On 03 Dec 2015, at 22:59, Jamie Lawrence
<jlawrence(a)squaretrade.com> wrote:
Hello,
I’m currently playing with scripted VM creation and have an issue getting the newly
created VM to PXE boot. My objective is to port some creation scripts from the environment
this will eventually replace and worry about making this more “Ovirt-ey” later.
Skipping the details, everything is happy through creation, and when I boot it, it
attempts to boot from a ‘disk' and fails, and I don’t understand why.
Following the creation of the VM, creation/attachment of the disk and net,
boot_dev = params.Boot(dev='network')
vm.os.set_boot([boot_dev])
vm.set_os(params.OperatingSystem(cmdline=kernel_cmd))
vm.update()
kernel_cmd there evaluates to a fairly typical PXEboot string that works with our legacy
setup - there isn’t really anything exotic going on. The BIOS doesn’t show any attempt at
a PXE boot - it goes straight to the disk, declares it unbootable (because it was just
created and is blank), and halts. It feels like the set_boot line is wrong or ignored, but
this is new to me.
Hi,
if you look for a one-time boot off PXE, and then from hdd after installation finishes
(which is the most frequent use case), you can use “Run Once” to override the “default”
boot order. Instead of modifying the VM configuration twice.
Nevertheless, the above should work, so I would start checking if the VM is indeed
configured correctly (bug on ovirt side) or we create it ok but there’s a bug in lower
layers (libvirt, qemu, ipxe). If you can check the VM in UI there is a “VM Devices” subtab
new in 3.6 which can tell you if it even has a NIC attached; and there is the final
libvirt domain xml being logged in /var/log/vdsm/vdsm.log when you launch the VM - there
you can doublecheck the VM is started as expected
Thanks,
michal
Anyone have a hint?
Thanks,
-j_______________________________________________
Users mailing list
Users(a)ovirt.org
http://lists.ovirt.org/mailman/listinfo/users