I'm testing ovirt node, both 4.1.5 and 4.1.6rc2.

I see that in both versions the network stack is configured with NetworkManager enabled.

With plain CentOS I always disabled NetworkManager and only used classical network service alone, so I have not background with NM, related to hypervisor.

I have configured a vlan (untagged) for VMs, using bonding mode=4 (802.3ad).

I have some problems with my VMs: they are reachable from some parts and not from others.

In parallel I have already asked network guys to crosscheck network configurations of the datacenter, but I would like to have confirmation that other ones are already using with success the combination of:

- NetworkManager on hypervisor

- bonding mode = 4 for VMs traffic

Another question for another aspect I never understood: with an environment in the past (before 3.5 times if I remember well) I used 802.3ad but I had to set lacp_rate=1 (fast) because I had connection problems with the default lacp_rate=0 (slow) parameter.

I see that still the default is lacp_rate=0

Could it be the same this time? Is there any reason to set default lacp_rate=0 in oVirt (or the reverse... ;-)?  

Thanks in advance,

Gianluca