10

I'm currently in the business of recovering data from a damaged---and dying---hard-drive (check this question for background info). The drive was dropped while the system was running; and while foolishly I was trying to run testdisk on it, I started hearing---what I assume is---the "click of death".

Now I have read in several places about the "freezer trick" (e.g. on AskDifferent, on Unix SE, on AskUbuntu). The crux of it is as follows (but do read the full instructions!):

  • Place the damaged hard drive inside of a ziploc freezer bag, so that no moisture gets in. (Use two or even three bags, and note that the bags do NOT need to be isotherm.)
  • Place the wrapped hard drive in the freezer
  • Leave the hard drive in the freezer for at least 12 hours
  • Then connect the drive to the computer and start copying data
  • At some point, the hard drive will fail again. When it does, redo the procedure until all data is copied or the drive dies.

(As per Harddrive in the freezer ever work for you? on SuperUser, two alternative-medicine remedies to try before freezing are: (1) "Place the hard drive on a smooth surface. Grab the ends and physically spin the whole unit around", "just spinning the unit sharply along the axis of the blades" and (2) "In situations where the head has crashed, you can temporarily get it up and running by turning the drive upsidedown. Keep the drive upsidedown while you back up your files.")

Question: Is freezing a legitimate strategy to rescue data from a dying hard-drive? Or is this simply a fancy way to definitively brick your disk?

And if this is legit, should one attempt it before trying ddrescue, after trying ddrescue, or should I combine the "freezer trick" with ddrescue (and an associated log file)?

  • This sounds like a quick way to break the read head motor if its frozen and it tries to move to read data. Haven't heard of this trick though. – Mgamerz Feb 1 '14 at 18:57
  • 2
    This seems off-topic to me. What does this have to do with unix? – Faheem Mitha Feb 1 '14 at 19:53
  • 2
    Well, I am trying to rescue a hard-drive from within Linux using Unix tools, and the present topic is part of the rescue process. Myriad other questions deemed on-topic deal with rescuing data from damaged hard-drives. And numerous answers involve "freezing". – landroni Feb 1 '14 at 19:55
  • It worked for me, although the problem with my drive started to appear after it crashed on the (hard) ground. I put it in a ziplock bag in the freezer and avoided any condensation on it. And now it works. Note that I used my freezer, not my refrigerator. Cheers – user96773 Jan 2 '15 at 13:52
7

I've done this with known good USB enclosures and operated the drive while it was inside the closed freezer (with my laptop on the kitchen counter).

It rarely helps, there are so many forms of failure, that it's a crapshoot. The Data Recovery experts tell you not to do this, but "seek professional help". How that help got to be professional is by reading and trying stuff with disks they didn't care about.

The most awesome recovery I managed to date was to open a drive and unstick the spindle with a screwdriver and pliers. Shockingly, it ran for a few hours while I recovered most of the data. I didn't even use a cleanroom. I did that after the freezer trick failed.

The most hopeless recovery I tried was a Mac disk where a head delaminated. There, running the disk or trying the freezer trick just continued to destroy the medium. A DR expert would have been able to do a partial recovery by sourcing identical drive mechanics and mechanically transfering the platters. My amateur efforts just destroyed the disk.

Not having the right tools can be a problem too. After my disk froze a second time, I tried to unstick it with the pliers and screwdriver again..... I slipped. The platter was hoplessly destroyed. The DR guys have proper tools to do this and minimize risk.

Best of all the DR guys have the experience to identify the problem without running the disk and listening to it click while your medium is scratched to hell.

If the data is worth anything, it needs to go to the DR guys. If not, the freezer is a good third-to-last effort.

After the freezer I would try the drop-test.

Then try a home-made cleanroom from plastic bags, tape and a clear box. Pop open the drive and see if the heads are gone, if htere are big scratches or if the spindle just won't turn. Be sure to have all your clean tools and your clean drive inside the box before you start. Also a cable to be able to connect the disk to your computer.

Most of this is just learning about disks at this point. Possibility of recovery is remote.

