Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 14 Sep 2004 11:38:52 +0200
From:      =?ISO-8859-1?Q?S=F8ren_Schmidt?= <sos@DeepCore.dk>
To:        Eirik Oeverby <ltning@anduin.net>
Cc:        current@freebsd.org
Subject:   Re: atacontrol and creating raid-1 arrays
Message-ID:  <4146BC2C.80604@DeepCore.dk>
In-Reply-To: <4146BB3E.2070804@anduin.net>
References:  <8AA611E5-05B1-11D9-831C-000D9335BCEC@anduin.net> <414695D0.1020202@DeepCore.dk> <4146BB3E.2070804@anduin.net>

next in thread | previous in thread | raw e-mail | index | archive | help
Eirik Oeverby wrote:
>> Since the RAID metadata has to be put on disk it wont work if you have=
=20
>> that area used for real data. Depending on what controller you use the=
=20
>> metadata can be stored in different places. Other than that it works=20
>> if you are *sure* the disks are identical (dd with a decnet blocksize =

>> is *much* faster tha cp).
>=20
> Yea, I used dd (blocksize 256kbyte proved optimal; ~40mbyte/sec) for my=
=20
> final operation; was testing with cp first to make sure. This is a=20
> Silicon Image SATA RAID controller which isn't supported by the ATA=20
> driver in RAID mode, so it's purely software. Which, according to=20
> another reply I got, uses the last 255 sectors on the disk. These were =

> already free, so I just tested - seems to work fine. Fsck has no=20
> complaints. Other things I should do to verify?

The Silicon Image SATA "RAID" controller is supported by ATA, and it has =

*NO* RAID capabilities whatsoever, its all done in software, I just dont =

support the particular metadata format that your BIOS uses.
Now that said metadata can take anywhere from 1 to 17 sectors either in=20
the front (old HPT) or the end (Promise/LSI/native) of the disk.
Anyhow if ataraid picks it up etc you should be fine :)

-S=F8ren



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4146BC2C.80604>