Re: Brocken Raid & LUKS

[Phil Turmel <philip@turmel.org>]

On 02/21/2013 05:20 PM, Chris Murphy wrote:
> 
> On Feb 21, 2013, at 9:32 AM, Stone <stone@heisl.org> wrote:
>>>> 
>> This is my ouput from the badblocks
>> 1073006628
>> 1073006629
>> 1073006630
>> 1073006631
>> 1073006632
>> 1073006633
>> 1073006634
>> 1073006635
>> 1073006636
>> 1073006637
>> 1073006638
>> 1073006639

> It's consistently reporting 12. This can't be LBA values if it's an 
> AF disk, or you'd get multiples of 8 (8*512=4096). I actually don't 
> recall off hand how to convert from ext block numbers to LBA. But dd
>  wants LBA.

These are default 1k block addresses returned by badblocks.  dd does not
want LBA.  It wants block addresses, with a default block size of 512.
If you specify a different block size with bs=, you must use that scale
for seek= or skip= or count=.

> I haven't read this whole thread, is there a backup? I did see more 
> than one disk with non-zero current pending sector values. So in my 
> opinion, I'd ATA secure erase all of these drives and start from 
> scratch if you have a backup. Actually, I'd ATA Secure Erase them, 
> and then do an extended SMART test to confirm. Or if they're under 
> warranty, RMA them. You shouldn't have so much bad sectors on a 
> disk.

No backup.

> If you keep them, you need to keep an eye on them with an extended 
> smart test every week or two. It sounds like there may be loose 
> material bouncing around in the disks causing these bad sectors, and 
> if that's true, more will go bad. And if more do show up in an 
> extended smart test, and the drives are under warranty, I'd bail out 
> on them. Get them replaced.

Read the whole thread.  A followup smartctl report will be useful, but
Stone's hands are full at the moment.

Phil