Hi,
On Sat, 17 Jun 2006, John Summerfied wrote:
Several of us have poked around but not been able to put together a method
by which we can do this within anaconda purely. I'm wondering if anyone
could suggest a means by which we could do this.
Years ago, I had some success partitioning in %pre by piping commands into
fdisk. I expect it wouldn't be too difficult to do so now.
It might be a little trickier if you want to use /dev/hda1 for some,
/dev/hda2 for others, but I _think_ you can generate some ks fragment(s) and
then %include them; if so then you can easily generate the required partition
layout in shell script and then include it further down in the ks file.
we've had some success in the past implementing different hacks... I was
hoping someone knew of something a bit more "proper".
In the past, in the source for init we've added a call to a shell script
in which we would diff to the known output of sfdisk; this was done in the
init in the initrd.img. As it's a bit of a pain to edit and repack that,
I was looking for something a little more straight forward.
As we're moving to larger volumes on the systems, we're using parted. I
was taking the opportunity to re-invent this process for us. We've poked
around in that image, but I/we couldn't find/think-of a way to diff the
output of parted with the "template" of what a partitioned drive might
look like, as diff's not included in the initrd.img...
C'est la vie...
Thanks!
Paul
