bug-ddrescue
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Bug-ddrescue] bug on non-hardbs rounded start position or size!


From: Scott D
Subject: Re: [Bug-ddrescue] bug on non-hardbs rounded start position or size!
Date: Mon, 25 May 2015 16:23:35 -0400
User-agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:31.0) Gecko/20100101 Thunderbird/31.7.0

On 5/25/2015 3:11 PM, Antonio Diaz Diaz wrote:
Also, my HDD being rescued contains areas with good
read rate (about 4Mb/s) sized about 600M, and between them - areas of
slow readable sectors (about 4 seconds per sector). I need some
options to immediately skip such slow areas to recover as much as
possible with minimal time spent. For example, such strategy as to
read forwards until slowing to min_read_rate or detecting an error,
then read backwards, until slowing or bad, too. Then divide the
maximum sized untried block by half and repeat.

Such "divide and conquer" strategies have already been tried in past versions of ddrescue with bad results. They produce a lot of head movement and no not reduce the total rescue time. Sorry.
This sounds like a bad head. If the slow sectors are unreadable (errors) then you can adjust the skip size. With ddrescue 1.19 you could add the option --skip-size=25Mi,50Mi (if I have the format correct). That will skip 25MB on the first error, with no more than 50MB skipped on each consecutive error. This will skip out of the bad head faster than normal and allow more reading of the good data first. If the disk has 2 platters then it has 4 heads, and 600MB of good read would be 3 heads of 200MB chunks, and a bad head of 200MB. This is an estimate without knowing what drive you actually have or the number of platters. So in theory the added option would skip out of the bad chunk after no more than 8 errors. On the reverse pass any trailing data that was skipped is picked up. I have bench tested this method and it has good results. Head movement is linear, and the data from the good heads is recovered much faster. Once it gets past the copying phases and starts the trimming phase and scraping phase expect it to be much slower as it is dealing with the bad head, and other than being patient and waiting for the recovery to finish (or the drive to die), the only other option is take it to a professional.

Scott



reply via email to

[Prev in Thread] Current Thread [Next in Thread]