Hi,

On Cent OS 6.0 I have 2 NIC's, eth0 and eth1. I have defined a bridge br0 for eth0 and br1 for eth1.

When defining new guests, I will be able to use either br0 or br1, regarding loadbalancing and security.

I have tried to simulate that eth0 breaks down. Pulling the cable out. Now the guests on br0 of course can't be accessed. The guests on br1 still works :-)

How do I change a NIC for an existing guest from br0 to br1 ?

I have tried using Virtual Machine Manager to add new NIC card, even using same MAC address af the old NIC or changing ifcfg-eth0 on the guest to new MAC addr, but this doesn't work. There no network working. Something more must be done ?

Regards.

if both NICs are on the same network, you should bond them together, and use a single bridge. You could use brctl to attach NICs to bridges, but this is not an automated or in any way a correct solution.

Thanks.

I'm okay with the non-automated solution.

You say I can use brctl instead of the Virtual Machine Manager, that doesn't work ?

I can't see why you cannot use bonding

Because I don't know how to set it up.

;-)

There are tons of guides out there, bonding is just above network basics, no magic involved

Made it work now ! :-)

Made a ifcfg-bond0. Changed ifcfg-eth0 + 1 to be af slave and master=bond0
Made a br0 for bond0 and all works !

Increased bandwidth and redundancy :-))

great, and congrats on learning something new!

Which bonding mode are you using, btw?

At the moment, I haven't used any Bonding-options yet. I know I MUST use either arp_interval and arp_ip_target OR miimon.

Have you any recommendations ?

My 4 files look like this at the moment:

DEVICE=bond0
#IPADDR=192.168.10.40
#NETMASK=255.255.255.0
ONBOOT=yes
BOOTPROTO=none
USERCTL=no
BONDING_OPTS=""
BRIDGE=br0

DEVICE=br0
NM_CONTROLLED="no"
BOOTPROTO=static
IPADDR=192.168.10.45
ONBOOT=yes
TYPE=Bridge
GATEWAY=192.168.10.1
DELAY=0
NETMASK=255.255.255.0

DEVICE=eth0
NM_CONTROLLED="no"
HWADDR=00:1B:21:C3:FD:88
ONBOOT=yes
TYPE=Ethernet
MASTER=bond0
SLAVE=yes

DEVICE=eth1
NM_CONTROLLED="no"
HWADDR=D0:27:88:33E:51
ONBOOT=yes
TYPE=Ethernet
MASTER=bond0
SLAVE=yes


Don't it look correct ? It seems to work, but I need to study the bonding-options.

cat /proc/net/bonding/bond0

Gives me this:

Ethernet Channel Bonding Driver: v3.5.0 (November 4, 2008)

Bonding Mode: load balancing (round-robin)
MII Status: up
MII Polling Interval (ms): 0
Up Delay (ms): 0
Down Delay (ms): 0

Slave Interface: eth0
MII Status: up
Link Failure Count: 0
Permanent HW addr: 5c:f3:fc:5b:02:08

Slave Interface: eth1
MII Status: up
Link Failure Count: 0
Permanent HW addr: 5c:f3:fc:5b:02:0a



dmesg say in the end:

kvm: 16154: cpu7 unhandled wrmsr: 0x198 data 0
kvm: 16154: cpu8 unhandled wrmsr: 0x198 data 0
kvm: 16154: cpu9 unhandled wrmsr: 0x198 data 0
kvm: 16154: cpu0 unimplemented perfctr wrmsr: 0xc1 data 0x0
kvm: 16154: cpu0 unimplemented perfctr wrmsr: 0x186 data 0x130079
kvm: 16154: cpu0 unimplemented perfctr wrmsr: 0xc1 data 0xffe18e99
kvm: 16154: cpu0 unimplemented perfctr wrmsr: 0x186 data 0x530079
kvm: 16154: cpu1 unimplemented perfctr wrmsr: 0xc1 data 0x0
kvm: 16154: cpu1 unimplemented perfctr wrmsr: 0x186 data 0x130079
kvm: 16154: cpu1 unimplemented perfctr wrmsr: 0xc1 data 0xffe18e99
kvm: 16154: cpu1 unimplemented perfctr wrmsr: 0x186 data 0x530079
kvm: 16154: cpu2 unimplemented perfctr wrmsr: 0xc1 data 0x0
kvm: 16154: cpu2 unimplemented perfctr wrmsr: 0x186 data 0x130079
bond0: received packet with own address as source address
bond0: received packet with own address as source address
bond0: received packet with own address as source address
bond0: received packet with own address as source address
bond0: received packet with own address as source address
bond0: received packet with own address as source address
bond0: received packet with own address as source address
bond0: received packet with own address as source address
bond0: received packet with own address as source address

Is it "normal" with theese messages ? Also from kvm ?

this is the problem - this mode is not compatible with a linux bridge. Another mode that has the same problem is balance-alb or mode - 6.

The best option for you is mode-4 or 802.3ad, but to use this mode, you need to make sure your switch supports it, and configure the switch for LACP

Ok..... well, the customers switch'es are not new and the one I just delivered with the server, just for the new rack isn't 802.3ad compliant.

So I must remove the bonding and go back to my first solution, with 2 bridges ? Without the increased bandwidth and redundancy :-(

I have google't a little, and found that some recommend using mode 5 when using bonding + bridging in a virt environment....

or mode 1 ....