Re: [sheflug] (no subject)

[Peter Humphrey <prh [at] xxxxxxxxxxxxx>]

On Tuesday 29 August 2006 15:21, david.willington [at] ntlworld.com wrote:

> I've been left in charge of a server with a RAID configuration

Is it in service now, or can you play with it some more?

> ... notwithstanding previous advice
> you've given me about using software RAID the machine was bought with
> hardware RAID, or so I was told. Having poked around various websites,
> the machine itself, and a Linux Format article, I'm fairly sure it's a
> fake raid setup.

So'm I. If it doesn't have a plug-in card to drive the disks, the 
functionality is in firmware in the BIOS, and AFAIK all those BIOSes were 
designed to operate with Windows. I've been advised not to trust my data to 
fake RAID, though I don't understand what risk that would imply, so I've a 
software RAID setup on this box now. That involved disabling RAID in the 
BIOS, creating partitions on the disks and assembling them into md devices. 
I had a lot of fun with it. Er ... yes, I think that's the right word.

> RAID was configured in the BIOS, but various 
> distributions insisted on identifying 2 hard drives. Only FC5 identified
> 1 hard drive when I installed it, which is what's on the the server now.

I think SuSE uses software RAID, i.e. mdN devices, but I can't speak for the 
others. I'm running Gentoo, which allows a choice.

> The output of dmraid -r is
>
> /dev/sda: nvidia, "nvidia_hacfjacf", mirror, ok, 586114702 sectors, data@
> 0 /dev/sdb: nvidia, "nvidia_hacfjacf", mirror, ok, 586114702 sectors,
> data@ 0
>
> and the output of mount is

Mine is:

-- 
$ mount
/dev/md/2 on / type reiserfs (rw,noatime,notail,user_xattr)
proc on /proc type proc (rw)
sysfs on /sys type sysfs (rw,nosuid,nodev,noexec)
udev on /dev type tmpfs (rw,nosuid)
devpts on /dev/pts type devpts (rw,nosuid,noexec)
/dev/md/3 on /home type reiserfs (rw,noatime,user_xattr)
/dev/md/4 on /home/prh/common type reiserfs (rw,noatime,user_xattr)
/dev/md/5 on /usr/local type reiserfs (rw,noatime,user_xattr)
/dev/md/6 on /tmp type reiserfs (rw,noatime,user_xattr)
/dev/md/7 on /srv type reiserfs (rw,noatime,user_xattr)
shm on /dev/shm type tmpfs (rw,nosuid,nodev)
usbfs on /proc/bus/usb type usbfs (rw,noexec,nosuid,devmode=0664,devgid=85)
-- 

> I'm not sure what's going on. Did FC5 identify and overrule the BIOS
> software, and is it now handling the RAID configuration? Is the BIOS
> software still in charge and am I likely to upset it by playing around
> with the RAID configuration using Linux tools?

It's hard to see it from here, but I'd say you'd be better off doing as I 
did and disabling BIOS RAID. Of course you'd then have to install again 
from scratch, whence my first question.

> How do I identify when one  of the hard drives fails? I've got something
> in mind like running dmraid -r as a cron job every minute, and looking
> for output which doesn't match the above lines, but as I'm not sure what
> software is doing what I'm not confident that this'll do the job. Is there
> any way of breaking the RAID setup to test my warning system?

You're outside my knowledge here (sorry), but it does seem a bit pointless 
running RAID on a single disk.

-- 
Rgds
Peter Humphrey

___________________________________________________________________

Sheffield Linux User's Group -
http://www.sheflug.co.uk/mailfaq.html

  GNU the choice of a complete generation.