  • 1
    What about replacing the circuit board? Isn't this also something one should try before opening the hard disk? – Giorgio Feb 3 '14 at 11:51
  • 1
    Only if you have an identical board. IMHO, if you have a second identical disk, you're probably a professional, or you care enough about the data that you should send it to one. – mgjk Feb 3 '14 at 14:41
  • I have the same problem right now and I have been to a local shop. I asked them if they could replace the circuit board and they told me they don't: I'd better look for an identical disk on ebay and try it out myself. Even then, there is the chance that the information on the two boards does not match exactly. They estimate the success probability around 10%. I am now thinking about the freezer trick as a last option before saying good-bye to my hard disk. Luckily I have a partial back-up of the data (but a complete one would be better). – Giorgio Feb 3 '14 at 19:56
  • The manufacturers change chips on boards and often don't update their model numbers. Consider too shelving the drive until you find the parts or have the money to deal with it. Don't shelve it tooo long though, the parts get harder and harder to find. – mgjk Feb 4 '14 at 12:06
3

I've tried this technique on several occasions and it would buy me minutes of access time to a drive that was exhibiting the "clicking" behavior. Not really enough time to pull data of any consequence from it.

The other method I've not tried but might extend the amount of time, is to put the drive in a cooling container (small refrigerator perhaps) and then operate it from within this.

But based on my experience of trying the freezer method, I'd say this technique is a pipe dream and may have worked on prior technologies but with newer technologies the tolerances are orders of magnitude smaller given technological advancements, and the freezing the drive to get access to it is just not able to do so any longer.

As an additional data point I did dig up this link to a report put together by DRG - Data Recovery Group, titled: - FREEZING YOUR HARD DRIVE - A BAD IDEA. This report touches on most of the same issues I did plus some additional ones like humidity and such which will end up working against your recovery attempts.

References

  • "put the drive in a cooling container (small refrigerator perhaps) and then operate it from within this" This sounds like a more reasonable and less destructive way to approach this. Do you have any ideas how to go about this? I thought of putting the drive in the fridge, and operate it while it's in it (cable through the door); but I'm worrying about the humidity. I also thought of using a laptop cooling pad, but would it be sufficient cold and could the fan vibrations hurt? How else could you operate the drive from within a cooling container? What would be the "cooling container" of choice? – landroni Feb 2 '14 at 6:24
  • 1
    I've seen youtube videos showing various setups. When I contemplated doing the cooling container I conceived of putting the drive on a cable that would go into my freezer/refridg. and also put the drive in a ziploc bag and sealing it with either duct tape or rubberbands to try and keep as much of the humidity out. – slm Feb 2 '14 at 6:27
  • Right! So it's not an isotherm bag; it's a simple ziploc bag! I could even use two bags, both sealed with duct tape (including around the cable where it passes). Sounds like worth a try. – landroni Feb 2 '14 at 6:44
  • 1
    @landroni - Yes jus ziplocs. If I'd ever had any indications from the 24hr. freezer cycle that the drive(s) would let me get any data off of them I'd of gone with that route. But as I said, my experiences were at most 5 mins. or so of disk access and then it would stop working. – slm Feb 2 '14 at 6:46
  • At SuperUser there is a useful comment concerning operating/storing temperatures for the drive. – landroni Feb 2 '14 at 18:21
1

As always it depends of the failure. I tried the freezing method a couple of times successfully. The hard driver is getting hot about 50-80°C if you cool it down for about 10°C, some sectors can be successfully read. Modern hard drivers are reallocating failed sectors transparently. Therefore it is always good to analyse the status of your drive with smartctl (smartmontools).

  • I did try to run smartctl on the damaged drive (see UPDATE3), but I'm still hoping that someone could explain to me what it actually means. In short the self-assessment test FAILED! and the most glaring error message is 240 Head_Flying_Hours 0x0001 001 001 001 Pre-fail Offline FAILING_NOW 3. – landroni Feb 2 '14 at 10:46
  • please post the output of smartctl -H /dev/yourdisk, smartctl -A /dev/yourdisk – user55518 Feb 2 '14 at 11:08
  • I posted the requested output in UPDATE4 of How to repair a corrupted HFS+ partition from a damaged hard-disk?. What do you think? – landroni Feb 2 '14 at 13:36
  • 1
    I see. When read write heads touch the surface of the disk, then they are toss and twist, so no read is possible anymore. My condolences. – user55518 Feb 2 '14 at 18:38
  • 1
    There are a number of companies who offer repair. For you would be important to analyze if you really experienced a head crash. Head crash and How to recover – user55518 Feb 7 '14 at 15:30

protected by Community Feb 24 '15 at 19:18

Thank you for your interest in this question. Because it has attracted low-quality or spam answers that had to be removed, posting an answer now requires 10 reputation on this site (the association bonus does not count).

Would you like to answer one of these unanswered questions instead?

Not the answer you're looking for? Browse other questions tagged or ask your own question.