Re: partman-auto bug for multiple disks
On Wed, Oct 04, 2006 at 02:01:44PM +0200, David Härdeman wrote:
> On Wed, October 4, 2006 13:33, Simon Huggins said:
> > partman-auto has been changed now to deal with multiple disks for other
> > methods but it's broken:
> >
> > case "$method" in
> > regular)
> > for disk in $disks; do
> > id=$(dev_to_partman "$disk") || true
> > if [ -n "$id" ]; then
> > autopartition "$id"
> > exit 0
> > fi
> > done
> > exit 1
> > ;;
> >
> > The exit 0 should be outside the for loop I think. In fact I would
> > propose the following patch but I'd like David/Frans to comment on it
> > before I just apply it.
> I just added support for *specifying* multiple disks. All partman-auto
> methods still need to be changed to actually take advantage of
> multiple disks.
> So I don't think (from a very quick reading) that the code is currently
> incorrect (or I'm daft and you need to provide more detail).
Since partman-auto version 55, multiple disks have been supported so
that partman-auto-raid would work.
> Currently, the code will go trough the specified disk(s), use the
> first one which is found and considered valid (by dev_to_partman) and
> partition it. If you did the partitioning on multiple disks right now,
> you'd get several independently partitioned disks (each with /boot
> etc).
Or you'd get multiple raid partitions created on multiple disks which
something like partman-auto-raid could turn into real raid devices.
> If no disk is found, an error will be generated (hence the exit 1 outside
> the loop).
Right ok.
I would still like this to go back to the way it was in version 55
(revision 39879 e.g.).
Otherwise I guess I can add a method = "raid" that does this.
--
_ huggie@earth.li -+*+- fou, con et anglais _
(_) "Ahh.. Very passable, this, very passable." - Monty Python (_)
(_) (_)
\___ ___/
Reply to: