1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Need (simple) help with WinMFS

Discussion in 'TiVo Underground' started by The Great Inert, Jun 21, 2012.

  1. unitron

    unitron Active Member

    16,387
    2
    Apr 28, 2006
    semi-coastal NC
    "I'm hoping to see something that indicates that files are being copied soon..."

    You won't exactly, even if it is, but I think you need to type

    CTRL+C

    to interupt it and get it to stop. We'll take a different approach.


    "How does the program indicate that it's copying files?"


    Actually it doesn't.


    It's not set up to copy files, it copies the first x number of bytes and writes them to the corresponding location on the target drive.

    Then it does it again, with the next x number of bytes.

    If those bytes are a file, or part of one, eventually you have files copied.

    If those bytes are a bunch of zeros, then it copies and writes zeros.

    (which is why one should be very careful about using /dev/zero with

    dd

    or any of its derivitives)

    If they're goobledygook, then that's what gets read from the source and written to the target.

    And if those bytes are stuff like partition boundaries and boot records, it copies them as well, which is what lets it "Xerox" drives regardless of operating system or file format.

    To it, the entire drive is a file, although, if desired, you can set it to only read part of that file--any part you can provide a starting address and length for.



    find a man page for

    dd_rescue

    and learn how to set it to copy only 512 bytes at a time, with a fallback to 1 byte.

    Assuming both drives have the same LBA number, also learn how to make the program work in reverse,i.e., start at the end of the drives and work towards the front.

    Then wrap paper towel around the bad drive after it's cooled, and stick it in the freezer overnight.

    The paper towel will let you pick it up the next day without losing skin and will keep humidity from condensing on it.

    You'll need to slip the towel off of it before applying electricity, because at that point you don't want insulation.

    You might even turn the drive upside down to place the flat top on something that can be a heatsink, like block of metal or marble.

    Then you try to run

    dd_rescue

    in reverse, but only if the LBA numbers match, or things won't be in the right place,


    and you run it taking very small nips at the source.

    If it doesn't get it the first time it goes back and tries again.

    Doing so in smaller samples increases the chance of getting a good read.

    Having had to have done this once, I can assure you it will take forever, 24 to 48 hours at that low speed.

    But having done it once successfully, I can recommend it if the stuff on there needs saving badly enough.

    The nice thing about working backwards is that it'll probably get a bunch of sectors copied before it has to slow down and retry a lot.

    You should probably find a way to help keep the temperature of the source drive down throughout the procedure, like a fan blowing straight on it, maybe put a bowl of ice behind the fan for it to pull cool air from and change out the bowl when it melts (and do not let it spill near all of that exposed electronic and electrical gear)

    If the backwards copy gets a lot of the drive copied, we might be able to cheat to get the rest, but that's down the road and complicated.
     
  2. The Great Inert

    The Great Inert New Member

    24
    0
    Jun 20, 2012
    Unitron and HomeUser, this is outstanding information. I can't say thanks enough.

    I've decided to approach this using both strategies. Right now, I'm an hour and a half away from the 24 hour mark; I'll let the program run for the next 12 to see where it goes. If the program is still processing bad sectors, I'll stop it there completely, get a decent night's sleep, and pick it up again using the 512/1 byte, run-in-reverse option.

    Unitron: I actually have had a fan blowing on both drives since this all started, and even after 24 hours they're both surprisingly cool to the touch. I guess the WD Green series (which I've always thought was a cheap way for WD to force consumers to use a slower rpm drive) do have one strong point.
     
  3. unitron

    unitron Active Member

    16,387
    2
    Apr 28, 2006
    semi-coastal NC
    5200, or 5600, or 5800 PRM is fast enough for a TiVo drive, so anything faster just generates unneccessary heat.
     
  4. The Great Inert

    The Great Inert New Member

    24
    0
    Jun 20, 2012
    dd_rescue has just entered hour 30. :eek::eek::eek:
     
  5. unitron

    unitron Active Member

    16,387
    2
    Apr 28, 2006
    semi-coastal NC
    If it's still reading bad sectors,

    CTRL+C

    to interrupt, then

    poweroff

    to shut down so it can cool enough for the freezer.

    If it's actually showing increasing numbers of bytes read and written, let it finish.
     
  6. The Great Inert

    The Great Inert New Member

    24
    0
    Jun 20, 2012
    Unitron, I'm into hour 43 and it's still reading bad sectors. What should I be looking for to tell if bytes are being read and written?
     
  7. HomeUser

    HomeUser Active Member

    2,376
    1
    Jan 12, 2003
    A2 Michigan
  8. unitron

    unitron Active Member

    16,387
    2
    Apr 28, 2006
    semi-coastal NC
    What he said.

    Sorry, I've been catching up on a day or 3 of sleep.
     
  9. The Great Inert

    The Great Inert New Member

    24
    0
    Jun 20, 2012
    I want to thank you both again for all of your help these last few days. I truly appreciate all of your input more than you can imagine.

