openstack/kayobe
Code Issues Proposed changes

Merge "docs: Improve all-in-one scenario"

Browse Source
This commit is contained in:
Zuul 2021-12-21 16:37:56 +00:00 committed by Gerrit Code Review
commit 719968f694
2 changed files with 195 additions and 44 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

83
doc/source/configuration/scenarios/all-in-one/index.rst
View File

@ -2,8 +2,39 @@
All in one scenario
===================
.. note::
This documentation is intended as a walk through of the configuration
required for a minimal all-in-one overcloud host. If you are looking
for an all-in-one environment for test or development, see
:ref:`contributor-automated`.
This scenario describes how to configure an all-in-one controller and compute
node using Kayobe.
node using Kayobe. This is a very minimal setup, and not one that is
recommended for a production environment, but is useful for learning about how
to use and configure Kayobe.
.. _configuration-scenario-aio-prerequisites:
Prerequisites
=============
This scenario requires a basic understanding of Linux, networking and
OpenStack.
It also requires a single host running a :ref:`supported operating system
<support-matrix-supported-os>` (VM or bare metal), with:
* 1 CPU
* 8GB RAM
* 40GB disk
* at least one network interface that has Internet access
You will need access to a user account with passwordless sudo. The default user
in a cloud image (e.g. ``centos`` or ``ubuntu``) is typically sufficient. This
user will be used to run Kayobe commands. It will also be used by Kayobe to
bootstrap other user accounts.
.. _configuration-scenario-aio-overview:
Overview
========
@ -18,11 +49,11 @@ required configuration in Kayobe::
| |
| |
| +-------------+ |
| | | |
| | Containers | |
| | | |
| +-------------+ |
| |
| | |+ |
| | Containers || |
| | || |
| +-------------+| |
| +-------------+ |
| |
+---------+-------+---------+
| |
@ -33,18 +64,40 @@ required configuration in Kayobe::
|
+-----------------+------------------+ Internet
The networking in particular is relatively simple. The main interface of the
overcloud host, labelled NIC 1 in the above diagram, will be used only for
connectivity to the host and Internet access. A single Kayobe network called
``aio`` carries all control plane traffic, and is based on virtual networking
that is local to the host.
Prerequisites
=============
Later in this tutorial, we will create a dummy interface called ``dummy0``, and
plug it into a bridge called ``br0``::
This scenario requires:
+--------------+
| |
| OVS |
| |
+--------------+
|
|
+--------------+
| |
| br0 |
| 192.168.33.3 |
| 192.168.33.2 |
+--------------+
| dummy0 |
+--------+
* a basic understanding of Linux, networking and OpenStack
* a single host running a :ref:`supported operating system
<support-matrix-supported-os>` (VM or bare metal)
* at least one network interface that has Internet access
* an IP subnet with a free IP address for the OpenStack API virtual IP, and a
range of free IP addresses for external network access
The use of a bridge here allows Kayobe to connect this network to the Open
vSwitch network, while maintaining an IP address on the bridge. Ordinarily,
``dummy0`` would be a NIC providing connectivity to a physical network. We're
using a dummy interface here to keep things simple by using a fixed IP subnet,
``192.168.33.0/24``. The bridge will be assigned a static IP address of
``192.168.33.3``, and this address will by used for various things, including
Ansible SSH access and OpenStack control plane traffic. Kolla Ansible will
manage a Virtual IP (VIP) address of ``192.168.33.2`` on ``br0``, which will be
used for OpenStack API endpoints.
Contents
========

