TripleO

Introduction

Supported operating systems: RHEL/CentOS >= 7, Fedora

TripleO means “OpenStack on OpenStack.” The Undercloud is first deployed onto a single node with the essential OpenStack services to handle baremetal deployments. That server is then used to create and manage a full production cloud called the Overcloud.

TripleO is a collection of many services. As part of the Transformation Squad’s goal, Undercloud services are being removed and/or refactored to provide a simpler deployment tool. These are the services that are used on the Undercloud [52]:

Service

Removed In

Replaced By

Description

Ansible

Used for deploying the Undercloud and Overcloud services.

Ceilometer

Stein

Collects information about the Overcloud nodes.

docker

Ussuri

Podman

Container runtime for OpenStack services.

Glance

Image management used by Ironic.

Gnocchi

Stein

A more efficient alternative to Ceilometer.

Heat

Heat parameters define the deployment settings.

Horizon

Stein

Web dashboard for deploying an Overcloud.

Ironic

Manages the bare-metal provisioning.

Keystone

Authentication of OpenStack services.

Kolla

Victoria

Ansible (TCIB)

Provides container images of OpenStack services.

MariaDB

Database for OpenStack services.

Mistral

Victoria

Ansible

Workflows are used to define and execute all of the deployment processes.

Neutron

Manages the Overcloud networks.

Nova

Manages the Overcloud nodes after provisioning.

Paunch

Victoria

Ansible

Container state management.

Podman

Container runtime for OpenStack services.

Puppet

Configuration management.

RabbitMQ

Messaging back-end for OpenStack services.

Swift

Storage for Heat deployment plans.

Zaqar

Victoria

Ansible

A messaging service used by Mistral.

In Pike, most of the Overcloud services are deployed as containers built by Kolla. The most notable service that lacked container support was Neutron due to it’s complexity. Starting in Queens, all of the Overcloud services are installed as containers. Support for also running the Undercloud services in containers was added as a technology preview in Queens and later became the default configuration for Rocky. Previously, instack-undercloud was used to setup and install the Undercloud services and now the same deployment method for the Overcloud is used for the Undercloud. [20]

Red Hat OpenStack Platform Releases

Red Hat provides most of the development to the core OpenStack services. The RPM Distribution of OpenStack (RDO) project is a community project lead by Red Hat to use the latest upstream code from OpenStack and package it to work and be distributable on Red Hat Enterprise Linux and Fedora based operating systems. [2]

The Red Hat OpenStack Platform (RHOSP) is a solution by Red Hat that takes the upstream OpenStack source code and makes it enterprise quality by hardening the security and increasing it’s stability. Upgrades from one major release of RHOSP to the next have been supported since RHOSP 8.

Release Cycle:

  • RHOSP < 10 = Each release is supported for up to 3 years.

  • RHOSP >= 10 = Starting with RHOSP 10, every third release of RHOSP is a long-life (LL) release with up to 5 years of support. In-between releases are supported for 1 year. Fast-forward upgrades are supported to upgrade directly from one LL release to the next (for example, 10 to 13).

  • RHOSP >= 16 = Every release of RHOSP is now a LL release. [43]

Releases:

  • RHOSP 3 (Grizzly)

    • Release: 2013-07-10

    • EOL: 2014-07-31

  • RHOSP 4 (Havana)

    • Release: 2013-12-19

    • EOL: 2015-06-19

  • RHOSP 5 (Icehouse)

    • Release: 2014-06-30

    • EOL: 2017-06-30

  • RHOSP 6 (Juno)

    • Release: 2015-02-09

    • EOL: 2018-02-17

  • RHOSP 7 (Kilo)

    • Release: 2015-08-05

    • EOL: 2018-08-05

  • RHOSP 8 (Liberty)

    • Release: 2016-04-20

    • EOL: 2019-04-20

  • RHOSP 9 (Mitaka)

    • Release: 2016-08-24

    • EOL: 2019-08-24

  • RHOSP 10 LL (Newton)

    • Release: 2016-12-15

    • EOL: 2021-12-15

  • RHOSP 11 (Ocata)

    • Release: 2017-05-18

    • EOL: 2018-05-18

  • RHOSP 12 (Pike)

    • Release: 2017-12-13

    • EOL: 2018-12-13

  • RHOSP 13 LL (Queens)

    • Release: 2018-06-27

    • EOL: 2023-06-27

  • RHOSP 14 (Rocky)

    • Release: 2019-01-10

    • EOL: 2020-01-10

  • RHOSP 15 (Stein)

    • Release: 2019-09-19

    • EOL: 2020-09-19

  • RHOSP 16 LL (Train)

    • Release: 2020-02-06

    • EOL: 2025-05-30

Starting with RHOSP 16, each minor release is tied to a specific version of RHEL. [1]

RHOSP

RHEL

16.0

8.1

16.1

8.2

16.2

8.4

RHOSP supports running a virtualized Undercloud on these platforms [3]:

  • Kernel-based Virtual Machine (QEMU with KVM acceleration)

  • Red Hat Virtualization (RHV)

  • Microsoft Hyper-V

  • VMware ESX and ESXi

RHOSP only supports using libvirt with KVM as the compute hypervisor’s virtualization technology. [28]

The version of RHOSP in use can be found on the Undercloud by viewing the “/etc/rhosp-release” file. OpenStack packages can also be tracked down to which major release it is a part of by using https://access.redhat.com/downloads/content/package-browser.

$ yum install rhosp-release
$ cat /etc/rhosp-release
Red Hat OpenStack Platform release 16.0.1 (Train)

Historical Milestones

Upstream

[57][58]

Downstream

[1]

Repositories

Upstream

The upstream TripleO project has three main repositories for each OpenStack release:

Name and Aliases

Testing Level

Use Case

General Availability (GA), Release, or Tested

High

Production

Testing, Test, or Buildlogs

Medium

Pre-production

Trunk, Current, Consistent, or Untested

Low

Development

If installing on RHEL, it is required to enable additional repositories:

  • RHEL 7 [40]:

    $ sudo subscription-manager repos --disable=*
    $ sudo subscription-manager repos --enable rhel-7-server-rpms --enable rhel-7-server-rh-common-rpms --enable rhel-7-server-extras-rpms
    
  • RHEL 8 [74]:

    $ sudo subscription-manager repos --disable=*
    $ sudo subscription-manager repos --enable=rhel-8-for-x86_64-baseos-eus-rpms --enable=rhel-8-for-x86_64-appstream-eus-rpms --enable=rhel-8-for-x86_64-highavailability-eus-rpms --enable=ansible-2-for-rhel-8-x86_64-rpms --enable=fast-datapath-for-rhel-8-x86_64-rpms
    

If installing on CentOS 8, it is required to enable both the high availability (for Pacemaker packages) and PowerTools (for Ruby packages) repositories.

$ sudo dnf config-manager --set-enabled HighAvailability
$ sudo dnf config-manager --set-enabled PowerTools
  • GA:

    • CentOS:

      $ sudo yum install centos-release-openstack-${OPENSTACK_RELEASE}
      
    • RHEL:

      $ sudo yum install https://repos.fedorapeople.org/repos/openstack/openstack-${OPENSTACK_RELEASE}/rdo-release-${OPENSTACK_RELEASE}-${RDO_RPM_RELEASE}.noarch.rpm
      
  • Testing

    • CentOS:

      $ sudo yum install centos-release-openstack-${OPENSTACK_RELEASE}
      $ sudo yum-config-manager --enable centos-openstack-${OPENSTACK_RELEASE}-test
      
    • RHEL:

      $ sudo yum install https://repos.fedorapeople.org/repos/openstack/openstack-${OPENSTACK_RELEASE}/rdo-release-${OPENSTACK_RELEASE}-${RDO_RPM_RELEASE}.noarch.rpm
      $ sudo yum-config-manager --enable openstack-${OPENSTACK_RELEASE}-testing
      
  • Trunk

    • Trunk builds are divided into different stages [54][65]:

      • current = The latest individually successfully built packages from every RDO and OpenStack project.

      • consistent = The current build passed the tripleo-ci promotion jobs.

      • current-tripleo = The consistent build passed phase 1 CI promotion jobs.

      • current-tripleo-rdo = The current-tripleo build passed phase 2 CI promotion jobs. This is also known as current-passed-ci because it has passed all of the available CI jobs.

    • RDO repository (current-tripleo-rdo):

      $ sudo yum install https://repos.fedorapeople.org/repos/openstack/openstack-${OPENSTACK_RELEASE}/rdo-release-${OPENSTACK_RELEASE}-${RDO_RPM_RELEASE}.noarch.rpm
      $ sudo yum-config-manager --enable rdo-trunk-${OPENSTACK_RELEASE}-tested
      
    • Or tripleo-repos [22]:

      • CentOS 7 (<= Train):

        $ sudo yum install "https://trunk.rdoproject.org/centos7/current-tripleo-rdo/$(curl -k https://trunk.rdoproject.org/centos7/current-tripleo-rdo/ | grep python2-tripleo-repos- | cut -d\" -f8)"
        
      • CentOS 8 (>= Train):

        $ sudo yum install "https://trunk.rdoproject.org/centos8-${OPENSTACK_RELEASE}/component/tripleo/current-tripleo-rdo/$(curl -k  https://trunk.rdoproject.org/centos8-${OPENSTACK_RELEASE}/component/tripleo/current-tripleo-rdo/ | grep "python3-tripleo-repos" | cut -d\" -f8)"
        
      • Setup the TripleO repositories:

        $ sudo tripleo-repos -b ${OPENSTACK_RELEASE} current-tripleo-rdo
        
    • Or manually:

      $ EL_VER="8"
      $ sudo curl -L -o /etc/yum.repos.d/delorean-${OPENSTACK_RELEASE}.repo https://trunk.rdoproject.org/centos${EL_VER}-${OPENSTACK_RELEASE}/current-tripleo-rdo/delorean.repo
      $ sudo curl -L -o /etc/yum.repos.d/delorean-deps-${OPENSTACK_RELEASE}.repo https://trunk.rdoproject.org/centos${EL_VER}-${OPENSTACK_RELEASE}/delorean-deps.repo
      
    • Create a container image prepare file that uses the current-tripleo (default), current-tripleo-rdo, or a Delorean hash tag. Container images are not built for current. Configure the undercloud.conf to use this file via the container_images_file parameter. Configure the Overcloud to use it by adding it as another Heat environment template: openstack overcloud deploy --templates -e ~/containers-prepare-parameters.yaml. [76]

      $ openstack tripleo container image prepare default --output-env-file ~/containers-prepare-parameters.yaml
      $ ${EDITOR} ~/containers-prepare-parameters.yaml
      
      ---
      parameter_defaults:
        ContainerImagePrepare:
           - set:
               namespace: docker.io/tripleo<OPENSTACK_RELEASE>
               tag: current-tripleo-rdo
               # Alternatively, to prevent unforeseen updates, the tag can be a Delorean hash.
               # For example, for
               # https://trunk.rdoproject.org/centos8-victoria/current-tripleo-rdo/7f/97/7f974e10d7184d5fc45445a3073333bd/
               # this tag can be used:
               #namespace: docker.io/tripleovictoria
               #tag: 7f974e10d7184d5fc45445a3073333bd
      

[53]

Downstream

It is recommended to disable any existing repositories to avoid package conflicts.

$ sudo subscription-manager repos --disable=*
  • RHOSP 10 [26]:

    $ sudo subscription-manager repos --enable=rhel-7-server-rpms --enable=rhel-7-server-extras-rpms --enable=rhel-7-server-rh-common-rpms --enable=rhel-ha-for-rhel-7-server-rpms --enable=rhel-7-server-nfv-rpms --enable=rhel-7-server-rhceph-2-tools-rpms --enable=rhel-7-server-rhceph-2-mon-rpms --enable=rhel-7-server-rhceph-2-osd-rpms --enable=rhel-7-server-openstack-10-rpms
    
  • RHOSP 13 [27]:

    $ sudo subscription-manager repos --enable=rhel-7-server-rpms --enable=rhel-7-server-extras-rpms --enable=rhel-7-server-rh-common-rpms --enable=rhel-ha-for-rhel-7-server-rpms --enable=rhel-7-server-nfv-rpms --enable=rhel-7-server-rhceph-3-tools-rpms --enable=rhel-7-server-rhceph-3-mon-rpms --enable=rhel-7-server-rhceph-3-osd-rpms --enable=rhel-7-server-openstack-13-rpms
    
  • RHOSP 16 [55]:

    $ sudo subscription-manager repos --enable=rhel-8-for-x86_64-baseos-rpms --enable=rhel-8-for-x86_64-appstream-rpms --enable=rhel-8-for-x86_64-highavailability-rpms --enable=ansible-2.8-for-rhel-8-x86_64-rpms --enable=openstack-16-for-rhel-8-x86_64-rpms --enable=fast-datapath-for-rhel-8-x86_64-rpms
    

Deployment (Quick)

Packstack

Supported operating system: RHEL/CentOS 7, Fedora

Packstack is part of Red Hat’s RDO project. It’s purpose is for providing small and simple demonstrations of OpenStack. This tool does not handle any upgrades of the OpenStack services.

Hardware requirements [9]:

  • 16GB RAM

Install

Disable NetworkManager. It is not compatible with Packstack.

$ sudo systemctl disable NetworkManager

Install the Packstack utility.

$ sudo yum -y install openstack-packstack

There are two network scenarios that Packstack can deploy. The default is to have an isolated network (1). Floating IPs will not be able to access the network on the public interface. For lab environments, Packstack can also configure Neutron to expose the network instead to allow instances with floating IPs to access other IP addresses on the network (2).