    An update: at hour 45 and a half, I simply stopped dd_rescue altogether. The only sector information I noticed had changed in nearly two days was the number of bad sectors on the old drive. Upon stopping the program, I ran pdisk -l /dev/sdb to see if anything had transferred, and the program reported that all 15 partitions had been moved. (/dev/sda and /dev/sdb reported the same number of partitions, but I didn't copy the data figures to determine if the partitions on sdb were mathematically identical to those on sda.) Thinking that the new drive had to have been at least partially cloned, I put it in the Tivo, started it up, and sure enough--Welcome, Powering Up--Almost There--GSOD--reboot. Again and again and again.

    At this point, after three days of trying to get the old drive back, I need (for sanity's sake) to stop trying. I'm going to wait until each of the drives, and the computer I was using to do the recovery, cool down...about three or four days...before I try again. The next attempt is the deal-breaker: if I can't get the information off the dying 1TB, and can't get the new 1TB to work in the Tivo, I'm simply going to clone the image of my 500GB drive (my first Tivo upgrade) to the new 1TB and admit temporary defeat. An unbelievable amount of information, including 76 episodes of a television series that will probably NEVER see the light of day on DVD because of Dick Wolf's licensing nonsense, will be lost, but I guess it serves me right for not actually backing up an image of the drive while it was working.

    After three days, I'm entitled to a rant I think you'll both appreciate. One thing that sucks about this situation is that I'm actually adept with hardware and could EASILY (really, really EASILY) replace the PCB on the drive...IF the PCB for my drive didn't cost $150 dollars. I swear to God, hard drive manufacturers have their consumers by the round, dangling objects. I know there's nothing mechanically wrong with the drive besides its age. If i could replace the board it would work again, even if I only ran it for the few hours it would take to clone the drive via WinMFS--but WD "won't replace parts." After all, why replace parts when you can just force people to buy new drives, right? Locking the market in this way lets third-party distributors charge a premium for replacement PCBs. Of course, in three years' time, the PCB for my drive will probably cost about 40 bucks...but then that means waiting three years to clone the drive. Which sucks. Period.

    (Actually, come to think of it... my electric bill is going to suck next month, too.)

    For now, I need to sleep, but I really want to thank you both again for all of the help. I'll be back at the end of the week if you're both still gracious enough to indulge me. As I said, if I could buy you both a drink of your choice I would.
     
  10. unitron

    unitron Active Member

    16,387
    2
    Apr 28, 2006
    semi-coastal NC

    What

    pdisk

    showed you was that the partition map had been copied.

    That doesn't mean that the actual partitions had been.

    Did you upgrade from the 40GB to the 500 and then from the 500 to the 1TB?


    To what season of what show do you refer?
     
  11. The Great Inert

    The Great Inert New Member

    24
    0
    Jun 20, 2012
    Unitron, happy to see you on!

    I indeed went from a 40GB to a 500GB to a 1TB. I still have the 40 and the 500, which both still work.

    About a hundred million years ago (okay, 20 years ago,) Dick Wolf produced a series called "New York Undercover" that was shot almost entirely in the Lower East Side, where I was born and raised. The show itself wasn't the greatest (some of the plots laid it on a bit thick and there was a subtle but nasty streak of reverse-racism to most of the show) but it serves, for me at least, as a reminder of what my neighborhood used to be like before gentrification, NYU, and Mike "Mayor Dips**t" Bloomberg turned it into a 24 hour dorm room. Alas, it's probably lost... unless I have better fortunes this weekend...
     
  12. bshrock

    bshrock New Member

    263
    0
    Jan 6, 2012
    Not that I am suggesting anything... Can you get an advance replacement for the drive? Western Digital used to put a hold on your credit card and send out the replacement drive so you could copy off your data.
     
  13. bshrock

    bshrock New Member

    263
    0
    Jan 6, 2012
    What math? Just start at the end address of the source drive.
     
  14. The Great Inert

    The Great Inert New Member

    24
    0
    Jun 20, 2012
    Welcome to the discussion, bshock. According to the ever-helpful people at WD, the company "doesn't replace parts". Even if I could get them to ship a replacement, the PCB of the replacement drive would have to be a 98-99% match to the PCB of the damaged drive before any data extraction could occur. As the dying drive is now 3 years old, the chance of that happening is rather slim... :(
     
  15. bshrock

    bshrock New Member

    263
    0
    Jan 6, 2012
    Probably even if the PCB was a 100% match the error map would be different.

    You might try breaking the copy in small chunks letting the drive cool in-between.

    copy blocks 00 through 1000 let the drive cool then copy blocks 1000 through 2000 etc.

    Note: 1000 is probably way to small a value just easier for the example adjust it to just below area where the first errors are returned.

    I have used the program DFSEE to binary copy a drive in just that way.
     
  16. unitron

    unitron Active Member

    16,387
    2
    Apr 28, 2006
    semi-coastal NC

    dd_rescue


    can do that, you just have to give it the right parameters.

    For that matter

    dd

    can do it as well, if you set up the options correctly.
     
  17. bshrock

    bshrock New Member

    263
    0
    Jan 6, 2012
    True just not the only method. I have not used dd_rescue enough to be comfortable with the options. DFSee can be run from a command prompt or by using a menu. It was what I used to recover the failed 250 Gig in my old Series2.
     

Share This Page