[docs] Update release notes for Hunter 8.1
[fuel.git] / docs / release / installation / installation.instruction.rst
index b279630..bdb52d8 100644 (file)
 .. http://creativecommons.org/licenses/by/4.0
 .. (c) Open Platform for NFV Project, Inc. and its contributors
 
 .. http://creativecommons.org/licenses/by/4.0
 .. (c) Open Platform for NFV Project, Inc. and its contributors
 
-========
+***********************************
+OPNFV Fuel Installation Instruction
+***********************************
+
 Abstract
 ========
 
 Abstract
 ========
 
-This document describes how to install the Euphrates release of
+This document describes how to install the ``Hunter`` release of
 OPNFV when using Fuel as a deployment tool, covering its usage,
 limitations, dependencies and required system resources.
 OPNFV when using Fuel as a deployment tool, covering its usage,
 limitations, dependencies and required system resources.
-This is an unified documentation for both x86_64 and aarch64
+
+This is an unified documentation for both ``x86_64`` and ``aarch64``
 architectures. All information is common for both architectures
 except when explicitly stated.
 
 architectures. All information is common for both architectures
 except when explicitly stated.
 
-============
 Introduction
 ============
 
 This document provides guidelines on how to install and
 Introduction
 ============
 
 This document provides guidelines on how to install and
-configure the Euphrates release of OPNFV when using Fuel as a
+configure the ``Hunter`` release of OPNFV when using Fuel as a
 deployment tool, including required software and hardware configurations.
 
 Although the available installation options provide a high degree of
 freedom in how the system is set up, including architecture, services
 and features, etc., said permutations may not provide an OPNFV
 compliant reference architecture. This document provides a
 deployment tool, including required software and hardware configurations.
 
 Although the available installation options provide a high degree of
 freedom in how the system is set up, including architecture, services
 and features, etc., said permutations may not provide an OPNFV
 compliant reference architecture. This document provides a
-step-by-step guide that results in an OPNFV Euphrates compliant
+step-by-step guide that results in an OPNFV ``Hunter`` compliant
 deployment.
 
 The audience of this document is assumed to have good knowledge of
 networking and Unix/Linux administration.
 
 deployment.
 
 The audience of this document is assumed to have good knowledge of
 networking and Unix/Linux administration.
 
-=======
-Preface
-=======
-
-Before starting the installation of the Euphrates release of
+Before starting the installation of the ``Hunter`` release of
 OPNFV, using Fuel as a deployment tool, some planning must be
 done.
 
 Preparations
 ============
 
 OPNFV, using Fuel as a deployment tool, some planning must be
 done.
 
 Preparations
 ============
 
-Prior to installation, a number of deployment specific parameters must be collected, those are:
+Prior to installation, a number of deployment specific parameters must be
+collected, those are:
 
 #.     Provider sub-net and gateway information
 
 
 #.     Provider sub-net and gateway information
 
-#.     Provider VLAN information
-
-#.     Provider DNS addresses
-
-#.     Provider NTP addresses
+#.     Provider ``VLAN`` information
 
 
-#.     Network overlay you plan to deploy (VLAN, VXLAN, FLAT)
+#.     Provider ``DNS`` addresses
 
 
-#.     How many nodes and what roles you want to deploy (Controllers, Storage, Computes)
-
-#.     Monitoring options you want to deploy (Ceilometer, Syslog, etc.).
-
-#.     Other options not covered in the document are available in the links above
+#.     Provider ``NTP`` addresses
 
 
+#.     How many nodes and what roles you want to deploy (Controllers, Computes)
 
 This information will be needed for the configuration procedures
 provided in this document.
 
 
 This information will be needed for the configuration procedures
 provided in this document.
 
-=========================================
-Hardware Requirements for Virtual Deploys
-=========================================
-
-The following minimum hardware requirements must be met for the virtual
-installation of Euphrates using Fuel:
-
-+----------------------------+--------------------------------------------------------+
-| **HW Aspect**              | **Requirement**                                        |
-|                            |                                                        |
-+============================+========================================================+
-| **1 Jumpserver**           | A physical node (also called Foundation Node) that     |
-|                            | will host a Salt Master VM and each of the VM nodes in |
-|                            | the virtual deploy                                     |
-+----------------------------+--------------------------------------------------------+
-| **CPU**                    | Minimum 1 socket with Virtualization support           |
-+----------------------------+--------------------------------------------------------+
-| **RAM**                    | Minimum 32GB/server (Depending on VNF work load)       |
-+----------------------------+--------------------------------------------------------+
-| **Disk**                   | Minimum 100GB (SSD or SCSI (15krpm) highly recommended |
-+----------------------------+--------------------------------------------------------+
-
-
-===========================================
-Hardware Requirements for Baremetal Deploys
-===========================================
-
-The following minimum hardware requirements must be met for the baremetal
-installation of Euphrates using Fuel:
-
-+-------------------------+------------------------------------------------------+
-| **HW Aspect**           | **Requirement**                                      |
-|                         |                                                      |
-+=========================+======================================================+
-| **# of nodes**          | Minimum 5                                            |
-|                         |                                                      |
-|                         | - 3 KVM servers which will run all the controller    |
-|                         |   services                                           |
-|                         |                                                      |
-|                         | - 2 Compute nodes                                    |
-|                         |                                                      |
-+-------------------------+------------------------------------------------------+
-| **CPU**                 | Minimum 1 socket with Virtualization support         |
-+-------------------------+------------------------------------------------------+
-| **RAM**                 | Minimum 16GB/server (Depending on VNF work load)     |
-+-------------------------+------------------------------------------------------+
-| **Disk**                | Minimum 256GB 10kRPM spinning disks                  |
-+-------------------------+------------------------------------------------------+
-| **Networks**            | 4 VLANs (PUBLIC, MGMT, STORAGE, PRIVATE) - can be    |
-|                         | a mix of tagged/native                               |
-|                         |                                                      |
-|                         | 1 Un-Tagged VLAN for PXE Boot - ADMIN Network        |
-|                         |                                                      |
-|                         | Note: These can be allocated to a single NIC -       |
-|                         | or spread out over multiple NICs                     |
-+-------------------------+------------------------------------------------------+
-| **1 Jumpserver**        | A physical node (also called Foundation Node) that   |
-|                         | hosts the Salt Master and MaaS VMs                   |
-+-------------------------+------------------------------------------------------+
-| **Power management**    | All targets need to have power management tools that |
-|                         | allow rebooting the hardware and setting the boot    |
-|                         | order (e.g. IPMI)                                    |
-+-------------------------+------------------------------------------------------+
-
-**NOTE:** All nodes including the Jumpserver must have the same architecture (either x86_64 or aarch64).
-
-**NOTE:** For aarch64 deployments an UEFI compatible firmware with PXE support is needed (e.g. EDK2).
-
-
-===============================
+Hardware Requirements
+=====================
+
+Mininum hardware requirements depend on the deployment type.
+
+.. WARNING::
+
+    If ``baremetal`` nodes are present in the cluster, the architecture of the
+    nodes running the control plane (``kvm01``, ``kvm02``, ``kvm03`` for
+    ``HA`` scenarios, respectively ``ctl01``, ``gtw01``, ``odl01`` for
+    ``noHA`` scenarios) and the ``jumpserver`` architecture must be the same
+    (either ``x86_64`` or ``aarch64``).
+
+.. TIP::
+
+    The compute nodes may have different architectures, but extra
+    configuration might be required for scheduling VMs on the appropiate host.
+    This use-case is not tested in OPNFV CI, so it is considered experimental.
+
+Hardware Requirements for ``virtual`` Deploys
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+The following minimum hardware requirements must be met for the ``virtual``
+installation of ``Hunter`` using Fuel:
+
++------------------+------------------------------------------------------+
+| **HW Aspect**    | **Requirement**                                      |
+|                  |                                                      |
++==================+======================================================+
+| **1 Jumpserver** | A physical node (also called Foundation Node) that   |
+|                  | will host a Salt Master container and each of the VM |
+|                  | nodes in the virtual deploy                          |
++------------------+------------------------------------------------------+
+| **CPU**          | Minimum 1 socket with Virtualization support         |
++------------------+------------------------------------------------------+
+| **RAM**          | Minimum 32GB/server (Depending on VNF work load)     |
++------------------+------------------------------------------------------+
+| **Disk**         | Minimum 100GB (SSD or 15krpm SCSI highly recommended)|
++------------------+------------------------------------------------------+
+
+Hardware Requirements for ``baremetal`` Deploys
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+The following minimum hardware requirements must be met for the ``baremetal``
+installation of ``Hunter`` using Fuel:
+
++------------------+------------------------------------------------------+
+| **HW Aspect**    | **Requirement**                                      |
+|                  |                                                      |
++==================+======================================================+
+| **1 Jumpserver** | A physical node (also called Foundation Node) that   |
+|                  | hosts the Salt Master and MaaS containers            |
++------------------+------------------------------------------------------+
+| **# of nodes**   | Minimum 5                                            |
+|                  |                                                      |
+|                  | - 3 KVM servers which will run all the controller    |
+|                  |   services                                           |
+|                  |                                                      |
+|                  | - 2 Compute nodes                                    |
+|                  |                                                      |
+|                  | .. WARNING::                                         |
+|                  |                                                      |
+|                  |     ``kvm01``, ``kvm02``, ``kvm03`` nodes and the    |
+|                  |     ``jumpserver`` must have the same architecture   |
+|                  |     (either ``x86_64`` or ``aarch64``).              |
+|                  |                                                      |
+|                  | .. NOTE::                                            |
+|                  |                                                      |
+|                  |     ``aarch64`` nodes should run an ``UEFI``         |
+|                  |     compatible firmware with PXE support             |
+|                  |     (e.g. ``EDK2``).                                 |
++------------------+------------------------------------------------------+
+| **CPU**          | Minimum 1 socket with Virtualization support         |
++------------------+------------------------------------------------------+
+| **RAM**          | Minimum 16GB/server (Depending on VNF work load)     |
++------------------+------------------------------------------------------+
+| **Disk**         | Minimum 256GB 10kRPM spinning disks                  |
++------------------+------------------------------------------------------+
+| **Networks**     | Mininum 4                                            |
+|                  |                                                      |
+|                  | - 3 VLANs (``public``, ``mgmt``, ``private``) -      |
+|                  |   can be a mix of tagged/native                      |
+|                  |                                                      |
+|                  | - 1 Un-Tagged VLAN for PXE Boot -                    |
+|                  |   ``PXE/admin`` Network                              |
+|                  |                                                      |
+|                  | .. NOTE::                                            |
+|                  |                                                      |
+|                  |     These can be allocated to a single NIC           |
+|                  |     or spread out over multiple NICs.                |
+|                  |                                                      |
+|                  | .. WARNING::                                         |
+|                  |                                                      |
+|                  |     No external ``DHCP`` server should be present    |
+|                  |     in the ``PXE/admin`` network segment, as it      |
+|                  |     would interfere with ``MaaS`` ``DHCP`` during    |
+|                  |     ``baremetal`` node commissioning/deploying.      |
++------------------+------------------------------------------------------+
+| **Power mgmt**   | All targets need to have power management tools that |
+|                  | allow rebooting the hardware (e.g. ``IPMI``).        |
++------------------+------------------------------------------------------+
+
+Hardware Requirements for ``hybrid`` (``baremetal`` + ``virtual``) Deploys
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+The following minimum hardware requirements must be met for the ``hybrid``
+installation of ``Hunter`` using Fuel:
+
++------------------+------------------------------------------------------+
+| **HW Aspect**    | **Requirement**                                      |
+|                  |                                                      |
++==================+======================================================+
+| **1 Jumpserver** | A physical node (also called Foundation Node) that   |
+|                  | hosts the Salt Master and MaaS containers, and       |
+|                  | each of the virtual nodes defined in ``PDF``         |
++------------------+------------------------------------------------------+
+| **# of nodes**   | .. NOTE::                                            |
+|                  |                                                      |
+|                  |     Depends on ``PDF`` configuration.                |
+|                  |                                                      |
+|                  | If the control plane is virtualized, minimum         |
+|                  | baremetal requirements are:                          |
+|                  |                                                      |
+|                  | - 2 Compute nodes                                    |
+|                  |                                                      |
+|                  | If the computes are virtualized, minimum             |
+|                  | baremetal requirements are:                          |
+|                  |                                                      |
+|                  | - 3 KVM servers which will run all the controller    |
+|                  |   services                                           |
+|                  |                                                      |
+|                  | .. WARNING::                                         |
+|                  |                                                      |
+|                  |     ``kvm01``, ``kvm02``, ``kvm03`` nodes and the    |
+|                  |     ``jumpserver`` must have the same architecture   |
+|                  |     (either ``x86_64`` or ``aarch64``).              |
+|                  |                                                      |
+|                  | .. NOTE::                                            |
+|                  |                                                      |
+|                  |     ``aarch64`` nodes should run an ``UEFI``         |
+|                  |     compatible firmware with PXE support             |
+|                  |     (e.g. ``EDK2``).                                 |
++------------------+------------------------------------------------------+
+| **CPU**          | Minimum 1 socket with Virtualization support         |
++------------------+------------------------------------------------------+
+| **RAM**          | Minimum 16GB/server (Depending on VNF work load)     |
++------------------+------------------------------------------------------+
+| **Disk**         | Minimum 256GB 10kRPM spinning disks                  |
++------------------+------------------------------------------------------+
+| **Networks**     | Same as for ``baremetal`` deployments                |
++------------------+------------------------------------------------------+
+| **Power mgmt**   | Same as for ``baremetal`` deployments                |
++------------------+------------------------------------------------------+
+
 Help with Hardware Requirements
 Help with Hardware Requirements
-===============================
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
 Calculate hardware requirements:
 