1. Isolated Network Install

Generate a configuration file referred to as the “answer” file. This can optionally be customized. Then install OpenStack using the answer file. By default, the network will be entirely isolated. [4]

$ sudo packstack --gen-answer-file <FILE>
$ sudo packstack --answer-file <FILE>

Packstack logs are stored in /var/tmp/packstack/. The administrator and demo user credentials will be saved to the user’s home directory.

$ source ~/keystonerc_admin
$ source ~/keystonerc_demo

Although the network will not be exposed by default, it can still be configured later. The primary interface to the lab’s network, typically eth0, will need to be configured as a Open vSwitch bridge to allow this. Be sure to replace the “IPADDR”, “PREFIX”, and “GATEWAY” with the server’s correct settings. Neutron will also need to be configured to allow “flat” networks.

File: /etc/sysconfig/network-scripts/ifcfg-eth0

DEVICE=eth0
ONBOOT=yes
DEVICETYPE=ovs
TYPE=OVSPort
OVS_BRIDGE=br-ex
BOOTPROTO=none
NM_CONTROLLED=no

File: /etc/sysconfig/network-scripts/ifcfg-br-ex

DEVICE=br-ex
ONBOOT=yes
DEVICETYPE=ovs
TYPE=OVSBridge
DEFROUTE=yes
IPADDR=192.168.1.200
PREFIX=24
GATEWAY=192.168.1.1
PEERDNS=no
BOOTPROTO=none
NM_CONTROLLED=no

2. Exposed Network Install

It is also possible to deploy OpenStack where Neutron can have access to the public network. Run the Packstack installation with the command below and replace “eth0” with the public interface name.

$ sudo packstack --allinone --provision-demo=n --os-neutron-ovs-bridge-mappings=extnet:br-ex --os-neutron-ovs-bridge-interfaces=br-ex:eth0 --os-neutron-ml2-type-drivers=vxlan,flat

Alternatively, use these configuration options in the answer file.

CONFIG_NEUTRON_ML2_TYPE_DRIVERS=vxlan,flat
CONFIG_NEUTRON_OVS_BRIDGE_MAPPINGS=extnet:br-ex
CONFIG_NEUTRON_OVS_BRIDGE_IFACES=br-ex:eth0
CONFIG_PROVISION_DEMO=n
$ sudo packstack --answer-file <ANSWER_FILE>

After the installation is finished, create the necessary network in Neutron as the admin user. In this example, the network will automatically allocate IP addresses between 192.168.1.201 and 192.168.1.254. The IP 192.168.1.1 is both the physical router and default gateway.

$ . keystonerc_admin
$ openstack network create --share --provider-physical-network physical_network --provider-network-type flat --router external external_network
$ openstack subnet create --subnet-range 192.168.1.0/24 --gateway 192.168.1.1 --network external_network --allocation-pool start=192.168.1.201,end=192.168.1.254 --no-dhcp public_subnet

The “external_network” can now be associated with a router in user accounts.

[5][90]

Answer File

The “answer” configuration file defines how OpenStack should be setup and installed. Using a answer file can provide a more customizable deployment.

Common options:

  • CONFIG_DEFAULT_PASSWORD = Any blank passwords in the answer file will be set to this value.

  • CONFIG_KEYSTONE_ADMIN_TOKEN = The administrator authentication token.

  • CONFIG_KEYSTONE_ADMIN_PW = The administrator password.

  • CONFIG_MARIADB_PW = The MariaDB root user’s password.

  • CONFIG_HORIZON_SSL = Configure an SSL for the Horizon dashboard. This requires that SSLs be generated manually and then defined in the configuration file [6]:

    $ for cert in selfcert ssl_dashboard ssl_vnc; do sudo openssl req -x509 -sha256 -newkey rsa:2048 -keyout /etc/pki/tls/private/${cert}.key -out /etc/pki/tls/certs/${cert}.crt -days 365 -nodes; done
    
    • CONFIG_SSL_CACERT_FILE=/etc/pki/tls/certs/selfcert.crt

    • CONFIG_SSL_CACERT_KEY_FILE=/etc/pki/tls/private/selfkey.key

    • CONFIG_VNC_SSL_CERT=/etc/pki/tls/certs/ssl_vnc.crt

    • CONFIG_VNC_SSL_KEY=/etc/pki/tls/private/ssl_vnc.key

    • CONFIG_HORIZON_SSL_CERT=/etc/pki/tls/certs/ssl_dashboard.crt

    • CONFIG_HORIZON_SSL_KEY=/etc/pki/tls/private/ssl_dashboard.key

    • CONFIG_HORIZON_SSL_CACERT=/etc/pki/tls/certs/selfcert.crt

  • CONFIG_<SERVICE>_INSTALL = Install a specific OpenStack service.

  • CONFIG_<NODE>_HOST = The host to setup the relevant services on.

  • CONFIG_<NODE>_HOSTS = A list of hosts to setup the relevant services on. This currently only exists for “COMPUTE” and “NETWORK.” New hosts can be added and Packstack re-run to have them added to the OpenStack cluster.

  • CONFIG_PROVISION_DEMO = Setup a demo project and user account with an image and network configured.

Uninstall

For uninstalling everything that is installed by Packstack, run this Bash script on all of the OpenStack nodes. Use at your own risk.

TripleO Quickstart

The TripleO Quickstart project was created to use Ansible to automate deploying a TripleO Undercloud and Overcloud. [7] The project recommends a minimum of 32GB RAM and 120GB of disk space when deploying with the default settings. [9] This deployment has to use a baremetal hypervisor. Deploying TripleO within a virtual machine that uses nested virtualization is not supported. [10]

  • Download the tripleo-quickstart script or clone the entire repository from OpenDev or GitHub.

    $ curl -O https://opendev.org/openstack/tripleo-quickstart/raw/branch/master/quickstart.sh
    

    OR

    $ git clone https://opendev.org/openstack/tripleo-quickstart.git
    $ cd tripleo-quickstart
    
  • Install dependencies for the quickstart script.

    $ sudo bash quickstart.sh --install-deps
    

TripleO can now be installed automatically with the default setup of 3 virtual machines. This will be created to meet the minimum TripleO cloud requirements: (1) an Undercloud to deploy a (2) controller and (3) compute node. [8] . Otherwise, a different node configuration from “config/nodes/” can be specified or created.

Common node variables:

  • {block|ceph|compute|control|default|objectstorage|undercloud}_{memory|vcpu} = Define the amount of processor cores or RAM (in megabytes) to allocate to the respective virtual machine type. Use “default” to apply to all nodes that are not explicitly defined.

Further customizations should be configured now before deploying the TripleO environment. Refer to the Undercloud Deploy role’s documentation on all of the Ansible variables for the Undercloud. Add any override variables to a YAML file and then add the arguments -e @<VARIABLE_FILE>.yaml to the “quickstart.sh” commands.

1. Automatic

  • Run the quickstart script to install TripleO. Use “127.0.0.2” for the localhost IP address if TripleO will be installed on the same system that the quickstart command is running on.

    $ bash quickstart.sh --release trunk/queens --tags all <REMOTE_HYPERVISOR_IP>
    

[7]

2. Manual

  • Common quickstart.sh options:

    • --clean = Remove previously created files from the working directory on the start of TripleO Quickstart.

