[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: OT: Re: Interesting new RAID card?
- To: misc_(_at_)_openbsd_(_dot_)_org
- Subject: Re: OT: Re: Interesting new RAID card?
- From: Nick Holland <nick_(_at_)_holland-consulting_(_dot_)_net>
- Date: Wed, 15 Sep 2004 22:51:39 -0400
Greg Thomas wrote:
> Nico Meijer
>> Hi all,
>>> Any of these puppies ought to work with OpenBSD (I have not tried any of
>>> these cards on OpenBSD but have had no problems with a variety of other
>>> systems and platforms).
>> I use systems similar to arcoide's raidcase product line (up to tomorrow
>> only in FreeBSD systems). Works every time; completely OS-agnostic. I
>> assume the same for arcoide's offerings.
> I'm just curious, how do you rebuild the array after a drive failure?
In the first and second generation Arcoide products, you use a boot
floppy to run a utility which remirrors. One nice feature is you don't
have to remirror the whole drive -- I put one in a client's old 486
voice mail computer, he was concerned that the 200M HD that was in there
might fail some day (heh), put in a pair of 40G drives (smallest we
could get), and set up a 1G partition for DOS ("you have five times the
space now, BTW you were using 30% of the 200M drive"), rather than
waiting for it to completely remirror, just told it to be happy after
mirroring the first 5% of the drive (as I bill by the hour, NOT my
The more modern Arco products and the Accusys ACS7500 remirror when you
replace the failed drive. I am much more familiar with the Accusys than
the modern ArcoIDE products, but the Accusys is just plain cool. Pull
the failed drive, insert the new drive, lock it down, remirroring
starts. The Accusys doesn't even include any software for any OS.
Only complaint: the Accusys products have no way to inform you of a
failed drive, though modifying it to route the error LEDs to a couple
inputs on the parallel port shouldn't be hard.
The first generation Arco products did not support UDMA at all. The
second genration Arco products claimed to support it, but has given me
(and a client) nothing but nightmares under OpenBSD and SCO (actually,
SCO was worse than OpenBSD: OpenBSD puked quickly if DMA transfers were
not disabled. SCO let you get it set up and running and would die in
operation *shudder*). Not had any problems with the Accusys products.
Back to the topic of the NetCell/SyncRAID products: I must say, I'm
curious. My guess is it probably would need a PCI identification diff
to work properly with OpenBSD, just to say that the PCI identifier the
card gives should be handled by pciide(4). I do believe Windows is
sloppier about some device recognition (and pickier about others, of
course) than OpenBSD...assuming they don't use a "known" front end (such
as a common (and apparently, OLD, if W2k supports it) Promise chip) and
matching PCI identifiers.
The mod might be so trivial even I could do it. :)
Yes, I find the idea interesting...very interesting...