Bug 1035470

Summary: virt-manager should allow use of arm arch for network install/boot
Product: [Fedora] Fedora Reporter: Paul Whalen <pwhalen>
Component: virt-managerAssignee: Cole Robinson <crobinso>
Status: CLOSED CANTFIX QA Contact: Fedora Extras Quality Assurance <extras-qa>
Severity: unspecified Docs Contact:
Priority: unspecified    
Version: 20CC: berrange, crobinso, danlipsitt, virt-maint
Target Milestone: ---Keywords: Reopened
Target Release: ---   
Hardware: All   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2014-01-18 00:43:29 UTC Type: Bug
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:

Description Paul Whalen 2013-11-27 21:07:59 UTC
Description of problem:
Virt-Manager only allows the import disk option when ARM is selected in ARCH, network installs and boots are also possible. 

Version-Release number of selected component (if applicable):
virt-manager-0.10.0-5.git1ffcc0cc.fc20.noarch

Comment 1 Cole Robinson 2013-12-01 23:25:40 UTC
It works for physical hardware, but it's not doable with qemu shipped in Fedora that I know of: we would need a uboot build or something that knows to kick off PXE, but that doesn't exist, so the only real option is to manually pass in a kernel/initrd.

Eventually I assume qemu will have some solution for this, maybe shipping a uboot build that works with mach-virt that gives us bios like ability to boot from disk image, pxe, etc. without having to do the manual kernel/initrd wrangling, but that doesn't exist yet.

So closing as CANTFIX (for now). I may have details wrong here so please reopen if I've missed something

Comment 2 Paul Whalen 2013-12-03 17:20:20 UTC
To test the F20-TC's on a F20 host I've been using:

# Image
sudo qemu-img create Fedora-qemu-armhfp-20-TC4.img 8G

# text or VNC install
virt-install --name Fedora-qemu-armhfp-20-TC4 \
--ram 1024 --arch armv7l --machine vexpress-a9 \ --location=http://dl.fedoraproject.org/pub/alt/stage/20-TC4/Fedora/armhfp/os/ \
--disk Fedora-qemu-armhfp-20-TC4.img '--extra-args=console=ttyAMA0 rw' \
--os-variant fedora20 --graphics none

Kickstart installations are also possible when added to the extra-args. 

While the PXE boot option will not work as you've mentioned, can we enable the Network Install option?

Comment 3 Cole Robinson 2013-12-03 17:32:52 UTC
Sure that option works for the first boot, but not for every subsequent boot because the downloaded kernel/initrd are the transient install kernels.

We basically have the problem that if we provide an 'install' option, we then need to pull out the kernel/initrd from inside the disk image after the install is completed, to manually pass them in on the host qemu command line. There isn't any clean way to do this with virt-manager and it's actually a maintenance nightmare since subsequent updates of the guest kernel aren't available on the host.

If all you want to do is anaconda validation, you can use that virt-install command line. But since there are subtle and impactful differences between what users expect of virt-manager URL install and what we can currently do with ARM URL install, I don't want to expose it in virt-manager because it will cause more user conversion than it's worth.

Comment 4 Cole Robinson 2014-01-18 00:43:29 UTC
Given comment #3, closing again as CANTFIX. We need a proper solution out of qemu before we can expose those install concepts in the UI