__/ [ yttrx ] on Thursday 10 August 2006 04:19 \__
> The Ghost In The Machine <ewill@xxxxxxxxxxxxxxxxxxxxxxx> wrote:
>> In comp.os.linux.advocacy, Roy Schestowitz
>> <newsgroups@xxxxxxxxxxxxxxx>
>> wrote
>> on Wed, 09 Aug 2006 23:53:50 +0100
>> <12315965.TpbfCCrinW@xxxxxxxxxxxxxxx>:
>>> __/ [ Chuck Lysaght ] on Wednesday 09 August 2006 17:56 \__
>>>
>>>> EXT3 filesystem, any tools to recover deleted data?
>>>
>>> I don't know any tools, but a way to prevent accidental data loss is
>>> duplication (rsynch, cp -r, or scp -r can be scripted to run
>>> periodically, even using a GUI front-end) and use of the wastebin. I know
>>> people who even alias their rm command to send files to a wastebin
>>> temporarily, until it is removed. And good practice is to always ls
>>> <wildcard> before em <wildcard>.
>>>
>>> If there is no route to easy recovery, consider the filesystem to be
>>> safe. There is a reason why deleted data should be deleted. The
>>> versioning and recovery feature in Vista (which Apple announced ahead of
>>> Leopard yesterday) raised many concerns.
>>
>> A rather clever method was once used by Auspex
>> file servers. AIUI, every hour, the directory (say,
>> /home/ewill3) was walked, and every file found during the
>> walk would be hard-linked in another, generally hidden
>> area. Presumably, the directory structure was mirrored as
>> well (the directories themselves were *not* hard-linked).
>> There is some space wastage, mostly because files that
>> are deleted in the working directory can't get the space
>> reclaimed until all of the hard links are removed from the
>> structure, or one waited until the temporary directories
>> were rotated out -- which was probably configurable and
>> may have depended on the user's needs.
>>
>> I don't know what modifications Auspex used on their file
>> systems, if any.
Ironically (and sadly too), I was virtually forced to send the following
message half an hour ago:
===
Hi,
I am afraid that I may have lost a few days' worth of work. I wrote to a
file (LyX) when disk quota exceeded, so its size was 0 bytes. *GASP*
Do you happen to have nightly backups of user accounts? My account name is:
schestr0
The files which were reduced to 0 bytes were under
schestr0/Thesis
Many thanks,
Roy
===
For the time being, I remain optimistic. But we shall see if it's time for
shock and panic. *hands jitter* I need to get this off my mind.
> Or, if youre already spending time and resources on things
> like that, just spend money instead and buy veritas volume
> manager for linux, and freakin run snapshots on your filesystem.
I don't run nightly snapshots of the filesystem above because I assume that
the CS Department already does that. I keep a daily 'stack' of my main
hard-drive though... enough to ensure nothing gets 'contaminated' and that
all is irreversible after less than a month or two. I still can't get the
problem off my mind. But I'll try...
Best wishes,
Roy
--
Roy S. Schestowitz | < http://en.opensuse.org >
http://Schestowitz.com | Open Prospects ¦ PGP-Key: 0x74572E8E
Tasks: 119 total, 1 running, 116 sleeping, 0 stopped, 2 zombie
http://iuron.com - knowledge engine, not a search engine
|
|