 
 Calculate hardware requirements:
 
-For information on compatible hardware types available for use, please see `Fuel OpenStack Hardware Compatibility List <https://www.mirantis.com/software/hardware-compatibility/>`_.
-
 When choosing the hardware on which you will deploy your OpenStack
 environment, you should think about:
 
 When choosing the hardware on which you will deploy your OpenStack
 environment, you should think about:
 
-- CPU -- Consider the number of virtual machines that you plan to deploy in your cloud environment and the CPUs per virtual machine.
+- CPU -- Consider the number of virtual machines that you plan to deploy in
+  your cloud environment and the CPUs per virtual machine.
 
 
-- Memory -- Depends on the amount of RAM assigned per virtual machine and the controller node.
+- Memory -- Depends on the amount of RAM assigned per virtual machine and the
+  controller node.
 
 
-- Storage -- Depends on the local drive space per virtual machine, remote volumes that can be attached to a virtual machine, and object storage.
+- Storage -- Depends on the local drive space per virtual machine, remote
+  volumes that can be attached to a virtual machine, and object storage.
 
 
-- Networking -- Depends on the Choose Network Topology, the network bandwidth per virtual machine, and network storage.
+- Networking -- Depends on the Choose Network Topology, the network bandwidth
+  per virtual machine, and network storage.
 
 
-================================================
-Top of the Rack (TOR) Configuration Requirements
-================================================
+Top of the Rack (``TOR``) Configuration Requirements
+====================================================
 
 The switching infrastructure provides connectivity for the OPNFV
 infrastructure operations, tenant networks (East/West) and provider
 connectivity (North/South); it also provides needed connectivity for
 the Storage Area Network (SAN).
 
 The switching infrastructure provides connectivity for the OPNFV
 infrastructure operations, tenant networks (East/West) and provider
 connectivity (North/South); it also provides needed connectivity for
 the Storage Area Network (SAN).
+
 To avoid traffic congestion, it is strongly suggested that three
 physically separated networks are used, that is: 1 physical network
 for administration and control, one physical network for tenant private
 and public networks, and one physical network for SAN.
 To avoid traffic congestion, it is strongly suggested that three
 physically separated networks are used, that is: 1 physical network
 for administration and control, one physical network for tenant private
 and public networks, and one physical network for SAN.
+
 The switching connectivity can (but does not need to) be fully redundant,
 in such case it comprises a redundant 10GE switch pair for each of the
 three physically separated networks.
 
 The switching connectivity can (but does not need to) be fully redundant,
 in such case it comprises a redundant 10GE switch pair for each of the
 three physically separated networks.
 
-The physical TOR switches are **not** automatically configured from
-the Fuel OPNFV reference platform. All the networks involved in the OPNFV
-infrastructure as well as the provider networks and the private tenant
-VLANs needs to be manually configured.
+.. WARNING::
 
 
-Manual configuration of the Euphrates hardware platform should
-be carried out according to the `OPNFV Pharos Specification
-<https://wiki.opnfv.org/display/pharos/Pharos+Specification>`_.
+    The physical ``TOR`` switches are **not** automatically configured from
+    the OPNFV Fuel reference platform. All the networks involved in the OPNFV
+    infrastructure as well as the provider networks and the private tenant
+    VLANs needs to be manually configured.
+
+Manual configuration of the ``Hunter`` hardware platform should
+be carried out according to the `OPNFV Pharos Specification`_.
 
 
-============================
 OPNFV Software Prerequisites
 ============================
 
 OPNFV Software Prerequisites
 ============================
 
+.. NOTE::
+
+    All prerequisites described in this chapter apply to the ``jumpserver``
+    node.
+
+OS Distribution Support
+~~~~~~~~~~~~~~~~~~~~~~~
+
 The Jumpserver node should be pre-provisioned with an operating system,
 The Jumpserver node should be pre-provisioned with an operating system,
-according to the Pharos specification. Relevant network bridges should
-also be pre-configured (e.g. admin, management, public).
+according to the `OPNFV Pharos specification`_.
 
 
-Fuel@OPNFV has been validated by CI using the following distributions
+OPNFV Fuel has been validated by CI using the following distributions
 installed on the Jumpserver:
 
 installed on the Jumpserver:
 
-   - CentOS 7 (recommended by Pharos specification);
-   - Ubuntu Xenial;
+- ``CentOS 7`` (recommended by Pharos specification);
+- ``Ubuntu Xenial 16.04``;
 
 
-**NOTE:** The install script expects 'libvirt' to be installed and running
-on the Jumpserver. In case the packages are missing, the script will install
-them; but depending on the OS distribution, the user might have to start the
-'libvirtd' service manually.
+.. TOPIC:: ``aarch64`` notes
 
 
-==========================================
-OPNFV Software Installation and Deployment
-==========================================
+    For an ``aarch64`` Jumpserver, the ``libvirt`` minimum required
+    version is ``3.x``, ``3.5`` or newer highly recommended.
 
 
-This section describes the process of installing all the components needed to
-deploy the full OPNFV reference platform stack across a server cluster.
+    .. TIP::
 
 
-The installation is done with Mirantis Cloud Platform (MCP), which is based on
-a reclass model. This model provides the formula inputs to Salt, to make the deploy
-automatic based on deployment scenario.
-The reclass model covers:
+        ``CentOS 7`` (``aarch64``) distro provided packages are already new
+        enough.
 
 
-   - Infrastucture node definition: Salt Master node (cfg01) and MaaS node (mas01)
-   - Openstack node defition: Controler nodes (ctl01, ctl02, ctl03) and Compute nodes (cmp001, cmp002)
-   - Infrastructure components to install (software packages, services etc.)
-   - Openstack components and services (rabbitmq, galera etc.), as well as all configuration for them
+    .. WARNING::
 
 
+        ``Ubuntu 16.04`` (``arm64``), distro packages are too old and 3rd party
+        repositories should be used.
 
