On 6/12/20 10:20 PM, Michael Hall wrote:
Just wondering what the current best platform is for building Fedora
cloud-base images (specifically, for AWS). We're looking at potentially doing this as
part of a community college IT course, then contributing the documented process back to
the Fedora Documentation project.
I've installed ImageFactory with libvirt/qemu on a couple of Fedora machines now but
cannot get an image to build. No matter what I do I get one of two error messages.
One error seems to read the value out of the <version></version> tages in the
TCL template file, and return that as the error. So if I make the version "32",
it returns "Error: '32'"
The second error mentions needing to run "add_image()' first, but no
documentation I can find addresses this issue so not sure where to go from here.
Is ImageFactory the current recommended tool, a lot of its documentation seems to
potentially be out of date?
In general if you are building images for Fedora (i.e. you want the changes you're
doing to make it back into
Fedora at some point) it's going to have to go through image factory at some point.
Is HashiCorp's Packer a useful alternative?
I've never used it but I hear a lot of people do.
Is it absolutely necessary to get the whole Koji toolchain involved?
No, you can do it with just imagefactory. I used to do this inside a vagrant box with
Vagrant.configure(2) do |config|
config.ssh.insert_key = 'true'
config.vm.synced_folder "/var/b/shared", "/var/b/shared", type:
config.vm.provider :libvirt do |domain|
domain.memory = 6096
domain.cpus = 2
host = 'imagefactory'
box = 'fedora/31-cloud-base'
config.vm.define host do | tmp |
tmp.vm.hostname = host
tmp.vm.box = box
# Install tools for rpm-ostree
config.vm.provision "shell", inline: <<-SHELL
dnf install -y git imagefactory imagefactory-plugins* libvirt-client pykickstart
systemctl enable libvirtd --now # the --now should work but doesn't for some
sed -i.orig 's/\# memory = 1024/memory = 2048/' /etc/oz/oz.cfg
sudo imagefactory --debug base_image --file-parameter install_script fedora-cloud-base.ks
--parameter offline_icicle true ./fedora-rawhide.tdl
qemu-img convert -O qcow2
For the tdl file you can grab one from a recent koji build:
you can replace the repos in there with other ones like the ones from release day if you
Can we just build a normal VM image on libvirt with a suitable kickstart file and use
If you are just messing around with an image for your use I would just use `virt-install
give anaconda a kickstart file. Something like:
virt-install --name builder1 --ram 4500 --vcpus 4 --disk
--initrd-inject /path/fo/kickstart.cfg --force --noreboot --network bridge=virbr0
Looking for solutions that aren't too arduous as most students have limited Linux
+1 - This stack is a bit daunting. Depending on what you are trying to do you might give
a try as it's much easier to get that building locally. See
Once you get up and running it would be `cosa build && cosa buildextend-aws` then
take that file and
upload it into ec2.
> cloud mailing list -- cloud(a)lists.fedoraproject.org
> To unsubscribe send an email to cloud-leave(a)lists.fedoraproject.org
> Fedora Code of Conduct:
> List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
> List Archives: