SATA Raid 1 Data Corruption - A7N8X / RocketRaid 1520

  • Thread starter Thread starter ice
  • Start date Start date
I

ice

Hi There

I have 2 x 250 SATA 150 Seagate drives. Attached them to the MOBO SATA
connections and the Promise BIOS never saw them. Spend 4 weeks with
tech support and eventually gave up and bought a Highpoint Rocketraid
1520. Connected everything up and the drives were found.

Initialised & formatted the HDD with Windows XP SP disktools, set up a
RAID 1 array and then copied over my data. Only to find it was mostly
corrupted.

Tried reinstalling and reformatting, but had the same corruption
problems. So uninstalled, and started again, but this time set up a
RAID 0 array, and all data copied over was fine. So uninstalled and
retried with RAID 1 only to find the same corruption problem.

Can any one provide a solution to this data corruption problem? Tech
support so far has been somewhat poor :(

TIA
Ice70
 
a week ago i had same confusion as you. and i gave up, too.
but yesterday i've saw some tech. through tom's and now i think
raid must at least with 3 hdd then raid work !!

good luck &
merry christmas
 
a week ago i had same confusion as you. and i gave up, too.
but yesterday i've saw some tech. through tom's and now i think
raid must at least with 3 hdd then raid work !!

good luck &
merry christmas
No.
There are a lot of different forms of 'RAID'. RAID3, and RAID5, require
three drives or more (which is what the 'reference' you are seeing is
about). RAID0, should never have been called RAID at all (it does not give
any 'redundancy', and given that the 'R' in 'RAID', stands for
'redundant', it is naughty, that it is included as a form of 'RAID' -
however this has happened, since used in combination with RAID1, as
'RAID1+0', or 'RAID10', it is one of the simplest/fastest forms of RAID).
However 'RAID1', gives redundancy, and works perfectly with many
controllers. It is more commonly called 'mirroring', and is 'borderline'
on whether it should be included with the normal RAID forms (which
normally have parity calculations involved), but it most certainly
_should_ work. The 'higher' RAID forms, all need three or more drives, but
RAID1, is perfectly legitimate.
So, RAID1, is perfectly possible, and can work. The downside of it is that
you lose 50% of your storage capacity. Now the original poster has the
very strange situation, that the form that should provide redundancy
(RAID1), is giving data corruption, while the form that doesn't (RAID0),
doesn't. I'd have to say that this sounds like a controller problem, or
some borderline I/O problem with the drives, which is showing up in the
RAID1 configuration. Now there have been a lot of data I/O problems with
SATA150, and some manufacturers ship their SATA150 drives, set to wake up
using SATA133 for this reason (Hitachi do this, and offer a software tool
to switch the interface speed up, if you are confident that your hardware
really will work - given that the only speed gain from 133 to 150, is
basically immeasurable, except by benchmark programs, since the basic
speeds of the drives themselves is still below 100MB/sec, this is a much
safer way to go...).
I'd suggest seeing if Seagate have a software tool to turn the interface
speed down to SATA133, and seeing if both problems disappear (the
inability for the motherboard controller to recognise the drives, could
well come from the same source)...

Best Wishes
 
Hi Roger

thank you for the reply.

Contacted Seagate who replied:

"The controller is what determines the speed (SATA 133) of the drive.
Try exchanging your cables also."

These are the second set of cables I have tried. I would guess that if
they work for RAID 0 they should really work for RAID 1?


I have noticed that the rocketRAID card and my nvidia 6800 GT share
the same IRQ (16) I don't really have any idea what that means, or if
its meaningful. Or even what to do if it is...

thank you

Ice70
 
ice said:
Hi Roger

thank you for the reply.

Contacted Seagate who replied:

"The controller is what determines the speed (SATA 133) of the drive.
Try exchanging your cables also."
I thought you said the drives were signalling at SATA150 rate?. The
utility _is_ available from Seagate (but you have to get them to
understand that you want to turn the rate _down_ from the maximum
supported by the controller and drive). It changes one of the mode page
settings in the drive, so the controller cannot use the faster rates.
These are the second set of cables I have tried. I would guess that if
they work for RAID 0 they should really work for RAID 1?
The problem is that if this is a really 'borderline' error, anything is
possible,and the slight timing differences between the two modes might
make a difference, but generally if the cable is reliable in one mode it
ought to be OK in the other.

I have noticed that the rocketRAID card and my nvidia 6800 GT share
the same IRQ (16) I don't really have any idea what that means, or if
its meaningful. Or even what to do if it is...

thank you

Ice70
It means I'd suggest moving the card to a different slot. While devices
should be able to share IRQ's, the AGP display cards, are commonly one of
the devices that has the most problems with this (remember at times there
will be a lot of interrupts from an AGP card). If it is showing '16', then
it suggests that one of the advanced interrupt controllers is present that
supports interrupts above 15, and should avoid this happening. Normally
the AGP shares with the top PCI slot, so a different sot may help.

Best Wishes
 
ice;

A couple of years ago I had the same problem -- data corruption in Raid1
with a RocketRaid card. In my case only files > 1 MB were corrupted. There
were no error messages. I no longer use Highpoint cards.

I've become a fan of 3ware RAID cards. They are easy to install, have good
documentation, are easy to monitor, give high throughput (the card reduces
the CPU load), and are Linux-friendly. 3ware RAID products are a bit more
expensive than their competition, and are worth it. You might look at
3ware's 8006-2LP, that will let you mirror your two SATA drives. It is about
$120.

http://www.provantage.com/buy-73WAR...a-raid-pci-controller-cables-kit-shopping.htm

Good luck.

TomC
 
Hi TomC

I think I will try and return the card... especially when you get this
kind of response form the 'tech support'

We can recommend and RMA replacement.
This is not a known controller issue.
We rarely hear about data corruption issues that are directly related
to the card (it is usually found to be related to a hard disk).
[/end quote]

Cheers
Ice
 
I think I will try and return the card... especially when you get this
kind of response form the 'tech support'

I was so disgusted with having wasted over a dozen hours trying to get that
little POS to work that I didn't even return it, I trashed it. The data
corruption, which was reproducible for large files, was NOT related to a
hard disk. It was caused by their card or their driver, since everything
else remained in place for the 3ware RAID card, which has worked perfectly
since.

TomC

ice said:
Hi TomC

I think I will try and return the card... especially when you get this
kind of response form the 'tech support'

We can recommend and RMA replacement.
This is not a known controller issue.
We rarely hear about data corruption issues that are directly related
to the card (it is usually found to be related to a hard disk).
[/end quote]

Cheers
Ice


ice;

A couple of years ago I had the same problem -- data corruption in Raid1
with a RocketRaid card. In my case only files > 1 MB were corrupted. There
were no error messages. I no longer use Highpoint cards.

I've become a fan of 3ware RAID cards. They are easy to install, have good
documentation, are easy to monitor, give high throughput (the card reduces
the CPU load), and are Linux-friendly. 3ware RAID products are a bit more
expensive than their competition, and are worth it. You might look at
3ware's 8006-2LP, that will let you mirror your two SATA drives. It is
about
$120.

http://www.provantage.com/buy-73WAR...a-raid-pci-controller-cables-kit-shopping.htm

Good luck.

TomC
 
Back
Top