 
-Automatic Installation of a Virtual POD
-=======================================
+    For convenience, Armband provides a DEB repository holding all the
+    required packages.
 
 
-For virtual deploys all the targets are VMs on the Jumpserver. The deploy script will:
+    To add and enable the Armband repository on an Ubuntu 16.04 system,
+    create a new sources list file ``/apt/sources.list.d/armband.list``
+    with the following contents:
 
 
-   - Create a Salt Master VM on the Jumpserver which will drive the installation
-   - Create the bridges for networking with virsh (only if a real bridge does not already exists for a given network)
-   - Install Openstack on the targets
-      - Leverage Salt to install & configure Openstack services
+    .. code-block:: console
 
 
-.. figure:: img/fuel_virtual.png
-   :align: center
-   :alt: Fuel@OPNFV Virtual POD Network Layout Examples
+        jenkins@jumpserver:~$ cat /etc/apt/sources.list.d/armband.list
+        deb http://linux.enea.com/mcp-repos/rocky/xenial rocky-armband main
 
 
-   Fuel@OPNFV Virtual POD Network Layout Examples
+        jenkins@jumpserver:~$ sudo apt-key adv --keyserver keys.gnupg.net \
+                                               --recv 798AB1D1
+        jenkins@jumpserver:~$ sudo apt-get update
 
 
-   +-----------------------+------------------------------------------------------------------------+
-   | cfg01                 | Salt Master VM                                                         |
-   +-----------------------+------------------------------------------------------------------------+
-   | ctl01                 | Controller VM                                                          |
-   +-----------------------+------------------------------------------------------------------------+
-   | cmp01/cmp02           | Compute VMs                                                            |
-   +-----------------------+------------------------------------------------------------------------+
-   | gtw01                 | Gateway VM with neutron services (dhcp agent, L3 agent, metadata, etc) |
-   +-----------------------+------------------------------------------------------------------------+
-   | odl01                 | VM on which ODL runs (for scenarios deployed with ODL)                 |
-   +-----------------------+------------------------------------------------------------------------+
+OS Distribution Packages
+~~~~~~~~~~~~~~~~~~~~~~~~
 
 
+By default, the ``deploy.sh`` script will automatically install the required
+distribution package dependencies on the Jumpserver, so the end user does
+not have to manually install them before starting the deployment.
 
 
-In this figure there are examples of two virtual deploys:
-   - Jumphost 1 has only virsh bridges, created by the deploy script
-   - Jumphost 2 has a mix of linux and virsh briges; when linux bridge exist for a specified network,
-     the deploy script will skip creating a virsh bridge for it
+This includes Python, QEMU, libvirt etc.
 
 
-**Note**: A virtual network "mcpcontrol" is always created. For virtual deploys, "mcpcontrol" is also used
-for PXE, leaving the PXE bridge unused.
+.. SEEALSO::
 
 
+    To disable automatic package installation (and/or upgrade) during
+    deployment, check out the ``-P`` deploy argument.
 
 
-Automatic Installation of a Baremetal POD
-=========================================
+.. WARNING::
 
 
-The baremetal installation process can be done by editing the information about
-hardware and enviroment in the reclass files, or by using a Pod Descriptor File (PDF).
-This file contains all the information about the hardware and network of the deployment
-the will be fed to the reclass model during deployment.
+    The install script expects ``libvirt`` to be already running on the
+    Jumpserver.
 
 
-The installation is done automatically with the deploy script, which will:
+In case ``libvirt`` packages are missing, the script will install them; but
+depending on the OS distribution, the user might have to start the
+``libvirt`` daemon service manually, then run the deploy script again.
 
 
-   - Create a Salt Master VM on the Jumpserver which will drive the installation
-   - Create a MaaS Node VM on the Jumpserver which will provision the targets
-   - Install Openstack on the targets
-      - Leverage MaaS to provision baremetal nodes with the operating system
-      - Leverage Salt to configure the operatign system on the baremetal nodes
-      - Leverage Salt to install & configure Openstack services
+Therefore, it is recommended to install ``libvirt`` explicitly on the
+Jumpserver before the deployment.
 
 
-.. figure:: img/fuel_baremetal.png
-   :align: center
-   :alt: Fuel@OPNFV Baremetal POD Network Layout Example
-
-   Fuel@OPNFV Baremetal POD Network Layout Example
-
-   +-----------------------+---------------------------------------------------------+
-   | cfg01                 | Salt Master VM                                          |
-   +-----------------------+---------------------------------------------------------+
-   | mas01                 | MaaS Node VM                                            |
-   +-----------------------+---------------------------------------------------------+
-   | kvm01..03             | Baremetals which hold the VMs with controller functions |
-   +-----------------------+---------------------------------------------------------+
-   | cmp001/cmp002         | Baremetal compute nodes                                 |
-   +-----------------------+---------------------------------------------------------+
-   | prx01/prx02           | Proxy VMs for Nginx                                     |
-   +-----------------------+---------------------------------------------------------+
-   | msg01..03             | RabbitMQ Service VMs                                    |
-   +-----------------------+---------------------------------------------------------+
-   | dbs01..03             | MySQL service VMs                                       |
-   +-----------------------+---------------------------------------------------------+
-   | mdb01..03             | Telemetry VMs                                           |
-   +-----------------------+---------------------------------------------------------+
-   | odl01                 | VM on which ODL runs (for scenarios deployed with ODL)  |
-   +-----------------------+---------------------------------------------------------+
-   | Tenant VM             | VM running in the cloud                                 |
-   +-----------------------+---------------------------------------------------------+
-
-In the baremetal deploy all bridges but "mcpcontrol" are linux bridges. For the Jumpserver, if they are already created
-they will be used; otherwise they will be created. For the targets, the bridges are created by the deploy script.
-
-**Note**: A virtual network "mcpcontrol" is always created. For baremetal deploys, PXE bridge is used for
-baremetal node provisioning, while "mcpcontrol" is used to provision the infrastructure VMs only.
+While not mandatory, upgrading the kernel on the Jumpserver is also highly
+recommended.
 
 
+.. code-block:: console
 
 
-Steps to Start the Automatic Deploy
-===================================
+    jenkins@jumpserver:~$ sudo apt-get install \
+                          linux-image-generic-hwe-16.04-edge libvirt-bin
+    jenkins@jumpserver:~$ sudo reboot
 
 
-These steps are common both for virtual and baremetal deploys.
+User Requirements
+~~~~~~~~~~~~~~~~~
 
 
-#. Clone the Fuel code from gerrit
+The user running the deploy script on the Jumpserver should belong to
+``sudo`` and ``libvirt`` groups, and have passwordless sudo access.
 
 
-   For x86_64
+.. NOTE::
 
 
-   .. code-block:: bash
+    Throughout this documentation, we will use the ``jenkins`` username for
+    this role.
 
 
-       $ git clone https://git.opnfv.org/fuel
-       $ cd fuel
+The following example adds the groups to the user ``jenkins``:
 
 
-   For aarch64
+.. code-block:: console
 
 
-   .. code-block:: bash
+    jenkins@jumpserver:~$ sudo usermod -aG sudo jenkins
+    jenkins@jumpserver:~$ sudo usermod -aG libvirt jenkins
+    jenkins@jumpserver:~$ sudo reboot
+    jenkins@jumpserver:~$ groups
+    jenkins sudo libvirt
 
 
-       $ git clone https://git.opnfv.org/armband
-       $ cd armband
+    jenkins@jumpserver:~$ sudo visudo
+    ...
+    %jenkins ALL=(ALL) NOPASSWD:ALL
 
 
-#. Checkout the Euphrates release
+Local Artifact Storage
+~~~~~~~~~~~~~~~~~~~~~~
 
 
-   .. code-block:: bash
+The folder containing the temporary deploy artifacts (``/home/jenkins/tmpdir``
+in the examples below) needs to have mask ``777`` in order for ``libvirt`` to
+be able to use them.
 
 
-       $ git checkout opnfv-5.0.2
+.. code-block:: console
 
 
-#. Start the deploy script
+    jenkins@jumpserver:~$ mkdir -p -m 777 /home/jenkins/tmpdir
 
 
-   .. code-block:: bash
+Network Configuration
+~~~~~~~~~~~~~~~~~~~~~
 
 
-       $ ci/deploy.sh -l <lab_name> \
-                      -p <pod_name> \
-                      -b <URI to configuration repo containing the PDF file> \
-                      -s <scenario> \
-                      -B <list of admin, management, private and public bridges>
+Relevant Linux bridges should also be pre-configured for certain networks,
+depending on the type of the deployment.
 
 
-Examples
---------
-#. Virtual deploy
++------------+---------------+----------------------------------------------+
+| Network    | Linux Bridge  | Linux Bridge necessity based on deploy type  |
+|            |               +--------------+---------------+---------------+
+|            |               | ``virtual``  | ``baremetal`` | ``hybrid``    |
++============+===============+==============+===============+===============+
+| PXE/admin  | ``admin_br``  | absent       | present       | present       |
++------------+---------------+--------------+---------------+---------------+
+| management | ``mgmt_br``   | optional     | optional,     | optional,     |
+|            |               |              | recommended,  | recommended,  |
+|            |               |              | required for  | required for  |
+|            |               |              | ``functest``, | ``functest``, |
+|            |               |              | ``yardstick`` | ``yardstick`` |
++------------+---------------+--------------+---------------+---------------+
+| internal   | ``int_br``    | optional     | optional      | present       |
++------------+---------------+--------------+---------------+---------------+
+| public     | ``public_br`` | optional     | optional,     | optional,     |
+|            |               |              | recommended,  | recommended,  |
+|            |               |              | useful for    | useful for    |
+|            |               |              | debugging     | debugging     |
++------------+---------------+--------------+---------------+---------------+
 
 
-   To start a virtual deployment, it is required to have the `virtual` keyword
-   while specifying the pod name to the installer script.
+.. TIP::
 
 
-   It will create the required bridges and networks, configure Salt Master and
-   install OpenStack.
+    IP addresses should be assigned to the created bridge interfaces (not
+    to one of its ports).
 
 
-      .. code-block:: bash
+.. WARNING::
 
 
-         $ ci/deploy.sh -b file:///home/jenkins/tmpdir/securedlab \
-                        -l ericsson \
-                        -p virtual_kvm \
-                        -s os-nosdn-nofeature-noha
+    ``PXE/admin`` bridge (``admin_br``) **must** have an IP address.
 
 
-   Once the deployment is complete, the OpenStack Dashboard, Horizon is
-   available at http://<controller VIP>:8078, e.g. http://10.16.0.101:8078.
-   The administrator credentials are **admin** / **opnfv_secret**.
+Changes ``deploy.sh`` Will Perform to Jumpserver OS
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
 
-#. Baremetal deploy
+.. WARNING::
 
 
-   A x86 deploy on pod2 from Linux Foundation lab
+    The install script will alter Jumpserver sysconf and disable
+    ``net.bridge.bridge-nf-call``.
 
 
-      .. code-block:: bash
+.. WARNING::
 
 
-          $ ci/deploy.sh -b file:///home/jenkins/tmpdir/securedlab \
-                         -l lf \
-                         -p pod2 \
-                         -s os-nosdn-nofeature-ha \
-                         -B pxebr,br-ctl
+    On Jumpservers running Ubuntu with AppArmor enabled, when deploying
+    on baremetal nodes (i.e. when MaaS is used), the install script
+    will disable certain conflicting AppArmor profiles that interfere with
+    MaaS services inside the container, e.g. ``ntpd``, ``named``, ``dhcpd``,
+    ``tcpdump``.
 
 
-      .. figure:: img/lf_pod2.png
-         :align: center
-         :alt: Fuel@OPNFV LF POD2 Network Layout
+.. WARNING::
 
 
-         Fuel@OPNFV LF POD2 Network Layout
+    The install script will automatically install and/or upgrade the
+    required distribution package dependencies on the Jumpserver,
+    unless explicitly asked not to (via the ``-P`` deploy arg).
 
 
-   Once the deployment is complete, the SaltStack Deployment Documentation is
-   available at http://<Proxy VIP>:8090, e.g. http://172.30.10.103:8090.
+OPNFV Software Configuration (``XDF``)
+======================================
 
 
-   An aarch64 deploy on pod5 from Arm lab
+.. versionadded:: 5.0.0
+.. versionchanged:: 7.0.0
 
 
-      .. code-block:: bash
+Unlike the old approach based on OpenStack Fuel, OPNFV Fuel no longer has a
+graphical user interface for configuring the environment, but instead
+switched to OPNFV specific descriptor files that we will call generically
+``XDF``:
 
 
-         $ ci/deploy.sh -b file:///home/jenkins/tmpdir/securedlab \
-                        -l arm \
-                        -p pod5 \
-                        -s os-nosdn-nofeature-ha \
-                        -B admin7_br0,mgmt7_br0,,public7_br0
+- ``PDF`` (POD Descriptor File) provides an abstraction of the target POD
+  with all its hardware characteristics and required parameters;
+- ``IDF`` (Installer Descriptor File) extends the ``PDF`` with POD related
+  parameters required by the OPNFV Fuel installer;
+- ``SDF`` (Scenario Descriptor File, **not** yet adopted) will later
+  replace embedded scenario definitions, describing the roles and layout of
+  the cluster enviroment for a given reference architecture;
 
 
-      .. figure:: img/arm_pod5.png
-         :align: center
-         :alt: Fuel@OPNFV ARM POD5 Network Layout
+.. TIP::
 
 
-         Fuel@OPNFV ARM POD5 Network Layout
+    For ``virtual`` deployments, if the ``public`` network will be accessed
+    from outside the ``jumpserver`` node, a custom ``PDF``/``IDF`` pair is
+    required for customizing ``idf.net_config.public`` and
+    ``idf.fuel.jumphost.bridges.public``.
 
 
-   Once the deployment is complete, the SaltStack Deployment Documentation is
-   available at http://<Proxy VIP>:8090, e.g. http://10.0.8.103:8090.
+.. NOTE::
 
 
+    For OPNFV CI PODs, as well as simple (no ``public`` bridge) ``virtual``
+    deployments, ``PDF``/``IDF`` files are already available in the
+    `pharos git repo`_. They can be used as a reference for user-supplied
+    inputs or to kick off a deployment right away.
 
 
-Pod Descriptor Files
-====================
++----------+------------------------------------------------------------------+
+| LAB/POD  | ``PDF``/``IDF`` availability based on deploy type                |
+|          +------------------------+--------------------+--------------------+
+|          | ``virtual``            | ``baremetal``      | ``hybrid``         |
++==========+========================+====================+====================+
+| OPNFV CI | available in           | available in       | N/A, as currently  |
+| POD      | `pharos git repo`_     | `pharos git repo`_ | there are 0 hybrid |
+|          | (e.g.                  | (e.g. ``lf-pod2``, | PODs in OPNFV CI   |
+|          | ``ericsson-virtual1``) | ``arm-pod5``)      |                    |
++----------+------------------------+--------------------+--------------------+
+| local or | ``user-supplied``      | ``user-supplied``  | ``user-supplied``  |
+| new POD  |                        |                    |                    |
++----------+------------------------+--------------------+--------------------+
+
+.. TIP::
+
+    Both ``PDF`` and ``IDF`` structure are modelled as ``yaml`` schemas in the
+    `pharos git repo`_, also included as a git submodule in OPNFV Fuel.
+
+    .. SEEALSO::
 
 
-Descriptor files provide the installer with an abstraction of the target pod
-with all its hardware characteristics and required parameters. This information
-is split into two different files:
-Pod Descriptor File (PDF) and Installer Descriptor File (IDF).
+        - ``mcp/scripts/pharos/config/pdf/pod1.schema.yaml``
+        - ``mcp/scripts/pharos/config/pdf/idf-pod1.schema.yaml``
 
 
+    Schema files are also used during the initial deployment phase to validate
+    the user-supplied input ``PDF``/``IDF`` files.
 
 
-The Pod Descriptor File is a hardware and network description of the pod
-infrastructure. The information is modeled under a yaml structure.
-A reference file with the expected yaml structure is available at
-*mcp/config/labs/local/pod1.yaml*
+``PDF``
+~~~~~~~
 
 
-A common network section describes all the internal and provider networks
-assigned to the pod. Each network is expected to have a vlan tag, IP subnet and
-attached interface on the boards. Untagged vlans shall be defined as "native".
+The Pod Descriptor File is a hardware description of the POD
+infrastructure. The information is modeled under a ``yaml`` structure.
 
 
-The hardware description is arranged into a main "jumphost" node and a "nodes"
-set for all target boards. For each node the following characteristics
+The hardware description covers the ``jumphost`` node and a set of ``nodes``
+for the cluster target boards. For each node the following characteristics
 are defined:
 
 are defined:
 
-- Node parameters including CPU features and total memory.
-- A list of available disks.
-- Remote management parameters.
-- Network interfaces list including mac address, speed and advanced features.
-- IP list of fixed IPs for the node
+- Node parameters including ``CPU`` features and total memory;
+- A list of available disks;
+- Remote management parameters;
+- Network interfaces list including name, ``MAC`` address, link speed,
+  advanced features;
+
+.. SEEALSO::
+
+    A reference file with the expected ``yaml`` structure is available at:
+
+    - ``mcp/scripts/pharos/config/pdf/pod1.yaml``
+
+    For more information on ``PDF``, see the `OPNFV PDF Wiki Page`_.
+
+.. WARNING::
+
+    The fixed IPs defined in ``PDF`` are ignored by the OPNFV Fuel installer
+    script and it will instead assign addresses based on the network ranges
+    defined in ``IDF``.
 
 
-**Note**: the fixed IPs are ignored by the MCP installer script and it will instead
-assign based on the network ranges defined under the pod network configuration.
+    For more details on the way IP addresses are assigned, see
+    :ref:`OPNFV Fuel User Guide <fuel-userguide>`.
 
 
+``PDF``/``IDF`` Role (hostname) Mapping
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
 
