Neutron Router - Not working [closed]

edit

Hi,

I'm not sure if this is related to neutron's network on the hyperv node running cloudbase nova driver or if this is related/isolated to the controller itself...

I am currently testing out OpenStack using a packstack AIO (Ocata) installation, however, I'm having a hard time to get networking up and running using neutron's router...

The packstack instance is running on a CentOS 7.3 hyperv guest with 6x NICs connected:

• eth0 = openstack data/management network
• eth1 = controller management ip
• eth2 = br-public-network
• eth3 = br-local-network
• eth4 = br-load-balancing-network
• eth5 = br-del-corp

I'm using OpenVSwitch (-vs- linuxbridge) and therefore, created the following bridges before running the packstack installer:

ovs-vsctl add-br br-public-network ; ovs-vsctl add-port br-public-network eth2
ovs-vsctl add-br br-local-network ; ovs-vsctl add-port br-local-network eth3
ovs-vsctl add-br br-load-balancing-network ; ovs-vsctl add-port br-load-balancing-network eth4
ovs-vsctl add-br br-del-corp ; ovs-vsctl add-port br-del-corp eth5

My packstack's answer file has the following details in regards to neutron/ovs:

CONFIG_NEUTRON_L3_EXT_BRIDGE=br-public-network
CONFIG_NEUTRON_ML2_TYPE_DRIVERS=vlan,flat
CONFIG_NEUTRON_ML2_TENANT_NETWORK_TYPES=vlan
CONFIG_NEUTRON_ML2_MECHANISM_DRIVERS=openvswitch,hyperv
CONFIG_NEUTRON_ML2_FLAT_NETWORKS=*
CONFIG_NEUTRON_ML2_VLAN_RANGES=physnet1,physnet2:500:2000,physnet3:2010:3010
CONFIG_NEUTRON_ML2_TUNNEL_ID_RANGES=
CONFIG_NEUTRON_ML2_VXLAN_GROUP=
CONFIG_NEUTRON_ML2_VNI_RANGES=10:100
CONFIG_NEUTRON_L2_AGENT=openvswitch
CONFIG_NEUTRON_LB_INTERFACE_MAPPINGS=
CONFIG_NEUTRON_OVS_BRIDGE_MAPPINGS=physnet1:br-public-network,physnet2:br-local-network,physnet3:br-load-balancing-network
CONFIG_NEUTRON_OVS_EXTERNAL_PHYSNET=physnet1

I've been able to create and use the following networks:

• public-network, physnet1 (flat), no router, public routeable cidr subnet, no dhcp
• private-network, physnet2 (vlan id: 501), no router, local non-routeable cidr subnet (10.0.10.0/24), no dhcp

** I'm injecting IPs instead of using DHCP..

Both networks are working fine. I can spin up instances and they can communicate to either the external network (eg: Internet) or the local network.

Now, when it comes to creating a public network that would be used for load balancing projects and therefore, would be providing a neutron router, this is where it becomes problematic... I'm creating the "load balancing network" using the following commands:

neutron net-create --provider:network_type=vlan --provider:physical_network=physnet3 --router:external=True PUBLIC-CLUSTER-NETWORK
neutron subnet-create PUBLIC-CLUSTER-NETWORK PUBLIC_CIDR_HERE/27 --gateway GW_IP_HERE --allocation-pool start=IP_START_HERE,end=IP_END_HERE --disable-dhcp --name PUBLIC-CLUSTER-SUBNET --dns-nameservers list=true 8.8.8.8 4.2.2.2
neutron router-create ROUTER-PUBLIC-CLUSTER-NETWORK
neutron router-gateway-set ID_ROUTER ID_NETWORK

Then, I'm creating a "local cluster network" that would be using the router created above:

neutron net-create --provider:network_type=vlan --provider:physical_network=physnet2 CLIENT0001-CLUSTER-NETWORK --tenant-id=e0f7fb96271f48588e2aac86d66ae42e
neutron subnet-create CLIENT0001-CLUSTER-NETWORK 192.168.23.0/24 --name CLIENT0001-CLUSTER-SUBNET --dns-nameservers list=true 8.8.8.8 4.2.2.2 --disable-dhcp
neutron router-interface-add ID_ROUTEUR ID_CLIENT_SUBNET

Result in the Neutron router IP configuration is:

# ip netns exec qrouter-78ab8780-6282-4b8b-b840-b92ba0916e62 ip a
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 scope host lo
       valid_lft forever preferred_lft forever
    inet6 ::1/128 scope host
       valid_lft forever preferred_lft forever
14: qg-0f38cb25-ae: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UNKNOWN
    link/ether fa:16:3e:76:7d:63 brd ff:ff:ff:ff:ff:ff
    inet 198.xxx.xxx.61/27 brd 198.xxx.xxx ...