You can configure router options like IP addresses or settings of routing protocol by ansible as you like. Note that ansible generally uses for initial configurations. Once you configure by ansible and start Beluganos, you can change the configuration by not only ansible but also Linux style.
- The installation is required in advance. Please refer install.md before proceeding.
- The setup of Beluganos is required in advance. Please refer setup.md before proceeding.
- In case you will use OpenNSL or OF-DPA:
- The installation of OpenNetworkLinux for your white-box switches is required in advance. Please refer setup-hardware.md before proceeding.
- The setup for ASIC API is required in advance. Please refer setup-onsl.md or setup-ofdpa.md.
The files under etc/playbooks are configuration files. In this page, lxd-*.yml and the files under roles/lxd are described.
beluganos/
etc/
playbooks/
hosts # Container name (inventory)
lxd-sample.yml # Sample playbook of container basic settings
lxd-<group-name>.yml # Container basic settings (playbook)
roles/
lxd/
vars/ # Do NOT edit
handlers/ # Do NOT edit
tasks/ # Do NOT edit
files/
common/ # Do NOT edit
sample/ # Sample files of router settings
<container-name>/ # Router settings
lxd_profile.yml
netplan.yaml
sysctl.conf
daemons
frr.conf
gobgp.comnf
gobgpd.conf
fibc.yml
ribxd.conf
At first, you should edit inventory file at etc/playbooks/hosts. The container name and group name should be described here. This name will be only used internal configurations.
In general, one-to-one correspondence between group name and container name is assumed except VRF environments. For example, if your group name is "lxd-group" and your container name (i.e. routing instance name) is "master", please edit as follows:
$ cd ~/beluganos/etc/playbooks/
$ vi hosts
---
[hosts] # DO NOT EDIT.
localhost # DO NOT EDIT.
[lxd-group]
master
For VRF environments only: To configure multiple routing-instances, you should describe multiple container name.
[lxd-group-vpn]
master
vrf10
vrf20
The sample playbook is etc/playbooks/lxd-sample.yml. You may copy this file and rename to your group name. For example, if your group name at inventory is "lxd-group", please edit following:
$ cd ~/beluganos/etc/playbooks
$ cp lxd-sample.yml lxd-group.yml
$ vi lxd-group.yml
---
- hosts: hosts
connection: local
vars:
bridges: []
roles:
- bridge
tags:
- bridge
- hosts: hosts
connection: local
roles:
- { role: lxd, mode: host }
tags:
- host
- hosts: hosts
connection: local
tasks:
- include_role:
name: lxd
vars:
mode: create
with_items: # Edit "lxd-group" to your group name in inventry file.
- "{{ groups['lxd-group'] }}"
loop_control:
loop_var: lxcname
tags:
- create
- lxd
- hosts: lxd-group # Edit "lxd-group" to your setting in inventry file.
connection: lxd
roles:
- { role: lxd, lxcname: "{{ inventory_hostname }}", mode: setup }
tags:
- setup
- lxd
The reflection of "Settings for Linux Containers" should be done after editing "Router settings".
The sample files are under etc/playbboks/roles/lxd/files/sample. You have copied this folder in previous. If your container name is master, you can copy by following commands:
$ cd ~/beluganos/etc/playbooks/roles/lxd/files
$ cp -r sample master
$ ls master
daemons fibc.yml frr.conf gobgp.conf lxd_proifle.yml gobgpd.conf netplan.yaml ribxd.conf sysctl.conf
In this files, you can determine interface name of your switch like "eth1". The sample is following:
- name: create profile
lxd_profile:
name: "{{ lxcname }}"
state: present
config: {"security.privileged": "true"}
devices:
eth0: # For internal use only.
