On Sat, Sep 15, 2001 at 11:29:41AM -0700, Matthew Jacob wrote:
>
> Don- too bad- I'm over in NAS in building N-258 and just took down the h/w
> where we could test this (Alpha 4100 with 8 Qlogic cards && 8 150GB MegaDrive
> RAID arrays)- We're excessing this equipment because the mass storage group's
> work was terminated due to budgetary goop.
>
> Typically there's no problem in *recognizing* raw SCSI disk, or RAID units
> that look like a SCSI disks- Mike or others will have to speak about how well
> specifical supported RAID cards for FreeBSD will handle something that large.
I'm not shure about recent SCSI definitions, but as long as a physical
sector is sized at 512 Bytes and SCSI uses 32Bit unsigned numbers a
(virtual) SCSI disk is limited to 2T.
Maybe recent SCSI specs already defined a new set of read/write...
> A more problematic issue is whether FFS can reliably handle > 1TB. We did some
> tests with NetBSD at 1TB a couple of years ago- you have to use larger block
> and frag sizes (we were using 32k/8k)- but I've never myelf gotten FreeBSD up
> above 500GB. I was going to do so when the rather flakey h/w I was on above
> was set to be excessed. If I'd known you needed the work done- I could have
> kept it going a couple more weeks and we could worked on sorting it out.
FFS in it's design can go up to 2G * fragsize - that means 2T with default
fragsize of 1k.
The Problem is that FreeBSD internaly uses 32bit signed Blocknumbers for
512Bytes/Block independend of it's physical size.
This limits to 1T.
I successfully build 600G partitions with vinum and I've heared of users
running partitions near the 1T limit.
--
B.Walter COSMO-Project http://www.cosmo-project.de
[EMAIL PROTECTED] Usergroup [EMAIL PROTECTED]
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-hackers" in the body of the message