    • --extra-vars supported_distro_check=false = Run on an unsupported hypervisor such as Fedora.

    • --no-clone = Use the current working directory for TripleO Quickstart. This should only be if the entire repository has been cloned.

    • --nodes config/nodes/<CONFIGURATION>.yml = Specify the configuration that determines how many Overcloud nodes should be deployed.

    • --playbook = Specify a Playbook to run.

    • --release = The OpenStack release to use. All of the available releases can be found in the OpenDev or GitHub project in the “config/release/” directory. Use “trunk/<RELEASE_NAME>” for the development version and “stable/<RELEASE_NAME>” for the stable version.

    • --retain-inventory = Use the existing inventory. This is useful for managing an existing TripleO Quickstart infrastructure.

    • --teardown {all|nodes|none|virthost} = Delete everything related to TripleO (all), only the virtual machines (nodes), nothing (none), or the virtual machines and settings on the hypervisor (virthost).

    • --tags all = Deploy a complete all-in-one TripleO installation automatically. If a Playbook is specified via -p, then everything in that Playbook will run.

    • -v = Show verbose output from the Ansible playbooks.

    • --config=~/.quickstart/config/general_config/containers_minimal.yml = Deploy the Overcloud from Kolla docker containers. [20]


  • Setup the Undercloud virtual machine.

    $ bash quickstart.sh --release trunk/queens --clean --teardown all --tags all --playbook quickstart.yml <REMOTE_HYPERVISOR_IP>
    
  • Install the Undercloud services.

    $ bash quickstart.sh --release trunk/queens --teardown none --no-clone --tags all --retain-inventory --playbook quickstart-extras-undercloud.yml <REMOTE_HYPERVISOR_IP>
    
  • Setup the Overcloud virtual machines.

    $ bash quickstart.sh --release trunk/queens --teardown none --no-clone --tags all --nodes config/nodes/1ctlr_1comp.yml --retain-inventory --playbook quickstart-extras-overcloud-prep.yml <REMOTE_HYPERVISOR_IP>
    
  • Install the Overcloud services.

    $ bash quickstart.sh --release trunk/queens --teardown none --no-clone --tags all --nodes config/nodes/1ctlr_1comp.yml --retain-inventory --playbook quickstart-extras-overcloud.yml <REMOTE_HYPERVISOR_IP>
    
  • Validate the installation.

    $ bash quickstart.sh --release trunk/queens --teardown none --no-clone --tags all --nodes config/nodes/1ctlr_1comp.yml --retain-inventory  --playbook quickstart-extras-validate.yml <REMOTE_HYPERVISOR_IP>
    

[11]

Standalone

Requirements:

  • 4 CPU cores

  • 8GB RAM

  • 50GB storage

OpenStack services:

  • Cinder

  • Glance

  • Keystone

  • Neutron

  • Nova

  • Placement

  • Swift

Starting with Rocky, an all-in-one Overcloud can be deployed without the need of an Undercloud. This is known as a Standalone deployment. It can be used for proof-of-concept TripleO deployments or for developers as an alternative to devstack. It is possible to deploy more than one Overcloud node using this method and to scale-up to more nodes at a later date. Although possible in upstream TripleO, in RHOSP this is unsupported by Red Hat.

The process is almost exactly the same as an Undercloud deployment. It deploys a fully functional Overcloud onto the local server. Unlike a typical Overcloud deployment (before the Victoria release where Mistral was removed), Mistral is not used. Instructions on how to setup a Standalone cloud are documented here. After the installation, the config-download Ansible playbooks will be available in the home directory as undercloud-ansible-<UUID>.

By default, some services, such as Heat, are disabled. Use this template to re-enable it.

---
resource_registry:
  OS::TripleO::Services::HeatApi: /usr/share/openstack-tripleo-heat-templates/deployment/heat/heat-api-container-puppet.yaml
  OS::TripleO::Services::HeatApiCfn: /usr/share/openstack-tripleo-heat-templates/deployment/heat/heat-api-cfn-container-puppet.yaml
  OS::TripleO::Services::HeatApiCloudwatch: /usr/share/openstack-tripleo-heat-templates/deployment/heat/heat-api-cloudwatch-disabled-puppet.yaml
  OS::TripleO::Services::HeatEngine: /usr/share/openstack-tripleo-heat-templates/deployment/heat/heat-engine-container-puppet.yaml

Updates

These steps apply to both Undercloud and Standalone cloud deployments.

  • Update:

    $ openstack {undercloud install|tripleo deploy} --force-stack-update
    
  • Upgrade:

    $ openstack {undercloud|tripleo} upgrade
    
  • Reinstall:

    $ openstack {undercloud install|tripleo deploy}  --force-stack-create
    

[48]

InfraRed 2

InfraRed uses Ansible playbooks to automate deploying downstream RHOSP packages and upstream RDO packages.

Install InfraRed into a Python 2 virtual environment.

$ virtualenv ~/venv_infrared
$ source ~/venv_infrared/bin/activate
$ git clone https://github.com/redhat-openstack/infrared.git
$ cd infrared
$ pip2 install --user .

As of 2019, these are the officially supported plugins in InfraRed.

  • provision

    • beaker

    • docker

    • foreman

    • openstack

    • virsh

  • install

    • build-packages

    • cloud-config

    • containers-sanity

    • install-ceph

    • oooq

    • packstack

    • patch-components

    • tripleo-overcloud

    • tripleo-standalone

    • tripleo-undercloud

  • test

    • browbeat

    • bzaf

    • gabbi

    • jordan

    • openstack-coverage

    • ospdui

    • pytest-runner

    • rally

    • robot

    • tempest

    • tripleo-config-changes

    • tripleo-post-tests

  • other

    • collect-logs

    • dellemc-idrac

    • list-builds

Use the infrared plugin search command to view the GitHub URL of each plugin. Then use infrared plugin add <GITHUB_URL> to install the plugin.

Alternatively, install plugins from the working directory of the infrared repository.

Install a provision plugin, such as virsh, along with the required plugins for deploying and managing a TripleO cloud.

$ infrared plugin add plugins/virsh
$ infrared plugin add plugins/tripleo-undercloud
$ infrared plugin add plugins/tripleo-overcloud
$ infrared plugin add plugins/cloud-config
  • Optionally create an answers file manually or by using the CLI and then import it. Otherwise, use the CLI arguments.

    $ infrared virsh --from-file=virsh_prov.ini
    
  • [virsh]

    • host-address = Required argument. Edit with any value, OR override with CLI: –host-address=<option>

    • host-memory-overcommit = Default: False.

    • host-key = Required argument. Edit with any value, OR override with CLI: –host-key=<option>

    • host-user = Default: root.

    • topology-nodes = The number of each node to deploy.

      • Minimal: "ovb_undercloud:1,controller:1,compute:1".

      • Minimal with OpenStack Virtual Baremetal (OVB) support for provisioning: "ovb_undercloud:1,ovb_controller:1,ovb_compute:1".

