diff --git a/doc/source/host-cloud/configuration.rst b/doc/source/host-cloud/configuration.rst index b81536c..4d461cb 100644 --- a/doc/source/host-cloud/configuration.rst +++ b/doc/source/host-cloud/configuration.rst @@ -31,25 +31,8 @@ host cloud. If this cannot be done (perhaps because you don't have access to make such a change on the host cloud), it will likely be necessary to - configure a smaller MTU on the deployed virtual instances. For a - TripleO undercloud, this can be done by setting the ``local_mtu`` - option in ``undercloud.conf`` to a smaller value (1450 will - usually work). - - .. note:: - In older versions of TripleO it may be necessary to do the MTU - configuration manually. That can be done with the following - commands (as root):: - - # Replace 'eth1' with the actual device to be used for the - # provisioning network - ip link set eth1 mtu 1350 - echo -e "\ndhcp-option-force=26,1350" >> /etc/dnsmasq-ironic.conf - systemctl restart 'neutron-*' - - If network isolation is used in the virtual deployment, the templates must - also configure mtu as discussed above, except the mtu should be set to 1350 - instead of 1550. + configure a smaller MTU on the deployed virtual instances. Details + on doing so can be found on the :doc:`../usage/usage` page. #. Restart ``nova-compute`` and ``neutron-openvswitch-agent`` to apply the changes above. diff --git a/doc/source/usage/usage.rst b/doc/source/usage/usage.rst index 920ee4c..eee475a 100644 --- a/doc/source/usage/usage.rst +++ b/doc/source/usage/usage.rst @@ -20,7 +20,52 @@ After an OVB environment has been deployed, there are a few things to know. .. note:: **Do not** rebuild the bmc. It is unnecessary and not guaranteed to work. -#. If using the full network isolation provided by OS::OVB::BaremetalNetworks - then a TripleO overcloud can be deployed in the OVB environment by using - the network templates in the ``network-templates`` (for ipv4) or - ``ipv6-network-templates`` (for ipv6) directories. +#. If the host cloud's tenant network MTU is 1500 or less, it will be necessary + to configure the deployed interfaces with a smaller MTU. The tenant network + MTU minus 50 is usually a safe value. For the undercloud this can be done + by setting ``local_mtu`` in ``undercloud.conf``. + + .. note:: + In Mitaka and older versions of TripleO it will be necessary to do the + MTU configuration manually. That can be done with the following + commands (as root):: + + # Replace 'eth1' with the actual device to be used for the + # provisioning network + ip link set eth1 mtu 1350 + echo -e "\ndhcp-option-force=26,1350" >> /etc/dnsmasq-ironic.conf + systemctl restart 'neutron-*' + +#. If using the full network isolation provided by one of the + ``all-networks*.yaml`` environments then a TripleO overcloud can be deployed + in the OVB environment by using the network templates in the + ``overcloud-templates`` directory. The names are fairly descriptive, but + this is a brief explanation of each: + + - **network-templates:** IPv4 multi-nic. Usable with the network layout + deployed by the ``all-networks.yaml`` and + ``all-networks-port-security.yaml`` environments. + - **ipv6-network-templates:** IPv6 multi-nic. Usable with the network layout + deployed by the ``all-networks.yaml`` and + ``all-networks-port-security.yaml`` environments. + - **bond-network-templates:** IPv4 multi-nic, with duplicate `public` + interfaces for testing bonded nics. Usable with the network layout + deployed by the ``all-networks-public-bond.yaml`` and + ``all-networks-public-bond-port-security.yaml`` environments. + + The undercloud's ``public`` interface should be configured with the address + of the default route from the templates in use. Firewall rules for + forwarding the traffic from that interface should also be added. The + following commands will make the necessary configuration:: + + cat >> /tmp/eth2.cfg <