0 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-sparse-bundle.html 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.

gsiener 15y ago

Well, to be fair, I've been dealing with a fubar'd system since Monday that is all Apple -- macbook air, time capsule for backup, full disk encryption w/ Lion. Apparently there's a huge bug that prevents the PGP key from properly mounting the drive on reboot (as of 10.7.2). Restoring from time machine AND doing a fresh lion install have both failed, so I'm not convinced the "Supported" flag helps much :/