-The Installer Descriptor File extends the PDF with pod related parameters
+Upcoming ``SDF`` support will introduce a series of possible node roles.
+Until that happens, the role mapping logic is hardcoded, based on node index
+in ``PDF``/``IDF`` (which should also be in sync, i.e. the parameters of the
+``n``-th cluster node defined in ``PDF`` should be the ``n``-th node in
+``IDF`` structures too).
+
++-------------+------------------+----------------------+
+| Node index  | ``HA`` scenario  | ``noHA`` scenario    |
++=============+==================+======================+
+| 1st         | ``kvm01``        | ``ctl01``            |
++-------------+------------------+----------------------+
+| 2nd         | ``kvm02``        | ``gtw01``            |
++-------------+------------------+----------------------+
+| 3rd         | ``kvm03``        | ``odl01``/``unused`` |
++-------------+------------------+----------------------+
+| 4th,        | ``cmp001``,      | ``cmp001``,          |
+| 5th,        | ``cmp002``,      | ``cmp002``,          |
+| ...         | ``...``          | ``...``              |
++-------------+------------------+----------------------+
+
+.. TIP::
+
+    To switch node role(s), simply reorder the node definitions in
+    ``PDF``/``IDF`` (make sure to keep them in sync).
+
+``IDF``
+~~~~~~~
+
+The Installer Descriptor File extends the ``PDF`` with POD related parameters
 required by the installer. This information may differ per each installer type
 required by the installer. This information may differ per each installer type
-and it is not considered part of the pod infrastructure. Fuel installer relies
-on the IDF model to map the networks to the bridges on the foundation node and
-to setup all node NICs by defining the expected OS device name and bus address.
+and it is not considered part of the POD infrastructure.
+
+``idf.*`` Overview
+------------------
+
+The ``IDF`` file must be named after the ``PDF`` it attaches to, with the
+prefix ``idf-``.
+
+.. SEEALSO::
+
+    A reference file with the expected ``yaml`` structure is available at:
+
+    - ``mcp/scripts/pharos/config/pdf/idf-pod1.yaml``
+
+The file follows a ``yaml`` structure and at least two sections
+(``idf.net_config`` and ``idf.fuel``) are expected.
+
+The ``idf.fuel`` section defines several sub-sections required by the OPNFV
+Fuel installer:
+
+- ``jumphost``: List of bridge names for each network on the Jumpserver;
+- ``network``: List of device name and bus address info of all the target nodes.
+  The order must be aligned with the order defined in the ``PDF`` file.
+  The OPNFV Fuel installer relies on the ``IDF`` model to setup all node NICs
+  by defining the expected device name and bus address;
+- ``maas``: Defines the target nodes commission timeout and deploy timeout;
+- ``reclass``: Defines compute parameter tuning, including huge pages, ``CPU``
+  pinning and other ``DPDK`` settings;
+
+.. code-block:: yaml
+
+    ---
+    idf:
+      version: 0.1     # fixed, the only supported version (mandatory)
+      net_config:      # POD network configuration overview (mandatory)
+        oob: ...       # mandatory
+        admin: ...     # mandatory
+        mgmt: ...      # mandatory
+        storage: ...   # mandatory
+        private: ...   # mandatory
+        public: ...    # mandatory
+      fuel:            # OPNFV Fuel specific section (mandatory)
+        jumphost:      # OPNFV Fuel jumpserver bridge configuration (mandatory)
+          bridges:                          # Bridge name mapping (mandatory)
+            admin: 'admin_br'               # <PXE/admin bridge name> or ~
+            mgmt: 'mgmt_br'                 # <mgmt bridge name> or ~
+            private: ~                      # <private bridge name> or ~
+            public: 'public_br'             # <public bridge name> or ~
+          trunks: ...                       # Trunked networks (optional)
+        maas:                               # MaaS timeouts (optional)
+          timeout_comissioning: 10          # commissioning timeout in minutes
+          timeout_deploying: 15             # deploy timeout in minutes
+        network:                            # Cluster nodes network (mandatory)
+          interface_mtu: 1500               # Cluster-level MTU (optional)
+          ntp_strata_host1: 1.pool.ntp.org  # NTP1 (optional)
+          ntp_strata_host2: 0.pool.ntp.org  # NTP2 (optional)
+          node: ...                         # List of per-node cfg (mandatory)
+        reclass:                            # Additional params (mandatory)
+          node: ...                         # List of per-node cfg (mandatory)
+
+``idf.net_config``
+------------------
+
+``idf.net_config`` was introduced as a mechanism to map all the usual cluster
+networks (internal and provider networks, e.g. ``mgmt``) to their ``VLAN``
+tags, ``CIDR`` and a physical interface index (used to match networks to
+interface names, like ``eth0``, on the cluster nodes).
+
+
+.. WARNING::
+
+    The mapping between one network segment (e.g. ``mgmt``) and its ``CIDR``/
+    ``VLAN`` is not configurable on a per-node basis, but instead applies to
+    all the nodes in the cluster.
+
+For each network, the following parameters are currently supported:
+
++--------------------------+--------------------------------------------------+
+| ``idf.net_config.*`` key | Details                                          |
++==========================+==================================================+
+| ``interface``            | The index of the interface to use for this net.  |
+|                          | For each cluster node (if network is present),   |
+|                          | OPNFV Fuel will determine the underlying physical|
+|                          | interface by picking the element at index        |
+|                          | ``interface`` from the list of network interface |
+|                          | names defined in                                 |
+|                          | ``idf.fuel.network.node.*.interfaces``.          |
+|                          | Required for each network.                       |
+|                          |                                                  |
+|                          | .. NOTE::                                        |
+|                          |                                                  |
+|                          |     The interface index should be the            |
+|                          |     same on all cluster nodes. This can be       |
+|                          |     achieved by ordering them accordingly in     |
+|                          |     ``PDF``/``IDF``.                             |
++--------------------------+--------------------------------------------------+
+| ``vlan``                 | ``VLAN`` tag (integer) or the string ``native``. |
+|                          | Required for each network.                       |
++--------------------------+--------------------------------------------------+
+| ``ip-range``             | When specified, all cluster IPs dynamically      |
+|                          | allocated by OPNFV Fuel for that network will be |
+|                          | assigned inside this range.                      |
+|                          | Required for ``oob``, optional for others.       |
+|                          |                                                  |
+|                          | .. NOTE::                                        |
+|                          |                                                  |
+|                          |     For now, only range start address is used.   |
++--------------------------+--------------------------------------------------+
+| ``network``              | Network segment address.                         |
+|                          | Required for each network, except ``oob``.       |
++--------------------------+--------------------------------------------------+
+| ``mask``                 | Network segment mask.                            |
+|                          | Required for each network, except ``oob``.       |
++--------------------------+--------------------------------------------------+
+| ``gateway``              | Gateway IP address.                              |
+|                          | Required for ``public``, N/A for others.         |
++--------------------------+--------------------------------------------------+
+| ``dns``                  | List of DNS IP addresses.                        |
+|                          | Required for ``public``, N/A for others.         |
++--------------------------+--------------------------------------------------+
+
+Sample ``public`` network configuration block:
+
+.. code-block:: yaml
+
+    idf:
+        net_config:
+            public:
+              interface: 1
+              vlan: native
+              network: 10.0.16.0
+              ip-range: 10.0.16.100-10.0.16.253
+              mask: 24
+              gateway: 10.0.16.254
+              dns:
+                - 8.8.8.8
+                - 8.8.4.4
+
+.. TOPIC:: ``hybrid`` POD notes
+
+    Interface indexes must be the same for all nodes, which is problematic
+    when mixing ``virtual`` nodes (where all interfaces were untagged
+    so far) with ``baremetal`` nodes (where interfaces usually carry
+    tagged VLANs).
+
+    .. TIP::
+
+        To achieve this, a special ``jumpserver`` network layout is used:
+        ``mgmt``, ``storage``, ``private``, ``public`` are trunked together
+        in a single ``trunk`` bridge:
+
+        - without decapsulating them (if they are also tagged on ``baremetal``);
+          a ``trunk.<vlan_tag>`` interface should be created on the
+          ``jumpserver`` for each tagged VLAN so the kernel won't drop the
+          packets;
+        - by decapsulating them  first (if they are also untagged on
+          ``baremetal`` nodes);
+
+    The ``trunk`` bridge is then used for all bridges OPNFV Fuel
+    is aware of in ``idf.fuel.jumphost.bridges``, e.g. for a ``trunk`` where
+    only ``mgmt`` network is not decapsulated:
+
+    .. code-block:: yaml
+
+        idf:
+            fuel:
+              jumphost:
+                bridges:
+                  admin: 'admin_br'
+                  mgmt: 'trunk'
+                  private: 'trunk'
+                  public: 'trunk'
+                trunks:
+                  # mgmt network is not decapsulated for jumpserver infra nodes,
+                  # to align with the VLAN configuration of baremetal nodes.
+                  mgmt: True
+
+.. WARNING::
+
+    The Linux kernel limits the name of network interfaces to 16 characters.
+    Extra care is required when choosing bridge names, so appending the
+    ``VLAN`` tag won't lead to an interface name length exceeding that limit.
+
+``idf.fuel.network``
+--------------------
+
+``idf.fuel.network`` allows mapping the cluster networks (e.g. ``mgmt``) to
+their physical interface name (e.g. ``eth0``) and bus address on the cluster
+nodes.
+
+``idf.fuel.network.node`` should be a list with the same number (and order) of
+elements as the cluster nodes defined in ``PDF``, e.g. the second cluster node
+in ``PDF`` will use the interface name and bus address defined in the second
+list element.
+
+Below is a sample configuration block for a single node with two interfaces:
+
+.. code-block:: yaml
+
+    idf:
+      fuel:
+        network:
+          node:
+            # Ordered-list, index should be in sync with node index in PDF
+            - interfaces:
+                # Ordered-list, index should be in sync with interface index
+                # in PDF
+                - 'ens3'
+                - 'ens4'
+              busaddr:
+                # Bus-info reported by `ethtool -i ethX`
+                - '0000:00:03.0'
+                - '0000:00:04.0'
+
+
+``idf.fuel.reclass``
+--------------------
+
+``idf.fuel.reclass`` provides a way of overriding default values in the
+reclass cluster model.
+
+This currently covers strictly compute parameter tuning, including huge
+pages, ``CPU`` pinning and other ``DPDK`` settings.
+
+``idf.fuel.reclass.node`` should be a list with the same number (and order) of
+elements as the cluster nodes defined in ``PDF``, e.g. the second cluster node
+in ``PDF`` will use the parameters defined in the second list element.
+
+The following parameters are currently supported:
+
++---------------------------------+-------------------------------------------+
+| ``idf.fuel.reclass.node.*``     | Details                                   |
+| key                             |                                           |
++=================================+===========================================+
+| ``nova_cpu_pinning``            | List of CPU cores nova will be pinned to. |
+|                                 |                                           |
+|                                 | .. NOTE::                                 |
+|                                 |                                           |
+|                                 |     Currently disabled.                   |
++---------------------------------+-------------------------------------------+
+| ``compute_hugepages_size``      | Size of each persistent huge pages.       |
+|                                 |                                           |
+|                                 | Usual values are ``2M`` and ``1G``.       |
++---------------------------------+-------------------------------------------+
+| ``compute_hugepages_count``     | Total number of persistent huge pages.    |
++---------------------------------+-------------------------------------------+
+| ``compute_hugepages_mount``     | Mount point to use for huge pages.        |
++---------------------------------+-------------------------------------------+
+| ``compute_kernel_isolcpu``      | List of certain CPU cores that are        |
+|                                 | isolated from Linux scheduler.            |
++---------------------------------+-------------------------------------------+
+| ``compute_dpdk_driver``         | Kernel module to provide userspace I/O    |
+|                                 | support.                                  |
++---------------------------------+-------------------------------------------+
+| ``compute_ovs_pmd_cpu_mask``    | Hexadecimal mask of CPUs to run ``DPDK``  |
+|                                 | Poll-mode drivers.                        |
++---------------------------------+-------------------------------------------+
+| ``compute_ovs_dpdk_socket_mem`` | Set of amount huge pages in ``MB`` to be  |
+|                                 | used by ``OVS-DPDK`` daemon taken for each|
+|                                 | ``NUMA`` node. Set size is equal to       |
+|                                 | ``NUMA`` nodes count, elements are        |
+|                                 | divided by comma.                         |
++---------------------------------+-------------------------------------------+
+| ``compute_ovs_dpdk_lcore_mask`` | Hexadecimal mask of ``DPDK`` lcore        |
+|                                 | parameter used to run ``DPDK`` processes. |
++---------------------------------+-------------------------------------------+
+| ``compute_ovs_memory_channels`` | Number of memory channels to be used.     |
++---------------------------------+-------------------------------------------+
+| ``dpdk0_driver``                | NIC driver to use for physical network    |
+|                                 | interface.                                |
++---------------------------------+-------------------------------------------+
+| ``dpdk0_n_rxq``                 | Number of ``RX`` queues.                  |
++---------------------------------+-------------------------------------------+
+
+Sample ``compute_params`` configuration block (for a single node):
+
+.. code-block:: yaml
+
+    idf:
+      fuel:
+        reclass:
+          node:
+            - compute_params:
+                common: &compute_params_common
+                  compute_hugepages_size: 2M
+                  compute_hugepages_count: 2048
+                  compute_hugepages_mount: /mnt/hugepages_2M
+                dpdk:
+                  <<: *compute_params_common
+                  compute_dpdk_driver: uio
+                  compute_ovs_pmd_cpu_mask: "0x6"
+                  compute_ovs_dpdk_socket_mem: "1024"
+                  compute_ovs_dpdk_lcore_mask: "0x8"
+                  compute_ovs_memory_channels: "2"
+                  dpdk0_driver: igb_uio
+                  dpdk0_n_rxq: 2
+
+``SDF``
+~~~~~~~
+
+Scenario Descriptor Files are not yet implemented in the OPNFV Fuel ``Hunter``
+release.
+
+Instead, embedded OPNFV Fuel scenarios files are locally available in
+``mcp/config/scenario``.
 
 
+OPNFV Software Installation and Deployment
+==========================================
 
 
-The file follows a yaml structure and a "fuel" section is expected. Contents and
-references must be aligned with the PDF file. The IDF file must be named after
-the PDF with the prefix "idf-". A reference file with the expected structure
-is available at *mcp/config/labs/local/idf-pod1.yaml*
+This section describes the process of installing all the components needed to
+deploy the full OPNFV reference platform stack across a server cluster.
 
 
+Deployment Types
+~~~~~~~~~~~~~~~~
 
 
-=============
-Release Notes
-=============
+.. WARNING::
 
 
-Please refer to the :ref:`Release Notes <fuel-release-notes-label>` article.
+    OPNFV releases previous to ``Hunter`` used to rely on the ``virtual``
+    keyword being part of the POD name (e.g. ``ericsson-virtual2``) to
+    configure the deployment type as ``virtual``. Otherwise ``baremetal``
+    was implied.
 
 
-==========
-References
-==========
+``Gambia`` and newer releases are more flexbile towards supporting a mix
+of ``baremetal`` and ``virtual`` nodes, so the type of deployment is
+now automatically determined based on the cluster nodes types in ``PDF``:
+
++---------------------------------+-------------------------------------------+
+| ``PDF`` has nodes of type       | Deployment type                           |
++---------------+-----------------+                                           |
+| ``baremetal`` | ``virtual``     |                                           |
++===============+=================+===========================================+
+| yes           | no              | ``baremetal``                             |
++---------------+-----------------+-------------------------------------------+
+| yes           | yes             | ``hybrid``                                |
++---------------+-----------------+-------------------------------------------+
+| no            | yes             | ``virtual``                               |
++---------------+-----------------+-------------------------------------------+
+
+Based on that, the deployment script will later enable/disable certain extra
+nodes (e.g. ``mas01``) and/or ``STATE`` files (e.g. ``maas``).
+
+``HA`` vs ``noHA``
+~~~~~~~~~~~~~~~~~~
 
 
-OPNFV
+High availability of OpenStack services is determined based on scenario name,
+e.g. ``os-nosdn-nofeature-noha`` vs ``os-nosdn-nofeature-ha``.
 
 
-1) `OPNFV Home Page <http://www.opnfv.org>`_
-2) `OPNFV documentation <http://docs.opnfv.org>`_
-3) `Software downloads <https://www.opnfv.org/software/download>`_
+.. TIP::
 
 
-OpenStack
+    ``HA`` scenarios imply a virtualized control plane (``VCP``) for the
+    OpenStack services running on the 3 ``kvm`` nodes.
 
 
-4) `OpenStack Ocata Release Artifacts <http://www.openstack.org/software/ocata>`_
-5) `OpenStack Documentation <http://docs.openstack.org>`_
+    .. SEEALSO::
 
 
-OpenDaylight
+        An experimental feature argument (``-N``) is supported by the deploy
+        script for disabling ``VCP``, although it might not be supported by
+        all scenarios and is not being continuosly validated by OPNFV CI/CD.
 
 
-6) `OpenDaylight Artifacts <http://www.opendaylight.org/software/downloads>`_
+.. WARNING::
 
 
-Fuel
+    ``virtual`` ``HA`` deployments are not officially supported, due to
+    poor performance and various limitations of nested virtualization on
+    both ``x86_64`` and ``aarch64`` architectures.
 
 
-7) `Mirantis Cloud Platform Documentation <https://docs.mirantis.com/mcp/latest>`_
+    .. TIP::
 
 
-Salt
+        ``virtual`` ``HA`` deployments without ``VCP`` are supported, but
+        highly experimental.
 
 
-8) `Saltstack Documentation <https://docs.saltstack.com/en/latest/topics>`_
-9) `Saltstack Formulas <http://salt-formulas.readthedocs.io/en/latest/develop/overview-reclass.html>`_
++-------------------------------+-------------------------+-------------------+
+| Feature                       | ``HA`` scenario         | ``noHA`` scenario |
++===============================+=========================+===================+
+| ``VCP``                       | yes,                    | no                |
+| (Virtualized Control Plane)   | disabled with ``-N``    |                   |
++-------------------------------+-------------------------+-------------------+
+| OpenStack APIs SSL            | yes                     | no                |
++-------------------------------+-------------------------+-------------------+
+| Storage                       | ``GlusterFS``           | ``NFS``           |
++-------------------------------+-------------------------+-------------------+
 
