Cloudstack appears to not obey physical network tags.

[Grayson Head <gr...@graysonhead.net>]

I'm trying to get a ACS environment running with two different guest
networks (one using VXLAN, the other using VLAN).

I have two physical network interfaces (cloudbr0 and cloudbr1) and intended
to use cloudbr0 for vlan isolation and cloudbr1 for vxlan isolation.

I have configured interface tags on the physical nics and network
offerings, and have confirmed in the logs that the vxlan interface on the
router is attempting to create the network on cloudbr1 as shown:

"nics": [
                    {
                        "broadcastType": "Vlan",
                        "broadcastUri": "vlan://901",
                        "defaultNic": true,
                        "deviceId": 2,
                        "dns1": "66.194.167.3",
                        "dns2": "66.194.167.4",
                        "gateway": "66.194.167.1",
                        "ip": "66.194.167.142",
                        "isSecurityGroupEnabled": false,
                        "isolationUri": "vlan://901",
                        "mac": "06:c4:e6:00:00:18",
                        "name": "cloudbr0",
                        "netmask": "255.255.255.0",
                        "networkRateMbps": 200,
                        "nicUuid": "843263cf-d69d-496f-84fd-b6bfbf55dbf9",
                        "pxeDisable": true,
                        "type": "Public",
                        "uuid": "47b26175-781e-47d8-8d37-577f6d1e1213"
                    },
                    {
                        "*broadcastType": "Vxlan", *
*                        "broadcastUri": "vxlan://1712", *
                        "defaultNic": false,
                        "deviceId": 0,
                        "dns1": "66.194.167.3",
                        "dns2": "66.194.167.4",
                        "ip": "10.1.1.1",
                        "isSecurityGroupEnabled": false,
                        "isolationUri": "vxlan://1712",
                        "mac": "02:00:01:47:00:02",
                        *"name": "cloudbr1", *
                        "netmask": "255.255.255.0",
                        "networkRateMbps": 200,
                        "nicUuid": "be7d02f5-2ef4-464d-8713-9823b3ae8bb7",
                        "pxeDisable": true,
                        "type": "Guest",
                        "uuid": "5e510ece-0afe-44a3-9cbc-9d4c587a0dce"
                    },
                    {
                        "broadcastType": "LinkLocal",
                        "defaultNic": false,
                        "deviceId": 1,
                        "gateway": "169.254.0.1",
                        "ip": "169.254.2.250",
                        "isSecurityGroupEnabled": false,
                        "mac": "0e:00:a9:fe:02:fa",
                        "netmask": "255.255.0.0",
                        "networkRateMbps": -1,
                        "nicUuid": "269431ed-c997-472c-95da-2ac26e129335",
                        "pxeDisable": true,
                        "type": "Control",
                        "uuid": "05975418-d018-449c-af0e-8948243aa76d"
                    }
                ],


For context, the full log entry is here: http://pastebin.com/TazJKFqd


However, when I look on the host, the following occurs:

ip -d link show vxlan1712
41: vxlan1712: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue
state UNKNOWN
    link/ether 8e:05:2c:dd:87:22 brd ff:ff:ff:ff:ff:ff
    vxlan id 1712 group 239.0.6.176 dev cloudbr0 port 32768 61000 ttl 10
ageing 300


Anyone have any idea what is up with this discrepancy?

I'm on ACS 4.5.2 and my hypervisors are KVM on Centos 6