OK. I will give this a look later today. Got some people over the pond
waiting for a report from me first...
On 03/09/2015 09:53 AM, Paul Whalen wrote:
----- Original Message -----
> The thread from Tomasz is interesting. I was jsut in the mode of
> thinking that armv7 builds would be copy an image to SD and go.
>
> I mean how much time to you save on an PXE build to SD versus xzcat to
> SD? I might think that for the basic build, xzcat on a host is faster.
Speeds can vary from board to board. Customization is the largest benefit.
> Then you run a script that does what kickstart does. The only
> consideration I have so far with this is sizing the partitions on the SD
> card. I would like a larger swap to prepare for when I can hibernate to
> swap and getting the / partition to fill the card.
>
> But then I started working with the partitions on the HDD. It is a
> little harder to move a HDD between systems than a mSD card, so minimum
> boot from an SD card that then did the kickstart to the HDD (or SSD)
> makes more sense.
>
> I can pull the F22-arm and repo to my local server; I do that for a
> number of distros every night (rsync changes) and do testing of a local
> install that way. Of course my repo is on a different subnet than my
> install testing; I would have to put up the PXE boot function, etc.
>
> So, lending a hand on a more Anaconda-style install to HD from a minimal
> SD card is something that I can help with, point me to what I can do to
> help.
Installation testing is greatly appreciated. You will need a tftp boot
server setup, rsyncing the entire repo isn't required but will speed
up the installations.
The tests can be found here[1]. Installation on the development boards
does require a couple of manual steps at the end of an interaction
installation or a section in %post to deal with the 'fdtdir' and re-write
uboot if you formatted the mSD/SD during installation.
Paul
[1] -
https://fedoraproject.org/wiki/Test_Results:Current_Summary