  • Deploy the virtual machines that will be used by the lab.

    • Virsh provisioner:

      $ infrared virsh --host-address 127.0.0.1 --host-key ~/.ssh/id_rsa --host-memory-overcommit yes --topology-nodes "ovb_undercloud:1,controller:1,compute:1"
      
    • OpenStack provisioner:

      $ infrared openstack --cloud ${OS_CLOUD} --prefix <OPTIONAL_RESOURCE_PREFIX> --key-file ~/.ssh/id_rsa --topology-network 3_nets_ovb --topology-nodes "ovb_undercloud:1,ovb_controller:1,ovb_compute:1" --anti-spoofing False --dns <DNS1>,<DNS2> --provider-network <EXTERNAL_PROVIDER_NETWORK> --image <RHEL_OR_CENTOS> --username <SSH_USER>
      
    • An Ansible inventory of the hosts will be generated here: ~/.infrared/.workspaces/active/hosts.

  • Deploy the Undercloud.

    • RHOSP:

      $ RHOSP_VERSION=16
      $ infrared tripleo-undercloud --version ${RHOSP_VERSION} --build ${PUDDLE_VERSION} --images-task rpm --ntp-server clock.redhat.com,clock2.redhat.com
      
    • RDO:

      $ RDO_VERSION=train
      $ infrared tripleo-undercloud --version ${RDO_VERSION} --images-task=import --images-url=https://images.rdoproject.org/${RDO_VERSION}/rdo_trunk/current-tripleo/stable/
      
  • Deploy the Overcloud.

    $ infrared tripleo-overcloud --deployment-files virt --version ${RDO_VERSION} --introspect yes --tagging yes --deploy yes
    
  • After the Overcloud is deployed, optionally configure resources on it.

    $ infrared cloud-config --deployment-files virt --tasks create_external_network,forward_overcloud_dashboard,network_time,tempest_deployer_input
    

[35]

Deployment (Full)

Minimum recommended requirements [8]:

  • Undercloud node:

    • 4 CPU cores

    • 8GB RAM (16GB recommended)

    • 60GB storage

    • 2 network interface cards (NICs) [21]

    • A fully qualified domain name (FQDN)

  • Overcloud nodes:

    • 4 CPU cores

    • 8GB RAM

    • 80GB storage

Here is an overview of the deployment process using TripleO:

  • Install the all-in-one Undercloud. This cloud will be used by the OpenStack operator to control and manage the Overcloud.

  • Import the Overcloud nodes into Ironic.

  • Configure those nodes to load both an initramfs and full kernel via a PXE boot.

  • Optionally set the nodes to be “manageable” and introspect the Overcloud nodes. This will report back detailed information about each node.

  • Set the Overcloud nodes to be “available” for provisioning.

  • Optionally configure settings for the Overcloud deployment (highly recommended).

  • Deploy the Overcloud. This cloud will be the production cloud that developers can use.

RHOSP enables high-availability (HA) for the control plane by default and requires having exactly 3 Controller nodes as part of the Overcloud. [45] TripleO can have HA enabled by setting the ControllerCount to 3 and including this template: -e /usr/share/openstack-tripleo-heat-templates/environments/docker-ha.yaml. [46]

Undercloud

The Undercloud can be installed onto a bare metal server or a virtual machine. Follow the “hypervisor” section to assist with automatically creating an Undercloud virtual machine. The Undercloud requires at least 2 NICs (typically eth0 and eth1). The first is used for external connectivity. The second is dedicated to provisioning the Overcloud nodes with Ironic. On those nodes, the related interface that can reach the Undercloud’s eth1 should be configured for PXE booting in the BIOS. [21]

Considerations before starting the Undercloud deployment:

  • The Undercloud server requires two network interfaces. One with public Internet/management access and the second dedicated to provisioning.

  • Configure the hostname.

  • Set push_destination: True in a custom container-image-prepare.yaml file.

  • undercloud.conf
    • The NTP and DNS resolvers need to be accurate and accessible.

    • If deploying or managing more than 250 hosts, it is required to change the ctlplane-subnet to a use a subnet mask with more available IP addresses.

    • Use the custom container-image-prepare.yaml file.

  • Undercloud (Automatic)

    • RDO provides pre-made Undercloud images.

      • <= Queens:

        $ curl -O https://images.rdoproject.org/queens/delorean/current-tripleo-rdo/undercloud.qcow2
        
      • >= Rocky:

        $ curl -O https://images.rdoproject.org/rocky/rdo_trunk/current-tripleo-rdo/undercloud.qcow2
        
    • TripleO Quickstart can build an Undercloud image.

      • Leave the overcloud_nodes variable blank to only deploy the Undercloud. Otherwise, provide a number of virtual machines that should be created for use in the Overcloud.

      $ curl -O https://opendev.org/openstack/tripleo-quickstart/raw/branch/master/quickstart.sh
      $ bash quickstart.sh --release trunk/queens --tags all --playbook quickstart.yml -e overcloud_nodes="" $VIRTHOST
      
    • Log into the virtual machine once TripleO Quickstart has completed setting up the environment.

      $ ssh -F ~/.quickstart/ssh.config.ansible undercloud
      
  • Undercloud (Manual)

    • It is recommended to create a user named “stack” with sudo privileges to manage the Undercloud.

      $ sudo useradd stack
      $ sudo passwd stack
      $ echo "stack ALL=(root) NOPASSWD:ALL" | sudo tee -a /etc/sudoers.d/stack
      $ sudo chmod 0440 /etc/sudoers.d/stack
      $ su - stack
      
    • Install TripleO. For <= Stein, install python-tripleoclient instead.

      $ sudo yum install python3-tripleoclient openstack-tripleo-common openstack-tripleo-heat-templates
      
    • Update the operating system and reboot the server.

      $ sudo yum update && sudo reboot
      
    • Copy the sample configuration to use as a base template. Optionally configure it.

      • <= Stein:

        $ cp /usr/share/instack-undercloud/undercloud.conf.sample ~/undercloud.conf
        
      • >= Train:

        $ cp /usr/share/python-tripleoclient/undercloud.conf.sample ~/undercloud.conf
        
    • Common Undercloud configuration options. If using an automated power management driver with Ironic, the IP address for the Undercloud’s provisioning NIC must use the same network and broadcast domain. [15]

      • enable_* = Enable or disable non-essential OpenStack services on the Undercloud.

      • dhcp_{start|end} = The range of IP addresses to temporarily use for provisioning Overcloud nodes. This range is a limiting factor in how many nodes can be provisioned at once.

      • local_interface = The network interface to use for provisioning new Overcloud nodes. This will be configured as an Open vSwitch bridge. Default: eth1.

      • local_ip = The local IP address of the Undercloud node to be used for using DHCP for providing IP addresses for Overcloud nodes during PXE booting. This should not be a public IP address.

      • inspection_iprange = The IP range to use for Ironic’s introspection of the Overcloud nodes. This range needs to unique from the DHCP start/end range.

