There is a new version of virt-manager available. Reproducible: Always
Version 0.6.0: http://virt-manager.et.redhat.com/download.html
Ok, but playing around with the ebuilds, I've managed to get this working nicely on my machine with kvm. I've never been able to get this thing working before, so I'm pleased this finally works. Here's what I had to do... 1) Get libvirt upgraded to 0.4.6 2) Install virtinst 0.400.0 3) Install virt-manager 0.6.0 Once everything was installed, I had some issues because the kvm and tun module wouldn't get automatically loaded. Once I loaded these, I added myself to the kvm group, and specified the following options in /etc/libvirt/libvirtd.conf: unix_sock_group = "kvm" unix_sock_rw_perms = "0770" For some reason, my regular user still wasn't able to modify virtual machines from within virt-manager and it kept saying I was in RO mode. After looking at the python source, I found the following in __init__: def __init__(self, config, uri, readOnly = None): self.__gobject_init__() self.config = config self.connectThread = None self.connectThreadEvent = threading.Event() self.connectThreadEvent.set() self.connectError = None self.uri = uri if self.uri is None or self.uri.lower() == "xen": self.uri = "xen:///" self.readOnly = readOnly if not self.is_remote() and os.getuid() != 0 and self.uri != "qemu:///session": hasPolkit = False for f in LIBVIRT_POLICY_FILES: if os.path.exists(f): hasPolkit = True # if not hasPolkit: # self.readOnly = True Notice near the bottom where I commented out the polkit check. It looks like virt-manager will only allow a regular user RW privs if polkit is used. Since we aren't using polkit on gentoo, I think we should have a patch for this. I've included a tarball of all this with the exception of the connection.py patch. Hope this helps!
Created attachment 166568 [details] ebuilds
*** Bug 239386 has been marked as a duplicate of this bug. ***
(In reply to comment #3) > Created an attachment (id=166568) [edit] > ebuilds > Michael Spiegle your ebuilds works fine, thanks for that. My current problem is that i can not see my eth0/br0 bridge in virt-manager, but i'm able to create new virbrX.
So when you start libvirtd, do you get a virbr0 device in ifconfig? I believe libvirtd is responsible for creating the virbr0 based on configs in /etc/libvirt/*/networks/default.xml. As long as virbr0 is there, I think it should work. Have you tried removing/creating your own NIC based on the "default" network? I noticed that I had to run virt-manager as root to get anything working right. I just upgraded to 2.6.27 and libvirtd/brctl appears to be having issues: [ 3770.008751] virbr0: Dropping NETIF_F_UFO since no NETIF_F_HW_CSUM feature. [ 3770.011207] virbr0: starting userspace STP failed, starting kernel STP [ 3835.784194] virbr0: Dropping NETIF_F_UFO since no NETIF_F_HW_CSUM feature. [ 3835.786165] virbr0: starting userspace STP failed, starting kernel STP [ 3873.536508] virbr0: Dropping NETIF_F_UFO since no NETIF_F_HW_CSUM feature. [ 3873.540052] virbr0: starting userspace STP failed, starting kernel STP [ 4006.405179] virbr0: Dropping NETIF_F_UFO since no NETIF_F_HW_CSUM feature. [ 4006.408368] virbr0: starting userspace STP failed, starting kernel STP Mike (In reply to comment #5) > (In reply to comment #3) > > Created an attachment (id=166568) [edit] > > ebuilds > > > > Michael Spiegle your ebuilds works fine, thanks for that. My current problem is > that i can not see my eth0/br0 bridge in virt-manager, but i'm able to create > new virbrX. >
Committed virt-manager-0.6.0 to the tree. It does not yet include a fix for the silly check for policykit but I've filed a bug with a crude patch for it upstream. Hopefully I will get some feedback on that soon.