Upgrading from RAID 5 to 6 or build native level 6?

Is a RAID 6 array made by upgrading a RAID 5 array to RAID 6 in any
way inferior to a RAID 6 array build from scratch? I read somewhere
that the last drive would be parity only. I'm not sure what to make of
it and i cant find the link again either. Is it going to be a
bottleneck like in RAID 4?

The story is i have a degraded RAID 6 array with 3 out of 5 drives
active because of random read errors on the drives. Danm you, Western
Digital. Anyway, i can get all my data out, so no worries. I need to
ship the two drives for replacement and when i get them back, i'm
going to rearrange the array to put the encryption layer on top of the
RAID layer instead of the other way around as it drops out drives from
the array from time to time. Thats a whole different story.

Now, when i get the drives back, i could use one drive for most of my
data and put the rest on a couple of laptops and spare drives, build a
4 drive RAID 6 array and then add the last drive to the array.
Or i could put my data on two drives a make a 3 drive RAID 5 array,
copy data in to it, add the last two drives and upgrade to RAID 6.
Which is better?

Thanks in advance.

--
Lasse Jensen (fafler at gmail dot com)
--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html

Re: Upgrading from RAID 5 to 6 or build native level 6?

On Wed, Feb 16, 2011 at 08:40, Lasse Jensen wrote:
> Is a RAID 6 array made by upgrading a RAID 5 array to RAID 6 in any
> way inferior to a RAID 6 array build from scratch? I read somewhere
> that the last drive would be parity only. I'm not sure what to make of
> it and i cant find the link again either. Is it going to be a
> bottleneck like in RAID 4?
>

That information is incomplete or out-of-date. During a RAID 5 to
RAID 6 reshape, there may be an intermediate step where all of the Q
parity blocks are stored on one disk, but as the reshape continues, it
will move all the Q parity blocks to a normal RAID 6 layout. IIRC, in
the very early stages of RAID 5 to RAID 6 reshape development, it
would leave it in the abnormal layout, but that has not been the case
for quite some time. And I believe in some cases a RAID 5 to RAID 6
reshape will not even use the intermediate step w/ the abnormal
layout. After the reshape fully completes, it is exactly the same as
a RAID 6 array built from scratch.

> The story is i have a degraded RAID 6 array with 3 out of 5 drives
> active because of random read errors on the drives. Danm you, Western
> Digital. Anyway, i can get all my data out, so no worries. I need to
> ship the two drives for replacement and when i get them back, i'm
> going to rearrange the array to put the encryption layer on top of the
> RAID layer instead of the other way around as it drops out drives from
> the array from time to time. Thats a whole different story.
>
> Now, when i get the drives back, i could use one drive for most of my
> data and put the rest on a couple of laptops and spare drives, build a
> 4 drive RAID 6 array and then add the last drive to the array.
> Or i could put my data on two drives a make a 3 drive RAID 5 array,
> copy data in to it, add the last two drives and upgrade to RAID 6.
> Which is better?
>

If I understand you correctly, I think the RAID 6 route would be
slightly better, because it would have enough parity info to recover
from a single drive failure, while the RAID 5 route you describe would
lose the array completely if a drive failed before you added the last
two drives.


Good luck!
--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html

Re: Upgrading from RAID 5 to 6 or build native level 6?

--Sig_/R+28nxhfxQ2XQ0H2HE+mDPF
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: quoted-printable

On Fri, 18 Feb 2011 19:02:07 -0700
Beolach wrote:

> That information is incomplete or out-of-date. During a RAID 5 to
> RAID 6 reshape, there may be an intermediate step where all of the Q
> parity blocks are stored on one disk, but as the reshape continues, it
> will move all the Q parity blocks to a normal RAID 6 layout. IIRC, in
> the very early stages of RAID 5 to RAID 6 reshape development, it
> would leave it in the abnormal layout, but that has not been the case
> for quite some time. And I believe in some cases a RAID 5 to RAID 6
> reshape will not even use the intermediate step w/ the abnormal
> layout. After the reshape fully completes, it is exactly the same as
> a RAID 6 array built from scratch.

See man mdadm --layout=3Dpreserve vs --layout=3Dnormalise.
I am not sure which one of these is the default, but it is definitely still
possible to convert RAID5 to RAID6 (while adding one disk) without restripi=
ng,
and to me this looks like a very useful feature, not something outdated.

--=20
With respect,
Roman

--Sig_/R+28nxhfxQ2XQ0H2HE+mDPF
Content-Type: application/pgp-signature; name=signature.asc
Content-Disposition: attachment; filename=signature.asc

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (GNU/Linux)

iEYEARECAAYFAk1ft3sACgkQTLKSvz+PZwgyTACfYuPZMGraOimTSoUvw3ZI AKVk
wN4Anj9+l7SbOzyqQs6Ji7KpYoha7G34
=WYlt
-----END PGP SIGNATURE-----

--Sig_/R+28nxhfxQ2XQ0H2HE+mDPF--
--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html