name: eth0
nictype: bridged
parent: lxdbr0
type: nic
eth1: # Interface name in Linux container
type: nic
name: eth1 # Interface name in Linux container
host_name: "{{ lxcname }}.1" # Interface name in Host OS
nictype: p2p
eth2:
type: nic
name: eth2
host_name: "{{ lxcname }}.2"
nictype: p2p
eth3:
type: nic
name: eth3
host_name: "{{ lxcname }}.3"
nictype: p2p
eth4:
type: nic
name: eth4
host_name: "{{ lxcname }}.4"
nictype: p2p
root:
path: /
pool: default
type: disk
The syntax is following:
- name: create profile
lxd_profile:
name: "{{ lxcname }}"
state: present
config: {"security.privileged": "true"}
devices:
eth0:
name: eth0
nictype: bridged
parent: lxdbr0
type: nic
<lxc-interface-name>:
type: nic
name: <lxc-interface-name>
host_name: "{{ lxcname }}.<lxc-interface-number>"
nictype: p2p
....
root:
path: /
pool: default
type: disk
- devices
eth0: The management interface for internal use. Do NOT edit.<lxc-interface-name>: The interface name of Linux container. This interface name should be unique between other containers.- type: nic. Do NOT edit.
- name (
<lxc-interface-name>): The interface name of Linux container. Should be same value as this section. - host_name (
{{ lxcname }}.<lxc-interface-number>): Beluganos's main module recognize container's interface by this value.<lxc-interface-number>should be a sequential number starting from 1 for each container. - nictype: p2p. Do NOT edit.
root: Storage pool settings for container. Do NOT edit.
In fibc.yml, you may set general settings. FIBC is one of the main components of Beluganos.
Especially, you can determine about interface mapping between your white-box switches and Linux containers. In Beluganos's architecture, the path information which was installed to Linux containers will be copied to white-box switches. This is because you have to configure interface mapping between switches and containers.
Note that mapping between physical interface and logical interface is sometimes different at OpenNSL case. For more detail, please check configure-portmapping.md.
Please note that only physical interface should be described at fibc.yml. Logical (VLAN) settings should be described another files (netplan.yml).
routers:
- desc: sample-router
re_id: 10.0.1.6 # Router entity id (=router-id)
datapath: switchA # dpname
ports:
- { name: eth1, port: 1 } # nid=0, lxc-iface=eth1, dp-port=1
- { name: eth2, port: 2 } # nid=0, lxc-iface=eth2, dp-port=2
Syntax is following:
routers:
- desc: <description>
re_id: <router-entity-id>
datapath: <switch-name>
ports:
- { name: <lxc-interface-name>, port: <switch-interface-number> } # For physical interface
- desc (
<description>): Your preferable descriptions. - re_id (
<router-entity-id>): Router entity id for internal use. In general, router-id is recommended.- For VRF environments only: Please set master instance's router-id for all routing instances.
- datapath (
<switch-name>): Your switch name which is declared atetc/playbooks/dp-sample.yml. Please refer setup.md. - ports
- name (
<lxc-interface-name>): Specify interface name in Beluganos's Linux container.- Note:
<lxc-interface-name>should be matched with the value inlxd_profile.yml.
- Note:
- port (
<switch-interface-number>): Specify interface number of white-box switches.
- name (
In latest Ubuntu, netplan is used for interface configurations. This file will be located to /etc/netplan/02-beluganos.yaml in linux container.
For example, you can configure subinterface (VLAN) settings by this file. "eth3", "eth3.10" and "eth4.20" will be configured in following examples:
---
network:
version: 2
ethernets:
eth3:
mtu: 1500 # note: MTU is not supported yet by netplan.
vlans:
eth3.10:
link: eth3 # Physical interface
id: 10 # VLAN-ID
eth4.20:
link: eth4 # Physical interface
id: 20 # VLAN-ID
For more detail, please refer netplan design.
This file will be located under /etc/sysctl.d/ in Linux container. If you want to enable MPLS for all interfaces, you have to disable rp_filter and set net.mpls.conf as follows:
# -*- coding: utf-8 -*-
# Max of MPLS label.
net.mpls.platform_labels=10240
# Disable rp filter
net.ipv4.conf.default.rp_filter=0
net.ipv4.conf.all.rp_filter=0
# Disable rp filter
net.ipv4.conf.eth1.rp_filter=0
net.ipv4.conf.eth2/100.rp_filter=0
# Enable MPLS
net.mpls.conf.eth1.input=1
net.mpls.conf.eth2/100.input=1
This is FRRouting setting. Please note that IP Multi-cast routing is not supported yet.
zebra=yes
bgpd=no
ospfd=yes
ospf6d=no
ripd=no
ripngd=no
isisd=no
pimd=no
ldpd=yes
nhrpd=no
For MPLS-VPN environments only: In MPLS-VPN, only GoBGP is supported. Do NOT enable bgpd of FRRouting.
