Fedora + KVM + Bridge = Fail
Okay, I would like to run virtual machines over KVM in Fedora 10, such that the VMs would be available on the local network (192.168.1.xxx). My server has three NICs. I would like eth1 to be bridged.
I have read multiple tutorials, and they all basically revolve around:
brctl addbr br0
ifconfig eth1 0.0.0.0
brctl addif br0 eth1
ifconfig br0 192.168.1.202 netmask 255.255.255.0 up
route add -net 192.168.1.0 netmask 255.255.255.0 br0
route add default gw 192.168.1.1 br0
tunctl
iplink set tap0 up
brctl addif br0 tap0
For iptables, I have tried:
iptables -I INPUT -i br0 -j ACCEPT
service iptables save
or:
iptables -F FORWARD
iptables -I FORWARD -m physdev --physdev-is-bridged -j ACCEPT
net.ipv4.ip_forward = 1 was set.
To run a VM, I do: qemu-kvm -hda /root/VM_Images/test.img -m 512 -cdrom /dev/cdrom -boot c -net nic,macaddr=DE:AD:BE:EF:19:11 -net tap,script=/etc/sysconfig/networkscripts/qemu-ifup
I get a VM, but cannot pull a DHCP on the VM's nic. When I set it static to the local network, I can't ping other machines.
Any suggestions?
Alphadogg,
Have you retried using the new mac generation script (macgen.py) referenced in post link below:
Achieving Maximum Network Performance usinig bridging and paravirtual drivers
Have a look at this page for some tips:
http://wiki.libvirt.org/page/Networking
I got this working using virt-manager and a hand rolled bridge setup:
I used the /etc/sysconfig/network-scripts config files so it comes up at boot time:
ifcfg-vmsrvbr0 contains:
DEVICE=vmsrvbr0
ONBOOT=yes
TYPE=Bridge
STP=on
NM_CONTROLLED=No
IPV6INIT=no
ifcfg-eth1 contains:
DEVICE=eth1
ONBOOT=yes
BRIDGE=vmsrvbr0
NM_CONTROLLED=No
IPV6INIT=no
My eth1 is plugged into a switch for my DMZ because my VM's sit on the Internet. In the virt-manager GUI I specify my bridge interface (vmsrvbr0) and I end up with this running config:
/usr/bin/qemu-kvm -S -M pc -m 512 -smp 1 -name mailserver -monitor pty -boot c -drive file=/dev/VIRTUAL/MAIL,if=ide,index=0,boot=on -drive file=,if=ide,media=cdrom,index=2 -net nic,macaddr=54:52:00:25:d2:f4,vlan=0 -net tap,fd=9,script=,vlan=0,ifname=vnet0 -serial pty -parallel none -usb -vnc 127.0.0.1:0 -k en-us -soundhw es1370 brctl show vmsrvbr0 produces: bridge name bridge id STP enabled interfaces virbr0 8000.000000000000 yes vmsrvbr0 80e8.0050da5d8066 yes eth1 vnet0 vnet1
brctl show vmsrvbr0 produces:
bridge name bridge id STP enabled interfaces
virbr0 8000.000000000000 yes
vmsrvbr0 80e8.0050da5d8066 yes eth1
vnet0
vnet1
So the real interface (eth1) has no IP address. It doesn't need one. Inside my virtual machines I can assign addreses as I see fit. I have two VM's running on this host (hence vnet0 and vnet1 part of the bridge) and each VM has its own statically assigned address. I'm betting I could have plugged eth1 into the switch for my LAN and gotten DHCP assigned addresses if I wanted.
Hope that helps
Post new comment