| From: Lennart Sorensen <lsorense@csclub.uwaterloo.ca> | On Thu, Jul 16, 2015 at 11:57:49AM -0500, o1bigtenor wrote: | > | > I have a large system with a fair number of resources which has just come | > home from the hospital (computer) when I had a disk drive fail abruptly. | > | > 1.The tech tells me that the hard drive can be accessed. How do I do a low | > level transfer to another hard drive? | > (This would mean that I would have empty sectors but there is some | > information that I would like to get.) DON'T make any sudden moves. They are likely to be wrong since you are upset and you are doing something new. | ddrescue can be handy for taking an image of a slightly defective | disk to somewhere else, so that you can then experiment on the copy to | recover things. Yes, I do this: 1) raw ddrescue from the whole failing disk to a whole new disk. (Make sure that the new disk is at least as large as the old one!) By copying the whole disk, you get the partition table and boot blocks. But you also get the same UUIDs so it is best to not have the two disks on the same machine after the copy is done (mount etc. will be confused). 2) Then put the failing disk on a shelf 3) Then play with the new disk. I find it best to do this from a live USB stick or CD so I'm not acutally booting off the bad hard drive, the new hard drive, or even just the same controller. Make sure that the live system does not use the sick disk's swap partition. Note: ddrescue is a bit tricky, but it is brilliant. Read the manual twice before using it. Google for examples of usage. I've posted to this list messages about how to use it. You DO want to let it use a logfile! Somewhere in there I do a smartctl -x /dev/sda and capture the output somewhere on another machine. I want to ask the drive what it thinks is going on. BTW, keep in mind that the identification of the problem is provisional. Subsequent observations may point the finger elsewhere. Eg. power supply, disk controller, RAM, sunspots.