[dm-devel] Re: Follow-up on your AIC-79xx Linux Patch...

[Darrick J. Wong]

Actually, there are two ways that I know of to deal with this.  I'm
assuming that you're using Adaptec HostRAID on that 7901 controller:

1) Deactivate HostRAID.  Set up a small /boot at the beginning of sda,
make a md-based RAID0 out of the rest of the two drives for /, and
(generally speaking) things should just work.

2) md doesn't know how to interpret HostRAID arrays correctly--what you
want is a program called dmraid because it _does_ know how to interpret
HostRAID arrays and set up the device-mapper driver correctly.  You'll
need to get the latest copy (1.0.0rc10) from Heinz Mauelshagen
(http://people.redhat.com/~heinzm/sw/dmraid/), build it, and then modify
your initrd to include /sbin/dmraid and call "dmraid -ay" prior to
mounting the root filesystem.

Neither approach requires the attach_HostRAID patch.  Incidentally, the
a320raid driver is a binary module from Adaptec.  You're not losing much
by using an open-source solution, because a320raid, md and dmraid are
all software RAID drivers.

--D

José Miguel Morales F. wrote:
> Hello Mr. Darrick. My name is José Miguel Morales, System Manager at
> Mining Center, P. Catholic University of Chile.
> 
>    I'm writing to you (in hope this IS your right address) about a
> thread you started on December 1st, 2005, at ussg.iu.edu
> 
>    Is about your "rejected" patch for "aic79xx_osm_pci.c" under kernel
> 2.6.14.
> 
>    I have a couple of Dell Precision 670n, with nice AIC-7901 on-board +
> 2 73 GiB. SCSI-3 10.000 RPM H.D.s, built as RAID-0 for QUITE A GOOD
> speed...
> 
>    Well, of course I had trouble installing Linux on that configuration.
> The driver ("aic79xx_osm_pci.c" of kernel 2.6.15, IDENTICAL to the
> 2.6.14 one) just recognized my two disks as /dev/sda & /dev/sdb, and NO
> RAID support.
> 
>    With your patch (and using "attach_HostRAID=1"), I finally made a
> /dev/md/0 visible, then wrote a SIMPLE /etc/raidtab (included below) and
> started it with "raid0run /dev/md/0".
> 
>    That way "fdisk /dev/md/0" was able to see the partitions, and WITH
> apropriate device names for each partition, but such devices don't exist
> under /dev/md/...
> 
>    Looking at /var/log/messages I found raidtools complaining about
> "errors with the superblock of the array", and I was unable to perform
> further actions without risking to destroy the array. (Which already has
> 2 partitions with a complete WinXP installation.)
> 
>    I noticed that you talk about an "a320raid" drier module, but I can't
> find any "direct" reference to it. Could it be possible to you to
> enlighten me a bit more on this issue, please?
> 
>    Sorry for bugging you, but as I read the whole thread I know I'll get
> no "official" help from Linux people, and I don't want to give up the
> array...
> 
>     Thank you very much in advance...    José Miguel Morales F.
> 
> P.D.: Here is my "simple" /etc/raidtab:
> 
>  - + - + - + -   Cut Here    - + - + - + -   Cut Here    - + - + - + -
> 
> raiddev /dev/md/0
>     raid-level        0
>     nr-raid-disks        2
>     persistent-superblock    0
>     chunk-size        16
> 
>     device            /dev/sda
>     raid-disk        0
>     device                  /dev/sdb
>     raid-disk        1
> 
>  - + - + - + -   Cut Here    - + - + - + -   Cut Here    - + - + - + -
> 
>    ... I really don't know, but the "persistent-superblock" statement
> always gives me problems, no matter how do I use (or don't use) it...   
> Regards !!!
> 

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 191 bytes
Desc: OpenPGP digital signature
URL: <http://listman.redhat.com/archives/dm-devel/attachments/20060515/d6145fe1/attachment.sig>