 
-Reclass
+Steps to Start the Automatic Deploy
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+These steps are common for ``virtual``, ``baremetal`` or ``hybrid`` deploys,
+``x86_64``, ``aarch64`` or ``mixed`` (``x86_64`` and ``aarch64``):
+
+- Clone the OPNFV Fuel code from gerrit
+- Checkout the ``Hunter`` release tag
+- Start the deploy script
+
+.. NOTE::
+
+    The deployment uses the OPNFV Pharos project as input (``PDF`` and
+    ``IDF`` files) for hardware and network configuration of all current
+    OPNFV PODs.
+
+    When deploying a new POD, one may pass the ``-b`` flag to the deploy
+    script to override the path for the labconfig directory structure
+    containing the ``PDF`` and ``IDF`` (``<URI to configuration repo ...>`` is
+    the absolute path to a local or remote directory structure, populated
+    similar to `pharos git repo`_, i.e. ``PDF``/``IDF`` reside in a
+    subdirectory called ``labs/<lab_name>``).
+
+.. code-block:: console
+
+    jenkins@jumpserver:~$ git clone https://git.opnfv.org/fuel
+    jenkins@jumpserver:~$ cd fuel
+    jenkins@jumpserver:~/fuel$ git checkout opnfv-8.1.0
+    jenkins@jumpserver:~/fuel$ ci/deploy.sh -l <lab_name> \
+                                            -p <pod_name> \
+                                            -b <URI to configuration repo containing the PDF/IDF files> \
+                                            -s <scenario> \
+                                            -D \
+                                            -S <Storage directory for deploy artifacts> |& tee deploy.log
+
+.. TIP::
+
+    Besides the basic options,  there are other recommended deploy arguments:
+
+    - use ``-D`` option to enable the debug info
+    - use ``-S`` option to point to a tmp dir where the disk images are saved.
+      The deploy artifacts will be re-used on subsequent (re)deployments.
+    - use ``|& tee`` to save the deploy log to a file
+
+Typical Cluster Examples
+~~~~~~~~~~~~~~~~~~~~~~~~
+
+Common cluster layouts usually fall into one of the cases described below,
+categorized by deployment type (``baremetal``, ``virtual`` or ``hybrid``) and
+high availability (``HA`` or ``noHA``).
+
+A simplified overview of the steps ``deploy.sh`` will automatically perform is:
+
+- create a Salt Master Docker container on the jumpserver, which will drive
+  the rest of the installation;
+- ``baremetal`` or ``hybrid`` only: create a ``MaaS`` container node,
+  which will be leveraged using Salt to handle OS provisioning on the
+  ``baremetal`` nodes;
+- leverage Salt to install & configure OpenStack;
+
+.. NOTE::
+
+    A Docker network ``mcpcontrol`` is always created for initial connection
+    of the infrastructure containers (``cfg01``, ``mas01``) on Jumphost.
+
+.. WARNING::
+
+    A single cluster deployment per ``jumpserver`` node is currently supported,
+    indifferent of its type (``virtual``, ``baremetal`` or ``hybrid``).
+
+Once the deployment is complete, the following should be accessible:
+
++---------------+----------------------------------+---------------------------+
+| Resource      | ``HA`` scenario                  | ``noHA`` scenario         |
++===============+==================================+===========================+
+| ``Horizon``   | ``https://<prx public VIP>``     | ``http://<ctl VIP>:8078`` |
+| (Openstack    |                                  |                           |
+| Dashboard)    |                                  |                           |
++---------------+----------------------------------+---------------------------+
+| ``SaltStack`` | ``http://<prx public VIP>:8090`` | N/A                       |
+| Deployment    |                                  |                           |
+| Documentation |                                  |                           |
++---------------+----------------------------------+---------------------------+
+
+.. SEEALSO::
+
+    For more details on locating and importing the generated SSL certificate,
+    see :ref:`OPNFV Fuel User Guide <fuel-userguide>`.
+
+``virtual`` ``noHA`` POD
+------------------------
+
+In the following figure there are two generic examples of ``virtual`` deploys,
+each on a separate Jumphost node, both behind the same ``TOR`` switch:
+
+- Jumphost 1 has only virsh bridges (created by the deploy script);
+- Jumphost 2 has a mix of Linux (manually created) and ``libvirt`` managed
+  bridges (created by the deploy script);
+
+.. figure:: img/fuel_virtual_noha.png
+   :align: center
+   :width: 60%
+   :alt: OPNFV Fuel Virtual noHA POD Network Layout Examples
+
+   OPNFV Fuel Virtual noHA POD Network Layout Examples
+
+   +-------------+------------------------------------------------------------+
+   | ``cfg01``   | Salt Master Docker container                               |
+   +-------------+------------------------------------------------------------+
+   | ``ctl01``   | Controller VM                                              |
+   +-------------+------------------------------------------------------------+
+   | ``gtw01``   | Gateway VM with neutron services                           |
+   |             | (``DHCP`` agent, ``L3`` agent, ``metadata`` agent etc)     |
+   +-------------+------------------------------------------------------------+
+   | ``odl01``   | VM on which ``ODL`` runs                                   |
+   |             | (for scenarios deployed with ODL)                          |
+   +-------------+------------------------------------------------------------+
+   | ``cmp001``, | Compute VMs                                                |
+   | ``cmp002``  |                                                            |
+   +-------------+------------------------------------------------------------+
+
+.. TIP::
+
+    If external access to the ``public`` network is not required, there is
+    little to no motivation to create a custom ``PDF``/``IDF`` set for a
+    virtual deployment.
+
+    Instead, the existing virtual PODs definitions in `pharos git repo`_ can
+    be used as-is:
+
+    - ``ericsson-virtual1`` for ``x86_64``;
+    - ``arm-virtual2`` for ``aarch64``;
+
+.. code-block:: console
+
+    # example deploy cmd for an x86_64 virtual cluster
+    jenkins@jumpserver:~/fuel$ ci/deploy.sh -l ericsson \
+                                            -p virtual1 \
+                                            -s os-nosdn-nofeature-noha \
+                                            -D \
+                                            -S /home/jenkins/tmpdir |& tee deploy.log
+
+``baremetal`` ``noHA`` POD
+--------------------------
+
+.. WARNING::
+
+    These scenarios are not tested in OPNFV CI, so they are considered
+    experimental.
+
+.. figure:: img/fuel_baremetal_noha.png
+   :align: center
+   :width: 60%
+   :alt: OPNFV Fuel Baremetal noHA POD Network Layout Example
+
+   OPNFV Fuel Baremetal noHA POD Network Layout Example
+
+   +-------------+------------------------------------------------------------+
+   | ``cfg01``   | Salt Master Docker container                               |
+   +-------------+------------------------------------------------------------+
+   | ``mas01``   | MaaS Node Docker container                                 |
+   +-------------+------------------------------------------------------------+
+   | ``ctl01``   | Baremetal controller node                                  |
+   +-------------+------------------------------------------------------------+
+   | ``gtw01``   | Baremetal Gateway with neutron services                    |
+   |             | (dhcp agent, L3 agent, metadata, etc)                      |
+   +-------------+------------------------------------------------------------+
+   | ``odl01``   | Baremetal node on which ODL runs                           |
+   |             | (for scenarios deployed with ODL, otherwise unused         |
+   +-------------+------------------------------------------------------------+
+   | ``cmp001``, | Baremetal Computes                                         |
+   | ``cmp002``  |                                                            |
+   +-------------+------------------------------------------------------------+
+   | Tenant VM   | VM running in the cloud                                    |
+   +-------------+------------------------------------------------------------+
+
+``baremetal`` ``HA`` POD
+------------------------
+
+.. figure:: img/fuel_baremetal_ha.png
+   :align: center
+   :width: 60%
+   :alt: OPNFV Fuel Baremetal HA POD Network Layout Example
+
+   OPNFV Fuel Baremetal HA POD Network Layout Example
+
+   +---------------------------+----------------------------------------------+
+   | ``cfg01``                 | Salt Master Docker container                 |
+   +---------------------------+----------------------------------------------+
+   | ``mas01``                 | MaaS Node Docker container                   |
+   +---------------------------+----------------------------------------------+
+   | ``kvm01``,                | Baremetals which hold the VMs with           |
+   | ``kvm02``,                | controller functions                         |
+   | ``kvm03``                 |                                              |
+   +---------------------------+----------------------------------------------+
+   | ``prx01``,                | Proxy VMs for Nginx                          |
+   | ``prx02``                 |                                              |
+   +---------------------------+----------------------------------------------+
+   | ``msg01``,                | RabbitMQ Service VMs                         |
+   | ``msg02``,                |                                              |
+   | ``msg03``                 |                                              |
+   +---------------------------+----------------------------------------------+
+   | ``dbs01``,                | MySQL service VMs                            |
+   | ``dbs02``,                |                                              |
+   | ``dbs03``                 |                                              |
+   +---------------------------+----------------------------------------------+
+   | ``mdb01``,                | Telemetry VMs                                |
+   | ``mdb02``,                |                                              |
+   | ``mdb03``                 |                                              |
+   +---------------------------+----------------------------------------------+
+   | ``odl01``                 | VM on which ``OpenDaylight`` runs            |
+   |                           | (for scenarios deployed with ``ODL``)        |
+   +---------------------------+----------------------------------------------+
+   | ``cmp001``,               | Baremetal Computes                           |
+   | ``cmp002``                |                                              |
+   +---------------------------+----------------------------------------------+
+   | Tenant VM                 | VM running in the cloud                      |
+   +---------------------------+----------------------------------------------+
+
+.. code-block:: console
+
+    # x86_x64 baremetal deploy on pod2 from Linux Foundation lab (lf-pod2)
+    jenkins@jumpserver:~/fuel$ ci/deploy.sh -l lf \
+                                            -p pod2 \
+                                            -s os-nosdn-nofeature-ha \
+                                            -D \
+                                            -S /home/jenkins/tmpdir |& tee deploy.log
+
+.. code-block:: console
+
+    # aarch64 baremetal deploy on pod5 from Enea ARM lab (arm-pod5)
+    jenkins@jumpserver:~/fuel$ ci/deploy.sh -l arm \
+                                            -p pod5 \
+                                            -s os-nosdn-nofeature-ha \
+                                            -D \
+                                            -S /home/jenkins/tmpdir |& tee deploy.log
+
+``hybrid`` ``noHA`` POD
+-----------------------
+
+.. figure:: img/fuel_hybrid_noha.png
+   :align: center
+   :width: 60%
+   :alt: OPNFV Fuel Hybrid noHA POD Network Layout Examples
+
+   OPNFV Fuel Hybrid noHA POD Network Layout Examples
+
+   +-------------+------------------------------------------------------------+
+   | ``cfg01``   | Salt Master Docker container                               |
+   +-------------+------------------------------------------------------------+
+   | ``mas01``   | MaaS Node Docker container                                 |
+   +-------------+------------------------------------------------------------+
+   | ``ctl01``   | Controller VM                                              |
+   +-------------+------------------------------------------------------------+
+   | ``gtw01``   | Gateway VM with neutron services                           |
+   |             | (``DHCP`` agent, ``L3`` agent, ``metadata`` agent etc)     |
+   +-------------+------------------------------------------------------------+
+   | ``odl01``   | VM on which ``ODL`` runs                                   |
+   |             | (for scenarios deployed with ODL)                          |
+   +-------------+------------------------------------------------------------+
+   | ``cmp001``, | Baremetal Computes                                         |
+   | ``cmp002``  |                                                            |
+   +-------------+------------------------------------------------------------+
+
+Automatic Deploy Breakdown
+~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+When an automatic deploy is started, the following operations are performed
+sequentially by the deploy script:
+
++------------------+----------------------------------------------------------+
+| **Deploy stage** | **Details**                                              |
++==================+==========================================================+
+| Argument         | enviroment variables and command line arguments passed   |
+| Parsing          | to ``deploy.sh`` are interpreted                         |
++------------------+----------------------------------------------------------+
+| Distribution     | Install and/or configure mandatory requirements on the   |
+| Package          | ``jumpserver`` node:                                     |
+| Installation     |                                                          |
+|                  | - ``Docker`` (from upstream and not distribution repos,  |
+|                  |   as the version included in ``Ubuntu`` ``Xenial`` is    |
+|                  |   outdated);                                             |
+|                  | - ``docker-compose`` (from upstream, as the version      |
+|                  |   included in both ``CentOS 7`` and                      |
+|                  |   ``Ubuntu Xenial 16.04`` has dependency issues on most  |
+|                  |   systems);                                              |
+|                  | - ``virt-inst`` (from upstream, as the version included  |
+|                  |   in ``Ubuntu Xenial 16.04`` is outdated and lacks       |
+|                  |   certain required features);                            |
+|                  | - other miscelaneous requirements, depending on          |
+|                  |   ``jumpserver`` distribution OS;                        |
+|                  |                                                          |
+|                  | .. SEEALSO::                                             |
+|                  |                                                          |
+|                  |     - ``mcp/scripts/requirements_deb.yaml`` (``Ubuntu``) |
+|                  |     - ``mcp/scripts/requirements_rpm.yaml`` (``CentOS``) |
+|                  |                                                          |
+|                  | .. WARNING::                                             |
+|                  |                                                          |
+|                  |     Mininum required ``Docker`` version is ``17.x``.     |
+|                  |                                                          |
+|                  | .. WARNING::                                             |
+|                  |                                                          |
+|                  |     Mininum required ``virt-inst`` version is ``1.4``.   |
++------------------+----------------------------------------------------------+
+| Patch            | For each ``git`` submodule in OPNFV Fuel repository,     |
+| Apply            | if a subdirectory with the same name exists under        |
+|                  | ``mcp/patches``, all patches in that subdirectory are    |
+|                  | applied using ``git-am`` to the respective ``git``       |
+|                  | submodule.                                               |
+|                  |                                                          |
+|                  | This allows OPNFV Fuel to alter upstream repositories    |
+|                  | contents before consuming them, including:               |
+|                  |                                                          |
+|                  | - ``Docker`` container build process customization;      |
+|                  | - ``salt-formulas`` customization;                       |
+|                  | - ``reclass.system`` customization;                      |
+|                  |                                                          |
+|                  | .. SEEALSO::                                             |
+|                  |                                                          |
+|                  |     - ``mcp/patches/README.rst``                         |
++------------------+----------------------------------------------------------+
+| SSH RSA Keypair  | If not already present, a RSA keypair is generated on    |
+| Generation       | the ``jumpserver`` node at:                              |
+|                  |                                                          |
+|                  | - ``/var/lib/opnfv/mcp.rsa{,.pub}``                      |
+|                  |                                                          |
+|                  | The public key will be added to the ``authorized_keys``  |
+|                  | list for ``ubuntu`` user, so the private key can be used |
+|                  | for key-based logins on:                                 |
+|                  |                                                          |
+|                  | - ``cfg01``, ``mas01`` infrastructure nodes;             |
+|                  | - all cluster nodes (``baremetal`` and/or ``virtual``),  |
+|                  |   including ``VCP`` VMs;                                 |
++------------------+----------------------------------------------------------+
+| ``j2``           | Based on ``XDF`` (``PDF``, ``IDF``, ``SDF``) and         |
+| Expansion        | additional deployment configuration determined during    |
+|                  | ``argument parsing`` stage described above, all jinja2   |
+|                  | templates are expanded, including:                       |
+|                  |                                                          |
+|                  | - various classes in ``reclass.cluster``;                |
+|                  | - docker-compose ``yaml`` for Salt Master bring-up;      |
+|                  | - ``libvirt`` network definitions (``xml``);             |
++------------------+----------------------------------------------------------+
+| Jumpserver       | Basic validation that common ``jumpserver`` requirements |
+| Requirements     | are satisfied, e.g. ``PXE/admin`` is Linux bridge if     |
+| Check            | ``baremetal`` nodes are defined in the ``PDF``.          |
++------------------+----------------------------------------------------------+
+| Infrastucture    | .. NOTE::                                                |
+| Setup            |                                                          |
+|                  |     All steps apply to and only to the ``jumpserver``.   |
+|                  |                                                          |
+|                  | - prepare virtual machines;                              |
+|                  | - (re)create ``libvirt`` managed networks;               |
+|                  | - apply ``sysctl`` configuration;                        |
+|                  | - apply ``udev`` configuration;                          |
+|                  | - create & start virtual machines prepared earlier;      |
+|                  | - create & start Salt Master (``cfg01``) Docker          |
+|                  |   container;                                             |
++------------------+----------------------------------------------------------+
+| ``STATE``        | Based on deployment type, scenario and other parameters, |
+| Files            | a ``STATE`` file list is constructed, then executed      |
+|                  | sequentially.                                            |
+|                  |                                                          |
+|                  | .. TIP::                                                 |
+|                  |                                                          |
+|                  |     The table below lists all current ``STATE`` files    |
+|                  |     and their intended action.                           |
+|                  |                                                          |
+|                  | .. SEEALSO::                                             |
+|                  |                                                          |
+|                  |     For more information on how the list of ``STATE``    |
+|                  |     files is constructed, see                            |
+|                  |     :ref:`OPNFV Fuel User Guide <fuel-userguide>`.       |
++------------------+----------------------------------------------------------+
+| Log              | Contents of ``/var/log`` are recursively gathered from   |
+| Collection       | all the nodes, then archived together for later          |
+|                  | inspection.                                              |
++------------------+----------------------------------------------------------+
+
+``STATE`` Files Overview
+------------------------
+
++---------------------------+-------------------------------------------------+
+| ``STATE`` file            | Targets involved and main intended action       |
++===========================+=================================================+
+| ``virtual_init``          | ``cfg01``: reclass node generation              |
+|                           |                                                 |
+|                           | ``jumpserver`` VMs (if present): basic OS       |
+|                           | config                                          |
++---------------------------+-------------------------------------------------+
+| ``maas``                  | ``mas01``: OS, MaaS configuration               |
+|                           | ``baremetal`` node commissioning and deploy     |
+|                           |                                                 |
+|                           | .. NOTE::                                       |
+|                           |                                                 |
+|                           |     Skipped if no ``baremetal`` nodes are       |
+|                           |     defined in ``PDF`` (``virtual`` deploy).    |
++---------------------------+-------------------------------------------------+
+| ``baremetal_init``        | ``kvm``, ``cmp``: OS install, config            |
++---------------------------+-------------------------------------------------+
+| ``dpdk``                  | ``cmp``: configure OVS-DPDK                     |
++---------------------------+-------------------------------------------------+
+| ``networks``              | ``ctl``: create OpenStack networks              |
++---------------------------+-------------------------------------------------+
+| ``neutron_gateway``       | ``gtw01``: configure Neutron gateway            |
++---------------------------+-------------------------------------------------+
+| ``opendaylight``          | ``odl01``: install & configure ``ODL``          |
++---------------------------+-------------------------------------------------+
+| ``openstack_noha``        | cluster nodes: install OpenStack without ``HA`` |
++---------------------------+-------------------------------------------------+
+| ``openstack_ha``          | cluster nodes: install OpenStack with ``HA``    |
++---------------------------+-------------------------------------------------+
+| ``virtual_control_plane`` | ``kvm``: create ``VCP`` VMs                     |
+|                           |                                                 |
+|                           | ``VCP`` VMs: basic OS config                    |
+|                           |                                                 |
+|                           | .. NOTE::                                       |
+|                           |                                                 |
+|                           |     Skipped if ``-N`` deploy argument is used.  |
++---------------------------+-------------------------------------------------+
+| ``tacker``                | ``ctl``: install & configure Tacker             |
++---------------------------+-------------------------------------------------+
+
+Release Notes
+=============
+
+Please refer to the :ref:`OPNFV Fuel Release Notes <fuel-releasenotes>`
+article.
+
+References
+==========
 