      • local_mtu = The MTU size to use for the local interface.

      • cidr = The CIDR range of IP addresses to use for the Overcloud nodes.

      • masquerade_network = The network CIDR that will be used for masquerading external network connections.

      • gateway = The default gateway to use for external connectivity to the Internet during provisioning. Use the “local_ip” when masquerading is used.

      • undercloud_admin_vip = The IP address to listen on for admin API endpoints.

      • undercloud_hostname = The fully qualified hostname to use for the Undercloud.

      • undercloud_nameservers = A list of DNS resolvers to use.

      • undercloud_ntp_servers = A list of NTP servers to use.

      • undercloud_public_vip = The IP address to listen on for public API endpoints.

      • enabled_hardware_types = The Ironic power management drivers to enable. For virtual lab environments, append “manual-management”.

    • Example of changing the control plane (provisioning) network details.

      [DEFAULT]
      undercloud_admin_host = 192.168.100.3
      undercloud_public_host = 192.168.100.2
      [ctlplane-subnet]
      cidr = 192.168.100.0/24
      dhcp_start = 192.168.100.4
      dhcp_end = 192.168.100.150
      gateway = 192.168.100.1
      inspection_iprange = 192.168.100.201,192.168.100.250
      masquerade = true
      
    • Deploy the Undercloud. Anytime the configuration for the Undercloud changes, this command needs to be re-ran to update the installation.

      $ openstack undercloud install
      
    • The installation will be logged to $HOME/.instack/install-undercloud.log.

    • After the installation, OpenStack user credentials will be saved to $HOME/stackrc. Source this file before running OpenStack commands to verify that the Undercloud is operational.

      $ source ~/stackrc
      $ openstack catalog list
      
    • All OpenStack service passwords will be saved to $HOME/undercloud-passwords.conf.

[12]

The next step is to optionally provision the Overcloud nodes and then deploy the OpenStack services.

Uninstall

Use the script provided here to uninstall the Undercloud services.

Overcloud (Provision Nodes with Ironic)

TripleO can provision a full CentOS or RHEL operating system onto a new baremetal server using the Ironic service. The normal TripleO deployment process is split into these steps [59]:

  • Upload pre-built Overcloud image files to Glance.

  • Import the instackenv file with power management details about the nodes.

  • Introspect the nodes. This will PXE/network boot the Overcloud nodes so that Ironic can gather hardware information used during provisioning.

  • Deploy the Overcloud. This will automatically provision the nodes. Provisioning can optionally be done manually before the deployment.


  • Image Preparation

    • GA releases do not have pre-built Overcloud image files. They must be manually created. [60]

      • Make and switch into an “images” directory.

        $ mkdir ~/images
        $ cd ~/images
        
      • Set the environment variable for the RDO repositories so all necessary packages can be installed. Then build all of the required images. [13]

        • <= Ocata

          $ export DIB_YUM_REPO_CONF="/etc/yum.repos.d/delorean*"
          $ openstack overcloud image build --all
          
        • >= Pike

          $ export DIB_YUM_REPO_CONF="/etc/yum.repos.d/delorean*"
          $ openstack overcloud image build
          
        • >= Train (CentOS 8)

          $ sudo dnf config-manager --set-enabled HighAvailability
          $ sudo dnf config-manager --set-enabled PowerTools
          $ export DIB_YUM_REPO_CONF="/etc/yum.repos.d/delorean* /etc/yum.repos.d/CentOS-HA.repo /etc/yum.repos.d/CentOS-PowerTools.repo"
          $ openstack overcloud image build
          
    • RDO Trunk (current-tripleo-rdo)

      $ export OS_RELEASE="train"
      $ export TRUNK_BRANCH="current-tripleo-rdo"
      $ mkdir ~/images
      $ cd ~/images
      $ curl -O https://images.rdoproject.org/${OS_RELEASE}/rdo_trunk/${TRUNK_BRANCH}/ironic-python-agent.tar
      $ curl -O https://images.rdoproject.org/${OS_RELEASE}/rdo_trunk/${TRUNK_BRANCH}/overcloud-full.tar
      $ tar -v -x -f ironic-python-agent.tar
      $ tar -v -x -f overcloud-full.tar
      
    • RHOSP [38]

      $ export OS_RELEASE="13.0"
      $ mkdir ~/images
      $ cd ~/images
      $ sudo yum install rhosp-director-images rhosp-director-images-ipa
      $ tar -v -x -f /usr/share/rhosp-director-images/overcloud-full-latest-${OS_RELEASE}.tar
      $ tar -v -x -f /usr/share/rhosp-director-images/ironic-python-agent-latest-${OS_RELEASE}.tar
      
  • These are the default image files that will be used.

    • ironic-python-agent.initramfs

    • ironic-python-agent.kernel

    • overcloud-full.initrd

    • overcloud-full.qcow2

    • overcloud-full.vmlinuz

  • Upload those images.

    $ openstack overcloud image upload --image-path /home/stack/images/
    
  • For using containers, the RDO images from Docker Hub are configured by default. Enable container caching on the Undercloud by generating this template. This will increase the Overcloud deployment time since container images will only have to be pulled from Docker Hub once. [33]

    $ openstack tripleo container image prepare default --output-env-file ~/templates/containers-prepare-parameter.yaml
    

Introspection

  • Create an instackenv.{json|yaml} file that describes the physical infrastructure of the Overcloud. [15] By default Ironic manages rebooting machines using the IPMI “pxe_ipmitool” driver. [18] Below are the common values to use that define how to handle power management (PM) for the Overcloud nodes via Ironic.

    • All

      • name = A descriptive name of the node.

      • pm_type = The power management driver type to use. Common drivers include “pxe_ipmitool” and “manual-management”.

      • capabilities = Set custom capabilities. For example, the profile and boot options can be defined here: "profile:compute,boot_option:local".

    • IPMI

      • pm_user = The PM user to use.

      • pm_password = The PM password to use.

      • pm_addr = The PM IP address to use.

    • Fake PXE

      • arch = The processor architecture. The standard is “x86_64”.

      • cpu = The number of processor cores.

      • mac = A list of MAC addresses that should be used for the PXE boot. This normally only contains one value.

      • memory = The amount of RAM, in MiB.

      • disk = The amount of disk space, in GiB. Set this to be 1 GiB less than the actual reported storage size. That will prevent partitioning issues during the Overcloud deployment.

