Andreas Rittershofer napsal(a):
My problem is:
1.) VM1 running on host A must have network access to VM3 when VM3 is
running on host B by default _and_
2.) when VM3 is running on host A because host B failed.
In some of my configurations 1.) worked but 2.) not, in some other
configurations 2.) worked but 1.) not.
I must have missed some detail ...
I do not see the problem. With the bridged scenario every VM has direct
access to the physical network and it is not important on which host it
runs. If the VMs are running on the same host, their ethernet frames
will just "turn around" in the bridge interface and if they are running
on different hosts they will go thru the real ethernet.
State 1)
A: VM1
B: VM3
Eth. frames will go like this:
VM1->BrA->PhysCardA->Ethernet->PhysCardB->BrB->VM3
State 2)
A: VM1+VM3
B: x
Eth. frames will go like this:
VM1->BrA->VM3
This is how it should work, for more debugging I would suggest running
tcpdump on all interfaces and pinging the hosts - tcpdump on bridge may
not work correctly (I am not sure about it) but on the other (vnet and
eth) interfaces it will work. Did you setup the firewall? I need to do
#iptables -I FORWARD -m physdev --physdev-is-bridged -j ACCEPT
to allow bridged packets thru the firewall.
Radek