more than 2TB?

[Steffen Grunewald]

On Sat, Dec 11, 2004 at 12:39:21PM +0100, Holger Levsen wrote:
> don't know much, but maybe I can give at least a little help...

Probably you know more about Debian than me. (Grew up with MCC, SLS,
Slack, then turned to Redhat, when RH introduced unstable PAM, on to
SuSE, later back to Redhat [for initial cluster setup] ...)

> > I'm trying to install Debian using FAI on a system equipped with a Very
> > Large (>2TB) RAID array.
> wow ;) 

:-)))

> > The install kernel doesn't take notice of the device because
> or (re-)build your own fai-kernel... ?! 

Too lazy. Been there, done that, but still it usually breaks more than
it fixes. I'm not complaining...

> > Unfortunately, neither fdisk nor sfdisk has been built with LBD support.
> and fdisk/sfdisk-packages ;) Then install those inside your nfsroot.

See above. I'm maintaining a FAI setup tree for production use and won't
take the risk to break something.

> > Should I consider missing LBD support a bug of Debian packages (and
> > consequently file a bug), or am I missing something? 
> I would suggest to file bugs with priority "wishlist". And I'm sure the bugs 
> against (s)fdisk won't be fixed before sarge is released...

Sure, there are other things which are more important. And Sarge must
come to a freeze some day.

> > utility that can handle large block devices out of the box, and can be
> > driven by a small script (fdisk isn't optimal to do this...)?
> don't know.

Let me fill in: parted seems to do the trick, but it's not suited very
well to run automatically - there is the '-s' option, sure, but there's
no way to automatically answer the "Ignore/Cancel" questions which woudl
come up every now and then.
I will do the partitioning/formatting by hand and leave it alone then,
keep the disks out of the disk_config (in particular since there's no
way to "put partitions aside" completely -> Thomas, ever thought of a
partition type "none"?) and add the fstab lines within a script.

Not nice, but pretty safe, and it's definitely an exception (OK, two
exceptions since the machine's got a twin...)




> 
> 
> regards,
>  Holger
> 

-- 
Steffen Grunewald * * * Merlin cluster admin (http://pandora.aei.mpg.de)
Albert-Einstein-Institut (MPI Gravitationsphysik, http://www.aei.mpg.de)
       Science Park Golm, Am Mühlenberg 1, 14476 Potsdam, Germany
e-mail: steffen.grunewald(*)aei.mpg.de * +49-331-567-{fon:7233,fax:7298}