From: unruh@invalid.ca
On 2012-11-16, Adam wrote:
> unruh wrote:
>> On 2012-11-16, Adam wrote:
>>> If you're going to do it again from the command line, I /strongly/
>>> suggest you do the r/w test as part of mke2fs, before you put important
>>> data on the drive and while it's still easy to return if there are
>>> problems with it.
>>
>> But I thought that most drives these days did a silent bad-block
>> remapping anyway. Thus you would not see anything on such a test, until
>> so many bad blocks had swamped the system that the spare blocks were all
>> used up. Ie, is there really any point at all in his huge waste of time?
>
> Maybe my thinking is a little old-fashioned, but if there are going to
> be a significant number of bad blocks, or the drive is going to fail
> early on ("infant mortality"), I figure it's better that it happen
> before the user's put any important data on it (and is depending on it)
> and while the drive is still relatively easy to return or exchange.
> Also, I don't think it's that much of a "waste of time" because the
> computer can still be used for everything else during this testing, and
> TJ mentioned he was doing this on his second computer anyway.
Well, IF my conjecture that bad block remapping is done in
harware/firmware anyway is correct, then this test will not find any
anyway.
You might want to do a stress test on the drive to do the same thing--
ie, fill the drive with some set of bytes, and then read them out, to
heat it up. Mind you running on usb, that drive can only take about 5W
anyway, which is not going to heat much of anything up.
Yes, it would be good to find those dud drives early. I am just not sure
that this will do it. But perhaps someone who knows drives better than
either of us will chime in.
>
> Adam
--- SoupGate-Win32 v1.05
* Origin: you cannot sedate... all the things you hate (1:229/2)
|
