Modular Layer 2 (ml2) configuration options

Modular Layer 2 (ml2) Flat Type configuration options
Modular Layer 2 (ml2) GRE Type configuration options
Modular Layer 2 (ml2) VLAN Type configuration options
Modular Layer 2 (ml2) VXLAN Type configuration options
Modular Layer 2 (ml2) Arista Mechanism configuration options
Modular Layer 2 (ml2) BigSwitch Mechanism configuration options
Modular Layer 2 (ml2) Brocade Mechanism configuration options
Modular Layer 2 (ml2) Cisco Mechanism configuration options
Modular Layer 2 (ml2) Mellanox Mechanism configuration options
Modular Layer 2 (ml2) OpenDaylight Mechanism configuration options
Modular Layer 2 (ml2) OpenFlow Mechanism configuration options
Modular Layer 2 (ml2) L2 Population Mechanism configuration options
Modular Layer 2 (ml2) Tail-f NCS Mechanism configuration options

The Modular Layer 2 (ml2) plug-in has two components: network types and mechanisms. You can configure these components separately. This section describes these configuration options.

MTU bug with VXLAN tunnelling

	MTU bug with VXLAN tunnelling
Due to a bug in Linux Bridge software maximum transmission unit (MTU) handling, using VXLAN tunnels does not work by default. A simple workaround is to increase the MTU value of the physical interface and physical switch fabric by at least 50 bytes. For example, increase the MTU value to 1550. This value enables an automatic 50-byte MTU difference between the physical interface (1500) and the VXLAN interface (automatically 1500-50 = 1450). An MTU value of 1450 causes issues when virtual machine taps are configured at an MTU value of 1500. Another workaround is to decrease the virtual ethernet devices' MTU. Set the `network_device_mtu` option to 1450 in the `neutron.conf` file, and set all guest virtual machines' MTU to the same value by using a DHCP option. For information about how to use this option, see Configure OVS plug-in.

Due to a bug in Linux Bridge software maximum transmission unit (MTU) handling, using VXLAN tunnels does not work by default.

A simple workaround is to increase the MTU value of the physical interface and physical switch fabric by at least 50 bytes. For example, increase the MTU value to 1550. This value enables an automatic 50-byte MTU difference between the physical interface (1500) and the VXLAN interface (automatically 1500-50 = 1450). An MTU value of 1450 causes issues when virtual machine taps are configured at an MTU value of 1500.
Another workaround is to decrease the virtual ethernet devices' MTU. Set the network_device_mtu option to 1450 in the neutron.conf file, and set all guest virtual machines' MTU to the same value by using a DHCP option. For information about how to use this option, see Configure OVS plug-in.

Table 7.11. Description of configuration options for ml2
Configuration option = Default value	Description
[ml2]
mechanism_drivers =	(ListOpt) An ordered list of networking mechanism driver entrypoints to be loaded from the neutron.ml2.mechanism_drivers namespace.
tenant_network_types = local	(ListOpt) Ordered list of network_types to allocate as tenant networks.
type_drivers = local, flat, vlan, gre, vxlan	(ListOpt) List of network type driver entrypoints to be loaded from the neutron.ml2.type_drivers namespace.

Questions? Discuss on ask.openstack.org
Found an error? Report a bug against this page

Legal notices