    • instackenv.json syntax:

      {
          "nodes": [
              {
                  "name": "<DESCRIPTIVE_NAME>",
                  "pm_type": "manual-management",
                  "arch": "x86_64",
                  "cpu": "<CPU_CORES>",
                  "memory": "<RAM_MB>",
                  "disk": "<DISK_GB>",
                  "capabilities": "profile:control,boot_option:local"
              },
              {
                  "name": "<DESCRIPTIVE_NAME>",
                  "pm_type": "pxe_ipmitool",
                  "pm_user": "<IPMI_USER>",
                  "pm_password": "<IPMI_PASSWORD>",
                  "pm_addr": "<IPMI_IP_ADDRESS>",
                  "mac": [
                      "AA:BB:CC:DD:EE:FF"
                  ],
                  "capabilities": "profile:compute,boot_option:local"
              }
          ]
      }
      
    • instackenv.yaml minimal syntax:

      ---
      nodes:
        - name: <DESCRIPTIVE_NAME>
          pm_type: manual-management
          mac:
            - "AA:BB:CC:DD:EE:FF"
          capabilities: "profile:<FLAVOR>,boot_option:local"
      
    • instackenv.yaml full syntax:

      ---
      nodes:
        - name: <DESCRIPTIVE_NAME>
          pm_type: manual-management
          arch: x86_64
          cpu: <CPU_CORES>
          memory: <RAM_MB>
          disk: <DISK_GB>
          mac:
            - "AA:BB:CC:DD:EE:FF"
          capabilities: "profile:control,boot_option:local"
        - name: <DESCRIPTIVE_NAME>
          pm_type: pxe_ipmitool
          pm_user: <IPMI_USER>
          pm_password: <IPMI_PASSWORD>
          pm_addr: <IPMI_IP_ADDRESS>
          capabilities: "profile:compute,boot_option:local"
      
    • Virtual lab environment:

      • The “manual-management” driver can be used. This will require the end-user to manually reboot the managed nodes.

      • Virtual machines deployed using Vagrant need to have vagrant-libvirt’s default eth0 management interface removed. The first interface on the machine (normally eth0) is used for introspection and provisioning and cannot be that management interface.

        $ sudo virsh detach-interface ${VM_NAME} network --persistent --mac $(sudo virsh dumpxml ${VM_NAME} | grep -B4 vagrant-libvirt | grep mac | cut -d "'" -f2)
        
  • Import the nodes and then introspect them immediately. [24]

    $ openstack overcloud node import --introspect --provide instackenv.json
    
  • Alternatively, import them and inspect them later.

    $ openstack overcloud node import instackenv.json
    Started Mistral Workflow tripleo.baremetal.v1.register_or_update. Execution ID: cf2ce144-a22a-4838-9a68-e7c3c5cf0dad
    Waiting for messages on queue 'tripleo' with no timeout.
    2 node(s) successfully moved to the "manageable" state.
    Successfully registered node UUID c1456e44-5245-4a4d-b551-3c6d6217dac4
    Successfully registered node UUID 9a277de3-02be-4022-ad26-ec4e66d97bd1
    
    • Verify that Ironic has successfully added the new baremetal nodes.

      $ openstack baremetal node list
      +--------------------------------------+-----------+---------------+-------------+--------------------+-------------+
      | UUID                                 | Name      | Instance UUID | Power State | Provisioning State | Maintenance |
      +--------------------------------------+-----------+---------------+-------------+--------------------+-------------+
      | c1456e44-5245-4a4d-b551-3c6d6217dac4 | control01 | None          | None        | manageable         | False       |
      | 9a277de3-02be-4022-ad26-ec4e66d97bd1 | compute01 | None          | None        | manageable         | False       |
      +--------------------------------------+-----------+---------------+-------------+--------------------+-------------+
      
  • Start the introspection. [24] Each Overcloud node requires at least 4GB of RAM or else the introspection will fail with a kernel panic during the network booted live session.

    • Method #1: Automatical introspection with a managed Ironic driver (such as IPMI). This command will introspect all nodes in the management state and set them to the available state when complete.

      $ openstack overcloud node introspect --all-manageable --provide
      Waiting for introspection to finish...
      Waiting for messages on queue 'tripleo' with no timeout.
      Introspection of node c1456e44-5245-4a4d-b551-3c6d6217dac4 completed. Status:SUCCESS. Errors:None
      Introspection of node 9a277de3-02be-4022-ad26-ec4e66d97bd1 completed. Status:SUCCESS. Errors:None
      Introspection completed.
      Waiting for messages on queue 'tripleo' with no timeout.
      2 node(s) successfully moved to the "available" state.
      
    • Method #2: Automatic but the connection details are given via the CLI instead of the instackenv file.

      • Automatically discover the available servers by scanning hardware devices (such as IPMI) via a CIDR range and using different logins.

        $ openstack overcloud node discover --range <CIDR> --credentials <USER1>:<PASSWORD1> --credentials <USER2>:<PASSWORD2>
        
    • Method #3: Lab environment using the manual-management driver.

      • In another terminal, verify that the “Power State” is “power on” and then manually start the virtual machines. The introspection will take a long time to complete.

        $ openstack overcloud node introspect --all-manageable --provide
        
        $ openstack baremetal node list
        +--------------------------------------+-----------+---------------+-------------+--------------------+-------------+
        | UUID                                 | Name      | Instance UUID | Power State | Provisioning State | Maintenance |
        +--------------------------------------+-----------+---------------+-------------+--------------------+-------------+
        | c1456e44-5245-4a4d-b551-3c6d6217dac4 | control01 | None          | power on    | manageable         | False       |
        | 9a277de3-02be-4022-ad26-ec4e66d97bd1 | compute01 | None          | power on    | manageable         | False       |
        +--------------------------------------+-----------+---------------+-------------+--------------------+-------------+
        
      • When the “Power State” becomes “power off” and the “Provisioning State” becomes “available” then manually shutdown the virtual machines.

        $ openstack baremetal node list
        +--------------------------------------+-----------+---------------+-------------+--------------------+-------------+
        | UUID                                 | Name      | Instance UUID | Power State | Provisioning State | Maintenance |
        +--------------------------------------+-----------+---------------+-------------+--------------------+-------------+
        | c1456e44-5245-4a4d-b551-3c6d6217dac4 | control01 | None          | power off   | available          | False       |
        | 9a277de3-02be-4022-ad26-ec4e66d97bd1 | compute01 | None          | power off   | available          | False       |
        +--------------------------------------+-----------+---------------+-------------+--------------------+-------------+
        
  • Configure the necessary flavors (mandatory for getting accurate results when using the manual-management Ironic driver). [25] Commonly custom “control” and “compute” flavors will need to be created.

    $ openstack flavor create --id auto --vcpus <CPU_COUNT> --ram <RAM_IN_MB> --disk <DISK_IN_GB_MINUS_ONE> --swap <SWAP_IN_MB> --property "capabilities:profile"="<FLAVOR_NAME>" <FLAVOR_NAME>
    
  • Configure the kernel and initramfs that the baremetal nodes should boot from.

    • Queens (optional) [24]:

      $ openstack baremetal node list
      $ openstack overcloud node configure <NODE_ID>
      
  • If the profile and/or boot option were not specified in the instackenv.json file then configure it now. Verify that the profiles have been applied. Valid default flavors are block-storage, ceph-storage, compute, control, and swift-storage.

