Felix, you are very astute. Maybe I
will contact you offline about raid recovery.
Thanks for answering my ddrescue question. I wasn't sure what
would happen when going from bigger to smaller, but it sounds like
it will fail gracefully when it hits the end of the smaller disk,
leaving my smaller disk in the same state as if I had copied
directly from the original of the same size.
-J
On 10/17/13 3:32 PM, Felix Ehlermann wrote:
Dear Jonathan Joseph,
one thing is not quite clear to me:
Is your RAID5- array "failed" (more than 1 disk is dead) or
"degraded" (1 disk dead)?
If the array is "degraded" you should be fine by just replacing
the faulty disk with a working one (same size or larger should
not matter. The array should be able to rebuild onto the new
drive (if it doesn't => something else is wrong). There is no
need for ddrescue in this situation.
In this case I would do a backup first anyway. Your array should
still be able to start and therefore the filesystem on it should
be accessible just fine. No data lost yet.
If the array has indeed "failed" you're looking at a rather
complicated situation as you basically have an even more ugly
situation than a failed raid-0 would be.
In that case just copying the sectors 1:1 onto a working drive
will NOT resolve your issue.
If the data on the array is important / valuable and you don't
have a backup of it I suggest that you hand this over to someone
with some experience in this field.
This is not meant as an offense, I just know how much we charge
our customers with their "we already tried to fix it before we
asked for help" kind of administrators - sometimes just to find
out that their data could have been salvaged if it had not been
destroyed by improper handling of the situation by their admin
:-(
Regarding your actual question:
If the new 1.5TB-drive is exactly the same size as the failed
drive, you should be fine writing the data from the 2TB disk
onto it with ddrescue or even dd.
It will of course abort at a certain point because the new 1.5TB
disk will run "out of space" - but the remaining data from the
2TB drive does not originate from your failed 1.5TB disk anyway,
so this doesn't matter.
However as written above this is very likely not going to
resolve the issue with your array. Probably the meta information
on the drives are out of sync (if you really have a "failed"
array). This can be fixed, but it is very specific to what kind
of raid you are using.
Kind Regards
Felix
On 17.10.2013 18:57, Jonathan Joseph wrote:
I have a failed disk in a failed raid-5 array that I am trying
to resurrect with ddrescue. The disk is 1.5TB, and I
successfully did a first pass ddrescue of it (using -f -n flags)
to a 2TB drive that I had available (3 errors reported). Since
this is one disk of a raid, there is no proper partition or file
system information on it. You can replace a disk in a raid-5
with a larger one and it will rebuild OK, so I thought the 2TB
disk would be fine, but it seems that that just cloning one of
the disks of the raid array onto a larger disk will probably not
work because in a raid array of this sort, some critical
information must be stored at the last block of the disk
The last block of the Anchor DDF Header
MUST be the last addressable logical block on a
physical disk
Microsoft Word - SNIA Technical Position - DDF v2.0.doc
So I want the ddrescue copy to go onto an identical 1.5TB drive.
I was able to get a matching 1.5TB drive so I would like to do
the ddrescue onto that drive now. However, now that I have
already done the ddrescue once, and presumably have most of the
data on the 2TB drive. I want to access the original failed
disk as little as possible to prevent making it worse - and
hopefully just access the original failed disk to try and
improve the error blocks as much as possible.
My question is can I now ddrescue from the 2TB drive to the
fresh 1.5TB drive and have everything end up in the right place
- including having that "last block" from the original 1.5TB
drive be in the right place on the new 1.5TB drive? Or do I
really need to redo the ddrescue from the original failed 1.5TB
disk to the new matching one?
Thanks.
-J
_______________________________________________
Bug-ddrescue mailing list
address@hidden
https://lists.gnu.org/mailman/listinfo/bug-ddrescue
_______________________________________________
Bug-ddrescue mailing list
address@hidden
https://lists.gnu.org/mailman/listinfo/bug-ddrescue
|