Red Hat Bugzilla – Bug 476732
fully automated qemu install fails at Network Manager
Last modified: 2009-04-20 10:31:38 EDT
Description of problem:
qemu installs through python-virtinst for F10 guests seem to be having a DHCP timeout issue.
Installing a kickstart that is proven to work on a /dev/kvm system with virtio or a PXEd physical host using cobbler/koan does not yield the same results for a regular qemu install.
Network Manager fails to configure the network interface on the first try and hitting "retry" in Anaconda solves the problem and the install continues normally.
-- host does not matter
-- guest must be running F-10
-- guest must be created using qemu
Version-Release number of selected component (if applicable):
(0.5.0-1.fc10.i386) F-10 latest everything at the time, non-testing
seems 100% Consistent
Steps to Reproduce:
1. See description
NetworkManager in Anaconda fails to configure the interface requiring the user to hit retry
Fully automated install is successful
Talked with this on IRC with crobinson in #virt/OFTC, said to file against libvirt and it could be triaged to other components as needed.
Does this still happen?
If you "virsh net-stop default; virsh net-start default" before starting the install, does that help?
Attach output from:
$> brctl show
$> iptables -L -v -n
$> ps -ef | grep dnsmasq
$> ifconfig -a
$> cat /proc/sys/net/ipv4/ip_forward
Missing info from reporter, closing
This seems to have been better lately, at least when using real bridges. I'll let you know if it happens again.