156
doc/source/configuration/scenarios/all-in-one/overcloud.rst
View File

@ -8,13 +8,35 @@ Overcloud
for an all-in-one environment for test or development, see
:ref:`contributor-automated`.
Preparation
===========
Use the bootstrap user described in :ref:`prerequisites
<configuration-scenario-aio-prerequisites>` to access the machine.
As described in the :ref:`overview <configuration-scenario-aio-overview>`, we
will use a bridge (``br0``) and a dummy interface (``dummy0``) for control
plane networking. Use the following commands to create them and assign the
bridge a static IP address of ``192.168.33.3``:
.. code-block:: console
sudo ip l add br0 type bridge
sudo ip l set br0 up
sudo ip a add 192.168.33.3/24 dev br0
sudo ip l add dummy0 type dummy
sudo ip l set dummy0 up
sudo ip l set dummy0 master br0
This configuration is not persistent, and must be recreated if the VM is
rebooted.
Installation
============
SSH to the overcloud machine, then follow the instructions in
:doc:`/installation` to set up an Ansible control host environment.
Typically this would be on a separate machine, but here we are keeping things
as simple as possible.
Follow the instructions in :doc:`/installation` to set up an Ansible control
host environment. Typically this would be on a separate machine, but here we
are keeping things as simple as possible.
Configuration
=============
@ -56,21 +78,19 @@ compute nodes would be used.
The ``inventory`` directory also contains group variables for network interface
configuration. In this example we will assume that the machine has a single
network interface called ``eth0``. We will create a bridge called ``breth0``
and plug ``eth0`` into it. This allows us to move the host's IP address to the
bridge, and pass traffic through to an Open vSwitch bridge for Neutron. Replace
the network interface configuration for the ``controllers`` group with the
following, replacing ``eth0`` with an appropriate interface:
network interface called ``dummy0``. We will create a bridge called ``br0``
and plug ``dummy0`` into it. Replace the network interface configuration for
the ``controllers`` group with the following:
.. code-block:: yaml
:caption: ``etc/kayobe/inventory/group_vars/controllers/network-interfaces``
# Controller interface on all-in-one network.
aio_interface: breth0
aio_interface: br0
# Interface eth0 is plugged into the all-in-one network bridge.
# Interface dummy0 is plugged into the all-in-one network bridge.
aio_bridge_ports:
- eth0
- dummy0
In this scenario a single network called ``aio`` is used. We must therefore set
the name of the default controller networks to ``aio``:
@ -170,7 +190,6 @@ address of the OpenStack API.
# All-in-one network.
aio_cidr: 192.168.33.0/24
aio_gateway: 192.168.33.1
aio_vip_address: 192.168.33.2
###############################################################################
@ -196,9 +215,31 @@ the ``os_distribution`` variable in ``etc/kayobe/globals.yml`` to ``ubuntu``.
.. code-block:: yaml
:caption: ``etc/kayobe/globals.yml``
---
os_distribution: "ubuntu"
Kayobe uses a bootstrap user to create a ``stack`` user account. By default,
this user is ``centos`` on CentOS, and ``ubuntu`` on Ubuntu, in line with the
default user in the official cloud images. If you are using a different
bootstrap user, set the ``controller_bootstrap_user`` variable in
``etc/kayobe/controllers.yml``. For example, to set it to ``cloud-user`` (as
seen in MAAS):
.. code-block:: yaml
:caption: ``etc/kayobe/controllers.yml``
controller_bootstrap_user: "cloud-user"
By default, on systems with SELinux enabled, Kayobe will disable SELinux and
reboot the system to apply the change. In a test or development environment
this can be a bit disruptive, particularly when using ephemeral network
configuration. To avoid rebooting the system after disabling SELinux, set
``disable_selinux_do_reboot`` to ``false`` in ``etc/kayobe/globals.yml``.
.. code-block:: yaml
:caption: ``etc/kayobe/globals.yml``
disable_selinux_do_reboot: false
In a development environment, we may wish to tune some Kolla Ansible variables.
Using QEMU as the virtualisation type will be necessary if KVM is not
available. Reducing the number of OpenStack service workers helps to avoid
