incremental disk wipe?

[rikona]

Hello Tony,

Saturday, March 7, 2015, 6:59:57 AM, Tony wrote:

> Rikona,

> On 07/03/15 14:25, rikona wrote:
>> Hello Rashkae,
>> 
>> Friday, March 6, 2015, 11:22:38 PM, Rashkae wrote:
>> 
>>> On 15-03-07 01:09 AM, rikona wrote:
>>>> I was checking the timing of securely wiping large disks - looks like
>>>> it takes a looooong time. Is there a way to do this in smaller
>>>> 'chunks', that might complete in an hour or so, eventually filling up
>>>> the entire disk? I'm not hiding plots from NSA, so I don't need
>>>> extreme security - just enough to discourage easy-tech snoops. :-))
>>>>
>>> dd if=/dev/zero of=/dev/sdX bs=4M
>> 
>>> This will be as fast as you can go, but it will still take a modern 
>>> large hard drive several hours to finish.  But a drive that writes 
>>> 200MB/s still only writes 200MB/s... the drive can't multiple chunks 
>>> simultaneously.
>> 
>> Thanks to you and others for the replies. I wasn't thinking of
>> multiple chunks simultaneously - rather sequentially. Let's say I'll
>> be away from the comp for 30 min or so. How can I keep it busy for
>> JUST 30 min - no more, to do *part* of the wipe? Then, later, do
>> another 30 min, eventually wiping the entire disk?

> Why not run the dd command as suggested in this thread, but in the
> background with a nice value of 19. That way the process will run when
> there is little else for the processor to do, so it should not affect
> the day to day running.

Again, thanks to all for the helpful suggestions!

This might fix the 'problem' - when I tried wiping, OO [and other
pgms] would NOT run correctly, or at all, crashing repeatedly. The box
was not useable while wiping. How can I get dd [or something else] to
run 'nice' as you suggest? Can I set it in htop after it starts, or is
it better to start it with the desired nice?

One of the pgm complaints seemed to be related to disk access. Does
nice also work to control disk access, since this is such a
disk-intensive process? Recoll accesses a few large xapian data bases
[~90GB], and I use it very often for searches - I need it to work well
while wiping. I manually update these Recoll DBs for the same reason -
it takes over the box. Would be good to have that be 'nice' too...

> I'm not sure what you are trying to achieve. You say wiping takes a long
> time, but doing it in chunks or as I've suggested above means it will
> take even longer?

> BTW, writing zeroes to the disk using dd will definitely prevent the
> 'easy-tech' snoops from recovering your data.

Thanks,

-- 

 rikona