Does anyone have experience using Red Hat 8, kvm and open vswitch? I have a RHEL8 host using kvm with open vswitch but I am unable to get connectivity working between guest machines. The setup is as follows: Physical Switch → trunk port → eno1 (host NIC) → ovs bridge0 → ovs port/interface (van 24) → guest The management interface is setup on the second physical NIC and I am only testing with vlan24 at the moment so once this up and working then I will replicate/migrate the other vlans I need. All required vlans are allowed on the trunk port of the physical switch, connectivity from the guests to the physical network is working as expected, I can ping from guest in vlan 24 to a physical host in vlan 24. The issue is that I cannot get traffic to pass directly between the guests in the same vlan (vlan24). I have both guests attached to the same ovs port/interface and the ovs port is configured with the correct vlan tag. I checked the ovs arp table and I can see the MAC addresses of the physical host and the guest machines so ovs is correctly learning the MAC addresses: Code: # ovs-appctl fdb/show bridge0 port VLAN MAC Age 12 24 3c:df:1e:97:ce:c2 11 12 0 3c:df:1e:97:ce:94 5 33 24 52:54:00:0b:19:1e 0 33 24 52:54:00:3d:cb:9d 0 Below is the output of the port/interface setup from nmcli: Code: # nmcli conn show NAME UUID TYPE DEVICE eno2 40309b82-edec-449e-a703-975288f2e8e8 ethernet eno2 bridge0 b4811e6f-29da-4625-aa72-e05c310c07bb ovs-bridge bridge0 bridge0-ovs-int b2317363-ae1e-437e-88a2-c6ad59385719 ovs-interface bridge0 bridge0-ovs-port 05170645-df46-4426-88da-3550ec754b77 ovs-port bridge0-ovs-port eno1 e2c3c790-34b3-403d-ab7e-e9ca6e30a6ab ethernet eno1 vlan100 e3102987-ab5a-48fa-9977-4ce141c8cc48 vlan vlan100 vlan23 59f9dafc-3a7f-48d5-95bd-2ce8fd080b44 vlan vlan23 vlan24-ovs-int 25ae35eb-9d5f-4036-8244-f13a41306829 ovs-interface vlan24 vlan24-ovs-port 34cb47d3-9649-4123-97e7-1b36d4af70dc ovs-port vlan24 Code: # nmcli device status DEVICE TYPE STATE CONNECTION eno2 ethernet connected eno2 eno1 ethernet connected eno1 bridge0 ovs-bridge connected bridge0 bridge0 ovs-interface connected bridge0-ovs-int vlan24 ovs-interface connected vlan24-ovs-int bridge0-ovs-port ovs-port connected bridge0-ovs-port vlan24 ovs-port connected vlan24-ovs-port vlan100 vlan connected vlan100 vlan23 vlan connected vlan23 lo loopback unmanaged -- macvtap27 macvlan unmanaged -- macvtap28 macvlan unmanaged -- eno1 ovs-port unmanaged -- The current ovs config: Code: # ovs-vsctl show dd0de2f1-4f76-41ed-8a98-605021f9667f Bridge "bridge0" Port "vlan24" tag: 24 Interface "vlan24" type: internal Port "bridge0-ovs-port" Interface "bridge0" type: internal Port "eno1" Interface "eno1" ovs_version: "2.12.0" I understand that in general linux implements some arp filtering on the host itself, but from everything that I have read this does not seem apply to ovs? In any event I have not been able to find out how to check for or disable any said filtering (specifically for RHEL8). I attempted to install ovs-tcpdump to see if this could tell me anything, but I have not been able to find a repo or rpm for RHEL8. I am using Red hat developer subscription, but it seems that this is limited when it comes to ovs/openstack repo access. This is not a complex setup, I am simply setting up network segmentation for my home labs, vpns, wireless appliances etc however I have spent a huge amount of time on this over the past couple of weeks for what really should be a pretty simple job (and learning way more than I ever thought I would about linux bridging), keeping in mind this is also the first time I have used ovs. I am just hoping that this issue is something simple I am overlooking. Appreciate if anyone has gotten a similar setup to work and could give me some hints. I have been reading a lot of ovs documentation but I have not found a reason for this problem, unless I am overlooking the obvious, to me this looks like it should be working.