Re: dd for failing discs?

[Bernhard Voelker]

re-adding the ML.

On 02/18/2014 07:43 AM, Bernhard Voelker wrote:
> On 02/17/2014 01:46 PM, Pádraig Brady wrote:
>> On 02/14/2014 01:55 AM, Bernhard Voelker wrote:
>>> Can short reads not also happen with block devices?
>>
>> Theoretically possible possible I suppose.
>> Though for regular files backed by block devices the task will go
>> into non-interruptible sleep from the read call, and thus not return early.
>> This is kind of a UNIX law which would be very awkward to change.
>>
>> Though POSIX does say:
>>
>> "The issue of which files or file types are interruptible is considered an
>> implementation design issue. This is often affected primarily by hardware
>> and reliability issues."
>>
>> So it is safer to use fullblock, and in your description I'd
>> mention that 'fullblock' is used to cater for short reads
>> (which traditionally never occur on disk based devices),
>> and 'sync' is used to pad out bad reads with NULs.
>>
>> Avoid nondescript phrases like "might be helpful".
>> Simply state in which cases the option is useful.
>> 'noerror' would be required for this use case for example.
> 
> Thanks for clarifying.
> 
> [...], so an amended
> version of the patch will take a few days.


>From e9e089cf4053e7f016e58deb12dbb530bddf18bb Mon Sep 17 00:00:00 2001
From: Bernhard Voelker <address@hidden>
Date: Tue, 25 Feb 2014 18:50:27 +0100
Subject: [PATCH] doc: add dd example for failing disks

* doc/coreutils.texi (dd invocation): Add an example for how to call
dd to save data from a failing disk.  Mention GNU 'ddrescue' as one
of the more specialized tools in such a case.
---
 doc/coreutils.texi | 18 ++++++++++++++++++
 1 file changed, 18 insertions(+)

diff --git a/doc/coreutils.texi b/doc/coreutils.texi
index 95453c7..5e3b402 100644
--- a/doc/coreutils.texi
+++ b/doc/coreutils.texi
@@ -8928,6 +8928,24 @@ tape=/dev/rmt/0
 (dd bs=4k seek=1 count=0 && dd bs=512k) <$tape >$disk
 @end example

+@cindex ddrescue
+@cindex disks, failing
+For failing disks, other tools come with a great variety of extra
+functionality to ease the saving of as much data as possible before the
+disk finally dies, e.g.
+@uref{http://www.gnu.org/software/ddrescue/, GNU @command{ddrescue}}.
+However, in some cases such a tool is not available or the administrator
+feels more comfortable with the handling of @command{dd}.
+As a simple rescue method, call @command{dd} as shown in the following
+example: the options @samp{conv=noerror,sync} are used to continue
+after read errors and to pad out bad reads with NULs, while
+@samp{iflag=fullblock} safely caters for short reads:
+
+@example
+# Rescue data from an (unmounted!) partition of a failing disk.
+dd conv=noerror,sync iflag=fullblock </dev/sda1 > /mnt/rescue.img
+@end example
+
 Sending an @samp{INFO} signal to a running @command{dd}
 process makes it print I/O statistics to standard error
 and then resume copying.  In the example below,
-- 
1.8.4.5