Each daemon configuration of FRRouting. See FRRouting user-guide.
This file is about gobgp booting, NOT gobgpd itself. In general, please do not edit.
# -*- coding: utf-8 -*-
CONF_PATH = /etc/frr/gobgpd.conf
CONF_TYPE = toml
LOG_LEVEL = debug
PPROF_OPT = --pprof-disable
API_HOSTS = 127.0.0.1:50051
GoBGP configurations. See GoBGP configurations.md.
For MPLS-VPN environments only: In MPLS-VPN, some optional settings about GoBGP is required. For more detail, please check feature-l3vpn.md.
This is the main configuration file of Beluganos itself. For example, following configuration will be assumed:
# -*- coding: utf-8; mode: toml -*-
[node]
nid = 0
reid = "10.0.1.6"
# label = 100000
allow_duplicate_ifname = false
[log]
level = 5
dump = 0
[nla]
core = "127.0.0.1:50061"
api = "127.0.0.1:50062"
[ribc]
fibc = "192.169.1.1:50070"
[ribs]
disable = true
# core = "sample:50071"
# api = "127.0.0.1:50072"
# [ribs.bgpd]
# addr = "127.0.0.1"
# # port = 50051
# [ribs.nexthops]
# mode = "translate"
# args = "1.1.0.0/24"
[ribp]
api = "127.0.0.1:50091"
interval = 5000
Except MPLS-VPN environments, the syntax is following:
# -*- coding: utf-8; mode: toml -*-
[node]
nid = 0
reid = "<router-entity-id>"
allow_duplicate_ifname = <allow-duplicate-ifname>
[log]
level = <beluganos-log-level>
dump = <beluganos-debug>
[nla]
core = "127.0.0.1:50061"
api = "127.0.0.1:50062"
[ribc]
fibc = "192.169.1.1:50070"
[ribs]
disable = true
[ribp]
api = "127.0.0.1:50091"
interval = <beluganos-ribp-interval>
- [node]
- nid (
<instance-number>): Linux container ID. Except VRF environments, you should set to0to instance number. - reid (
<router-entity-id>): Router entity id for internal use. In general, router-id is recommended. This value should match the value infibc.yml. - allow_duplicate_ifname (
<allow-duplicate-ifname>): The settings to allow overlapping of interface names between different containers, except "lo" and "eth0". Default settings is "false".
- nid (
- [log]
- level (
<beluganos-log-level>): Log level. 0〜5. 5 is the most detailed value. - dump (
<beluganos-debug>): Debug flag. If you don't need debug log, please set to 0.
- level (
- [nla]
- The setting about netlink abstraction module.
- core: set
localhost:50061expect MPLS-VPN environments. - api: NLA API address. set
localhost:50062. Do not change.
- [ribc]
- The setting about RIB controller.
- fibc: FIBC address. set
192.169.1.1:50070.
- [ribs]
- The setting about RIBS module. This module will enable route redistribution function between routing instances.
- disable: set true. Only in MPLS-VPN environments, this module will be used.
- [ribp]
- The setting about RIBP which is one of the components of RIB controller.
- api: set
localhost:50091. Do NOT change. - interval (
<beluganos-ribp-interval>): The interval value which synchronize interface status between Linux container and main module of Beluganos. By our verification,5000is recommended value for stable operations is assumed.
Note: In MPLS-VPN environments, more configurations are required. Please refer to feature-l3vpn.md.
In case group name is "lxd-group" and container name is "master":
$ cd ~/beluganos/etc/playbooks
$ ansible-playbook -i hosts -K lxd-group.yml
$ lxc stop master
This ansible-playbook commands will send your files to LXC like fibc.yml, ribxd.conf, and frr.conf.
Before configure router settings like IP address or routing protocol, you have to start the process of Beluganos. Required steps are following:
Please refer operation.md for more detail.
In ansible, only startup-config can be edited. If you want to change the running-config directly, please refer configure.md.