<div class="gmail_quote">On Fri, Sep 10, 2010 at 12:36 PM, Justin Clacherty <<a href="mailto:justin@redfish.com.au">justin@redfish.com.au</a>> wrote: <blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"> Thanks Nicolas, On 10/09/2010 6:04 PM, Nicolas Ochem wrote: <blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"> 1. managing images using cobbler I have not tried to do this. I import the NFS share in ovirt hardware pool. Then, the images are imported as libvirt storage volumes and they are available when I create a VM. Cobbler version has been upgraded between fedora 11 and fedora 13. You are probably the first one trying this under f13. Is there a particular reason why you are using cobbler for this purpose ? </blockquote> I was using cobbler because that's the way I first did it many moons ago when I started using ovirt (2008 I think), and in the absence of documentation to the contrary.... :-) I did try the NFS way but it didn't seem to work for me. I most likely didn't do it correctly. What I did was add the NFS share in the storage pool section. When creating the VM I selected the appropriate iso in the storage section as well as the iscsi volume I wanted to install too. When I started the VM it just said the disk wasn't bootable. What's the correct way to do this? Is there a way to specifiy boot order when you've selected multiple storage pools for the VM? </blockquote><div> No, because libvirt has limited support for that. You can select whether to boot from fd, hd, cdrom, or network, but you cannot specify a device in particular. See <a href="http://libvirt.org/formatdomain.html#elementsOSBIOS">http://libvirt.org/formatdomain.html#elementsOSBIOS</a> What could be done is to add "cdrom" to ovirt boot options. Then it would pass it to libvirt. But I'm afraid you have to submit the patch yourself :( I assume the general idea is to prepare your VM using your desktop virt management GUI and then move the prepared VMs to ovirt. That's how I do it. </div><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"> <blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"> 2. booting from cdrom I assume adding the cdrom iso in ovirt would also fix that issue. Besides you will just encounter it once, for installation. Afterwards, you will always boot from HD. </blockquote> Yes, switched to boot from HD now. If cobbler is the expected way to install though then this problem really needs to be fixed. Though from (1) I'm assuming cobbler isn't the expected way to install from an ISO. </blockquote><div> This "~" issue sure sounds like a bug, but again, probably you're the best candidate to submit a fix. </div><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"> <blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"> 3. too old qemu-kvm You have several choices here : a. download a more recent RPM and install it in the nodes manually each time you start them b. enable the fedora virtualization preview repository in your build environment and build a newer ovirt-node-image.iso c. try to build ovirt under fedora 14 </blockquote> That one was a false alarm, I sent a separate email to the list on that one. The node was out of memory (oom kernel panic in /var/log/messages), not sure why as there's only 6GB assigned to VMs and there's 8GB in the machine. Anyway, restarting the node seems to have fixed that issue. Cheers, Justin. </blockquote></div>