@ -207,7 +248,6 @@ using too much memory.
.. code-block:: yaml
:caption: ``etc/kayobe/kolla/globals.yml``
---
# Most development environments will use nested virtualisation, and we can't
# guarantee that nested KVM support is available. Use QEMU as a lowest common
# denominator.
@ -217,10 +257,62 @@ using too much memory.
# processes to one per-service.
openstack_service_workers: "1"
We can see the changes that have been made to the configuration.
.. code-block:: console
cd <base path>/src/kayobe-config
git status
On branch master
Your branch is up to date with 'origin/master'.
Changes to be committed:
(use "git restore --staged <file>..." to unstage)
deleted: etc/kayobe/inventory/hosts.example
Changes not staged for commit:
(use "git add <file>..." to update what will be committed)
(use "git restore <file>..." to discard changes in working directory)
modified: etc/kayobe/globals.yml
modified: etc/kayobe/inventory/group_vars/controllers/network-interfaces
modified: etc/kayobe/kolla/globals.yml
modified: etc/kayobe/networks.yml
Untracked files:
(use "git add <file>..." to include in what will be committed)
etc/kayobe/inventory/hosts
etc/kayobe/network-allocation.yml
The ``git diff`` command is also helpful. Once all configuration changes have
been made, they should be committed to the kayobe-config git repository.
.. code-block:: console
cd <base path>/src/kayobe-config
git add etc/kayobe/inventory/hosts etc/kayobe/network-allocation.yml
git add --update
git commit -m "All in one scenario config"
In a real environment these changes would be pushed to a central repository.
Deployment
==========
We are now ready to perform a deployment.
Activate the Kayobe virtual environment:
.. code-block:: console
cd <base path>/venvs/kayobe
source bin/activate
Activate the Kayobe configuration environment:
.. code-block:: console
cd <base path>/src/kayobe-config
source kayobe-env
Bootstrap the control host:
@ -235,13 +327,19 @@ Configure the overcloud host:
kayobe overcloud host configure
The previous command is likely to reboot the machine to disable SELinux. SSH
again when it has booted, activate the Kayobe environment and complete the host
configuration:
After this command has run, some files in the kayobe-config repository will
have changed. Kayobe performs static allocation of IP addresses, and tracks
them in ``etc/kayobe/network-allocation.yml``. Normally there may be changes to
this file, but in this case we manually added the IP address of ``controller0``
earlier. Kayobe uses tools provided by Kolla Ansible to generate passwords, and
stores them in ``etc/kayobe/kolla/passwords.yml``. It is important to track
changes to this file.
.. code-block:: console
kayobe overcloud host configure
cd <base path>/src/kayobe-config
git add etc/kayobe/kolla/passwords.yml
git commit -m "Add autogenerated passwords for Kolla Ansible"
Pull overcloud container images:
@ -255,13 +353,8 @@ Deploy overcloud services:
kayobe overcloud service deploy
There is an issue with Docker where it changes the default policy of the
``FORWARD`` chain to ``DROP``. This prevents traffic traversing the bridge.
Revert this change:
.. code-block:: console
sudo iptables -P FORWARD ACCEPT
Testing
=======
The ``init-runonce`` script provided by Kolla Ansible (not for production) can
be used to setup some resources for testing. This includes:
@ -281,18 +374,23 @@ pool range containing free IP addresses:
pip install python-openstackclient
export EXT_NET_CIDR=192.168.33.0/24
export EXT_NET_GATEWAY=192.168.33.1
export EXT_NET_GATEWAY=192.168.33.3
export EXT_NET_RANGE="start=192.168.33.4,end=192.168.33.254"
source "${KOLLA_CONFIG_PATH:-/etc/kolla}/admin-openrc.sh"
${KOLLA_SOURCE_PATH}/tools/init-runonce
Create a server instance, assign a floating IP address, and check that it is
accessible. The floating IP address is displayed after it is created, in this
example it is ``192.168.33.4``:
accessible.
.. code-block:: console
openstack server create --image cirros --flavor m1.tiny --key-name mykey --network demo-net demo1
openstack floating ip create public1
The floating IP address is displayed after it is created, in this example it is
``192.168.33.4``:
.. code-block:: console
openstack server add floating ip demo1 192.168.33.4
ssh cirros@192.168.33.4