Re: Promise Supertrak 8350
It is not, khym, real fakeraid. But it is definitely PainInTheAss.
Actually, i did had some progress (several months ago), after about 2
day part-time poking around with SX and this EX card, but the guys
preferred to purchase real one rather than rely on something not-so-well
supported under Linux.
The info I've had from the hardware guys was that this EX is almost
'real', not pure fakeraid, and it is worth using it, although it is
worse than the real ones.
If I dig back i can, eventually, give more info, BUT i would suggest you
to throw this one to some windows machine requiring RAID, it has good
win drivers, and get a real one.
Not because this is or is not 'real' raid, but you'll have hard time
with Linux. Maybe in the newest 2.6 there was something new (too late
throw it away to some windows machine, and get real card, for example, a
3ware or something.
Craig Sanders wrote:
On Fri, May 11, 2007 at 10:18:09AM -0500, firstname.lastname@example.org wrote:
Has anyone had any luck (good or bad) with the Promise Supertrak 8350
specifically related to an Etch install? I can't seem to get the
installer to recognize the disks as being in an array.
The SATA site at linuxmafia seems to think that this
card should/might be supported generally for Linux:
that page says the Promise 8350 is a real hardware raid card, and that it
might use the Marvell driver.
"(link) Promise SuperTrak EX8300 8-port SATA-II PCI-X card and
SuperTrak EX8350 8-port SATA-II PCI Express card \u2014 real
hardware RAID(?). Uses Intel IOP331 I/O processor (XScale family)
and Marvell 88SX6081 SATA chipset. Press releases says there's a
"full open source Linux driver" \u2014 which might be the Marvell
driver(?). EX8500 card was demoed on 2005-08-23."
oddly, the info on the Marvell driver says that it is fakeraid - i.e. not
hardware raid at all, just software raid with minimal support in the card.
you're better off using linux software raid than fakeraid, even if it does
happen to be directly supported with a linux driver.
"(link) Marvell Technology Group Ltd. 88SX7xxx, 88SX6xxx ("Hercules
II") and 88SX5xxx/88SX48 ("Hercules I") chip series chips \u2014
fakeraid. The 88SX50xx series supports TCQ, but not NCQ or port
multipliers. The 88SX60x1 series supports TCQ, NCQ, and port
multipliers. libata driver "sata_mv" became available 2005-06 (beta
quality as of 2007-02).
For Marvell chipsets 88SX5040, 88SX5041, 88SX5080, and 88SX5081
"Hercules I" (all SATA-I); and 88SX6041 4-port & 88SX6081 8-port
"Hercules II" (both SATA-II), there is a "mvSata" driver from Marvell
International Ltd. Like Garzik's libata "sata_mv" driver, Marvell's
"mvSata" driver uses the kernel's SCSI layers.
As of May 2005's v. 3.4.1, that driver changed from proprietary
licensing to GPLv2; probably the best revised source code is now
Carlos Vidal's file tree, which also has precompiled kernels for
Fedora Core and a HOWTO that should generalise well to other
distributions. A patch may be necessary to run this driver properly
on newer 2.6 kernels.
Supermicro's ftp site offers proprietary drivers for Marvell 4-port
and 8-part SATA chips, written by Adaptec."
if i had to bet, i'd bet that the info about the marvell driver is correct
while the info about the card is wrong. i.e. it's fakeraid.
I've poked around looking for updated docs on how to build a custom
boot/install CD for etch but everything I've seen appears pretty much
tailored to Sarge and even Woody.
Any pointers to help get this card working would be appreciated.
1. throw away the card and get a real hardware raid card. adaptec 2820
or 3ware or something. that page has a pretty good explanation of the
difference and a list of real hardware raid cards supported by linux.
2. use linux software raid. the debian install disks can handle this.