[arch-releng] LVM Installation Issues

[Dieter Plaetinck]

On Tue, 11 May 2010 11:23:14 +0200
Thomas Bächler <thomas at archlinux.org> wrote:

> Am 11.05.2010 10:49, schrieb Lukas Grässlin:
> > I tested archlinux-2010.04.19-core-x86_64.iso and have some issues
> > with installing with a lvm:
> > 
> > * It would be very very nice if I could say when creating a logical
> > volume: "Take all the free space in the volumegroup" (aka
> > -l100%FREE)
> 
> Good idea!
> 
> I didn't run into this yet because I tend to only assign a fraction of
> the available space during installation (i.e. on a 300GB volume, I use
> 10GB for system, 4GB for swap and 20GB for /home, let the rest
> unused).

same here.
I think this is the most common way to use lvm.

see  http://bugs.archlinux.org/task/13425 for -l feature request.

> > * If I type a value greater than the free space in the vg the
> > installer says it encountered errors. Would be better if it is
> > checked, if the value exceeds the free space.
> 
> Good idea. Also didn't run into this for the same reasons.

patch welcome. although in combination with the -l option, this would
get needlessly complex; know that the lvm setup is only created after
everything is defined. so either you get the feature above or this one,
both at the same time gets messy, but i hope i can be proven wrong.

> > At the current state I wouldn't use the Installers capability to
> > create a LVM because it's a bit glitchy.
> 
> Always worked for me so far.
> 
> > The same is true of the capability of creating dm-crypt volumes.
> > Till now I was not able to set up a working scenario with LVM atop
> > of dm-crypt. The Installers also fails with some errors. (I don't
> > have more information about this yet because it's a bit longer ago.
> > I will retest it)
> 
> I installed a system using AIF from git in February or March I think
> (the current CDs contain a newer AIF version). I had no trouble
> setting this up back then.

Recently some issues have been fixed where additional options specified
by the user caused aif to break. other then that, the code hasn't much
changed reliability-wise. and for me personally it works fine.

Dieter