The Great Leopard Time Machine External Drive Failure Mystery

Dear Forum Members,

Once bitten, twice shy, they say – so, as I await my refund from the Apple Online Store and contemplate what to buy as a replacement, I wanted to post this thread and a try to get an answer to the following question:

“Is pretty much any/every external drive attached to a newer Mac running Leopard with Time Machine at serious risk of failing without notice?”

This seems quite broad, but I have spent a weekend’s worth of spare time scouring the web and it seems to me there this something very very wrong with a large number of combinations of Macs, Leopard and External drives when used with Time Machine. Given that Apple has been pushing this combo as a great way to safe data heaven – it seems only fair we’re informed if it is dangerous, and better yet that a patch is around the corner.

Specifics of my own issue: new iMac 20” (latest iteration, firmware up-to-date), OSX 10.5.7. Pretty much standard configuration – not messed with/tweaked – usually left on i.e. std power management does its thing. Purchased: WD My Book Studio 1Tb (pre-formatted HFS+). Plugged in via Firewire 800, turned power on, recognised by Mac, Time Machine sees the drive pops up & is turned on. Everything seems fine – Mac and drive appear to be working together well and the drive knows when to spin down and wake up in concert with Mac; backups happen. Two weeks later – Kaboom! Time Machine locked in perpetual “preparing” loop for 24+ hours – only solution is to shut down (but no force quit required). Drive spins down after Mac is turned off – unplugged from power, re-plugged – spins up. Mac back on – drive spins up and the first error hits: ”The disk you inserted was not readable by this computer”, options to Initialize, Ignore or Eject. I Eject and reconnect. Same. I Ignore and try to locate in Finder – nada. Power everything down unplug, replug, same again. Since the disk contains nothing but Time Machine back-ups I can live without (new Mac, not really used yet) I try Disk Utility – it can see the drive (or at least the controller) but cannot mount. All First Aid options greyed out. Erase and Partition both return “Resource busy” errors. I try fiddling with the power management to make sure it’s not being sent to sleep, then a variety of combinations of hardware and software resets, with all connection combinations open to me (800/USB). Nothing. Finally try the drive in Vista, then XP. Comprehensively dead – various disk utilities get nowhere past the controller, Old Skool Chkdsk tells me drive is not ready. It spins, the pretty WD lights flicker, buy no one is home..

Causes hinted at across the blogosphere (and my comments):

• WD drives are pants: that’s certainly one opinion, and it would be fair to say that a disproportionate number of identical or similar issues appear to be with My Books. It is also the case that the drive is too clever by half, with firmware-based power management, a soft on/off button and multi-function LED status meter. It feels a lot like a PC product that has been adapted for Mac – always a risky business.
• Controller chipsets – some appear not standards compliant / compatible with newer Macs, or are otherwise unstable across multiple interfaces (seems particular prevalent across quad interface implementations). Too techy for me – but since there aren’t millions of these you’d think culprits would be spotted.
• Time Machine is pants: another possible opinion – but it’s also quite clever and a linchpin of Leopard marketing – would Apple really not have spotted for two years that it kills drives indiscriminately? There are instances of this happening back to 2007.
• Power Management turns off drives in the middle of backing up: a definite possibility – but again, this is something so obvious that, if it were the issue, the system should alter the configuration once you turn on TM.
• Sleep mode interrupts TM / knackers drive: as above.
• People pulling out cables without ejecting: agree is never helpful – but mine was just sitting there. I should not have to manually spin down, eject and power off external HDD before hitting ‘shut down’ on the Mac. I didn’t have to on my Amiga 20Mb HDD – and that was 18 years ago.
• Use GUID instead of APM – possibly good point, but if it was that simple surely OSX or HDD makers could detect your configuration and suggest a reformat when you connect.

I am not expecting a ‘cure’ for my particular drive – it’s a goner and on its way to Drive Hell. I know there are plenty of ideas around (it might be quite helpful if any of you that have saved the best forum / tech support links in relation to this issue to repeat them in one post so they’re in one place). What I want is some root cause analysis and someone to own up this is a problem, then commit to fixing it – because I’m miffed and do not feel like loaning the Apple Store $2/300 again for a few months before another one fails. Stuff sold on the Apple Store – Apple or third-party – should work with Macs reliably (that means no more than 1 failure in 100k in Year 1, 1 in 10k in Y2, 1 in 1k in Year 3 etc btw).

Apologies for the essay length!

Not having a go at TM - just exploring possibility of a bug existing

maflynn et al (yeah, sorry, haven't quite mastered the art of the short post yet) to be clear, this is not a go at TM, am suggesting that in a small but signifcant number of circumstances operation of TM is the trigger, but as per my list it may or may not be the actual cause i.e. could be that it does the voodoo it do but interaction with some other varible (e.g. FW chipset, power mgt) kills the drive.

Really not the first post by a long shot, not even on MacRumors; google any variation on "leopard time machine external drive failures / errors" and you'll see 100s of posts going back to 2007. Not suggesting that TM isn't predominantly successful / stable - clearly it is otherwise there wouldn't be hundreds of post but 10s of thousands.

Willing to accept I may just have gotten a duff drive, it happens - but right now there's enough 'chatter' to explore other reasons. And the whole point about an as yet unidentified bug is that it is far fetched - otherwise routine regression testing would have picked it up prior to release.