 
-10) `Reclass model <http://reclass.pantsfullofunix.net>`_
+For more information on the OPNFV ``Hunter`` 8.1 release, please see:
+
+#. `OPNFV Home Page`_
+#. `OPNFV Documentation`_
+#. `OPNFV Software Downloads`_
+#. `OPNFV Hunter Wiki Page`_
+#. `OpenStack Rocky Release Artifacts`_
+#. `OpenStack Documentation`_
+#. `OpenDaylight Artifacts`_
+#. `Mirantis Cloud Platform Documentation`_
+#. `Saltstack Documentation`_
+#. `Saltstack Formulas`_
+#. `Reclass`_
+
+.. FIXME: cleanup unused refs, extend above list
+.. _`OpenDaylight`: https://www.opendaylight.org
+.. _`OpenDaylight Artifacts`: https://www.opendaylight.org/software/downloads
+.. _`MCP`: https://www.mirantis.com/software/mcp/
+.. _`Mirantis Cloud Platform Documentation`: https://docs.mirantis.com/mcp/latest/
+.. _`fuel git repository`: https://git.opnfv.org/fuel
+.. _`pharos git repo`: https://git.opnfv.org/pharos
+.. _`OpenStack Documentation`: https://docs.openstack.org/rocky
+.. _`OpenStack Rocky Release Artifacts`: https://www.openstack.org/software/rocky
+.. _`OPNFV Home Page`: https://www.opnfv.org
+.. _`OPNFV Hunter Wiki Page`: https://wiki.opnfv.org/display/SWREL/Hunter
+.. _`OPNFV Documentation`: https://docs.opnfv.org
+.. _`OPNFV Software Downloads`: https://www.opnfv.org/software/downloads
+.. _`Apache License 2.0`: https://www.apache.org/licenses/LICENSE-2.0
+.. _`Saltstack Documentation`: https://docs.saltstack.com/en/latest/topics/
+.. _`Saltstack Formulas`: https://salt-formulas.readthedocs.io/en/latest/
+.. _`Reclass`: https://reclass.pantsfullofunix.net
+.. _`OPNFV Pharos Specification`: https://wiki.opnfv.org/display/pharos/Pharos+Specification
+.. _`OPNFV PDF Wiki Page`: https://wiki.opnfv.org/display/INF/POD+Descriptor