Use existing software RAID with new Rocketraid 2320

Zotty · 05-26-2007, 04:17 PM

I've been using a RAID5 software raid for a while now without any hickups. As I needed more SATA ports I ordered a Highpoint Rocketraid 2320 (8 SATA ports) which arrived today. Now the problem is that it I can't seem to use the existsing array with the new piece of hardware.

The software raid was created using mdadm and consists out of 4 drives. When hooked up to the mobo SATA ports or a Promise TX4 card the drives get recognized as sdb, sdc, sdd and sde. But attached to the Rocketraid no devices are assigned. The Rocketraid sees the drive as uninitialized and won't do anything except initializing the drive and thus destroying the data on disk already
When hooking up a single non software raid drive it is assigned a device and I can mount and use the disk without problems. Tried this with multiple non raid drives and all are recognized as legacy drives.
So it's just the existing software raid I can't read. Any clues to how I can maintain the existing data and just continue using mdadm? I don't care for hardware raid, it's just the extra 8 SATA ports I'm interrested in. And the existing data offcourse.

Is this a normal behaviour for a highpoint card? Have I missed something? What could be the problem?

The system is running Debian testing with the latest kernel (2.6.21.3). Compiled the latest highpoint driver, combined with build-in hpiopt driver.

Quakeboy02 · 05-26-2007, 04:24 PM

You're probably going to have to send it back and exchange it for a non-raid controller.

Zotty · 05-26-2007, 04:32 PM

Quote:

Originally Posted by Quakeboy02

You're probably going to have to send it back and exchange it for a non-raid controller.

That's what I looked for in the first place. But I've been unable to find a non-raid controller with the same amount of ports (and that's available). Only choice was a cheap highpoint card (which I choose) or a 600 euro real hardware raid card. That would be a major waste of money for just a bunch of extra ports. Aside from that I don't want to unable to access/recover the data when the hardware fails.
Hence the choice for software raid and a fake raid card with 8 ports. But the current situation was something I did not anticipate. I've got a week to get it to work or I'll be sending it back. If this won't work the card will be completely useless for my purposes...

Quakeboy02 · 05-26-2007, 04:42 PM

The problem is that this appears to be a "real" raid controller. If it's any consolation, I've got a Promise FastTrak66 using fakeraid on my test system. I tried adding a Promise S150 TX2Plus to get access to a SATA drive, and it doesn't play well with the FastTrak66.

Zotty · 05-27-2007, 04:50 AM

Running short on ideas I flashed the card with the latest firmware in hopes it would change something. But alas, no difference.

Btw, what I don't get is the sata_mv driver is supposed to support this card:

From http://linux-ata.org/driver-status.html#marvell

Quote:

Marvell 88SXxxxx
Driver name: sata_mv

Summary: Similar to ServerWorks "frodo": per-device queues, full SATA control including hotplug.
The 88SX50xx series supports TCQ, but not NCQ or PM.
The 88SX6xxx series supports TCQ, NCQ, and PM.
The 88SX7xxx series supports TCQ, NCQ, and PM.

HighPoint (HPT)
Driver name: sata_mv or hptiop (sometimes)

Some of the recent HighPoint cards are based on the Marvell 88SX50xx chips. These will be supported by the Marvell libata driver (in progress, see above).

Newer cards pretend they are SCSI, and are supported by the hptiop driver.

The rocketraid card has a Marvell 88SX6081 chip, but no matter what I do the sata_mv module doesn't appear to do anything. I even looked in the source and the chip is in there, including my revision (rev 9).

And is the highpoint driver (rr232x) always needed? What if the sata_mv driver would work, will the rr232x module no longer be needed? At least not for the sata ports to work?