    $ openstack baremetal node set --property capabilities='profile:control,boot_option:local' c1456e44-5245-4a4d-b551-3c6d6217dac4
    $ openstack baremetal node set --property capabilities='profile:compute,boot_option:local' 9a277de3-02be-4022-ad26-ec4e66d97bd1
    $ openstack overcloud profiles list --all
    +--------------------------------------+-----------+-----------------+-----------------+-------------------+-------+
    | Node UUID                            | Node Name | Provision State | Current Profile | Possible Profiles | Error |
    +--------------------------------------+-----------+-----------------+-----------------+-------------------+-------+
    | c1456e44-5245-4a4d-b551-3c6d6217dac4 | control01 | available       | control         |                   |       |
    | 9a277de3-02be-4022-ad26-ec4e66d97bd1 | compute01 | available       | compute         |                   |       |
    +--------------------------------------+-----------+-----------------+-----------------+-------------------+-------
    
  • Set a DNS nameserver on the control plane subnet. Starting with Rocky, this is automatically set to the value of undercloud_nameservers from the undercloud.conf configuration.

    $ openstack subnet set --dns-nameserver 8.8.8.8 --dns-nameserver 1.1.1.1 ctlplane-subnet
    

Deployment

  • Configure the networking Heat templates that define the physical and virtual network interface settings.

    • Scenario #1 - Default templates:

      $ cd /usr/share/openstack-tripleo-heat-templates/
      $ mkdir /home/stack/templates/
      $ /usr/share/openstack-tripleo-heat-templates/tools/process-templates.py -o /home/stack/templates/
      
    • Scenario #2 - Variables can be customized via the “roles_data.yaml” and “network_data.yml” files. Example usage can be found here.

      $ mkdir /home/stack/templates/
      $ cp /usr/share/openstack-tripleo-heat-templates/roles_data.yaml /home/stack/templates/roles_data_custom.yaml
      $ cp /usr/share/openstack-tripleo-heat-templates/network_data.yml /home/stack/templates/network_data_custom.yaml
      $ /usr/share/openstack-tripleo-heat-templates/tools/process-templates.py --roles-data ~/templates/roles_data_custom.yaml --roles-data ~/templates/network_data_custom.yaml
      
    • Scenario #3 - No templates:

      • If no custom network settings will be used, then the Heat templates do not need to be generated. By default, TripleO will configure different subnets to separate traffic (instead of also using VLANs) onto the default network interface of the Overcloud nodes.

  • In a YAML Heat template, set the number of controller, compute, Ceph, and/or any other nodes that should be deployed.

    ---
    parameter_defaults:
      OvercloudControllerFlavor: control
      OvercloudComputeFlavor: compute
      OvercloudCephStorageFlavor: ceph
      ControllerCount: <NUMBER_OF_CONTROLLER_NODES>
      ComputeCount: <NUMBER_OF_COMPUTE_NODES>
      CephStorageCount: <NUMBER_OF_CEPH_NODES>
    
  • Alternatively, the initial default count can be set in the roles_data.yaml file.

    - name: Controller
      CountDefault: <NUMBER_OF_CONTROLLER_NODES>
    - name: Compute
      CountDefault: <NUMBER_OF_COMPUTE_NODES>
    - name: CephStorage
      CountDefault: <NUMBER_OF_CEPHSTORAGE_NODES>
    
  • Deploy the Overcloud with any custom Heat configurations. [13] Starting with the Pike release, most services are deployed as containers by default. For preventing the use of containers, remove the “docker.yaml” and “docker-ha.yaml” files from ${TEMPLATES_DIRECTORY}/environments/. [14]

    $ openstack help overcloud deploy
    $ openstack overcloud deploy --templates ~/templates -r ~/templates/roles_data_custom.yaml
    
    • Virtual lab environment:

      • When the “Provisioning State” becomes “wait call-back” then manually start the virtual machines. The relevant Overcloud image will be copied to the local drive(s). At this point, Nova will have already changed the servers to have the “Status” of “BUILD”.

        $ openstack baremetal node list
        +--------------------------------------+-----------+--------------------------------------+-------------+--------------------+-------------+
        | UUID                                 | Name      | Instance UUID                        | Power State | Provisioning State | Maintenance |
        +--------------------------------------+-----------+--------------------------------------+-------------+--------------------+-------------+
        | c1456e44-5245-4a4d-b551-3c6d6217dac4 | control01 | 16a09779-b324-4d83-bc7d-3d24d2f4aa5d | power on    | wait call-back     | False       |
        | 9a277de3-02be-4022-ad26-ec4e66d97bd1 | compute01 | 5c2d1374-8b20-4af6-b114-df15bbd3d9ca | power on    | wait call-back     | False       |
        +--------------------------------------+-----------+--------------------------------------+-------------+--------------------+-------------+
        $ openstack server list
        +--------------------------------------+-------------------------+--------+------------------------+----------------+---------+
        | ID                                   | Name                    | Status | Networks               | Image          | Flavor  |
        +--------------------------------------+-------------------------+--------+------------------------+----------------+---------+
        | 9a277de3-02be-4022-ad26-ec4e66d97bd1 | overcloud-novacompute-0 | BUILD  | ctlplane=192.168.24.35 | overcloud-full | compute |
        | c1456e44-5245-4a4d-b551-3c6d6217dac4 | overcloud-controller-0  | BUILD  | ctlplane=192.168.24.34 | overcloud-full | control |
        +--------------------------------------+-------------------------+--------+------------------------+----------------+---------+
        
      • The nodes will then be in the “Provisioning State” of “deploying”. At this phase the operating system image is copied over, partitions are resized, and SSH keys are configured for access to the heat-admin user account.

        $ openstack baremetal node list
        +--------------------------------------+-----------+--------------------------------------+-------------+--------------------+-------------+
        | UUID                                 | Name      | Instance UUID                        | Power State | Provisioning State | Maintenance |
        +--------------------------------------+-----------+--------------------------------------+-------------+--------------------+-------------+
        | c1456e44-5245-4a4d-b551-3c6d6217dac4 | control01 | 16a09779-b324-4d83-bc7d-3d24d2f4aa5d | power on    | deploying          | False       |
        | 9a277de3-02be-4022-ad26-ec4e66d97bd1 | compute01 | 5c2d1374-8b20-4af6-b114-df15bbd3d9ca | power on    | deploying          | False       |
        +--------------------------------------+-----------+--------------------------------------+-------------+--------------------+-------------+
        
      • After that is complete, the virtual machines will power off. Ironic will report that the “Power State” is now “power on” and the Provisioning State” is now “active.” The nodes have now been provisioned with the Overcloud image. Change the boot order of each machine to start with the hard drive instead of the network interface card. Manually start the virtual machines after that.

        $ openstack baremetal node list
        +--------------------------------------+-----------+--------------------------------------+-------------+--------------------+-------------+
        | UUID                                 | Name      | Instance UUID                        | Power State | Provisioning State | Maintenance |
        +--------------------------------------+-----------+--------------------------------------+-------------+--------------------+-------------+
        | c1456e44-5245-4a4d-b551-3c6d6217dac4 | control01 | 16a09779-b324-4d83-bc7d-3d24d2f4aa5d | power on    | active             | False       |
        | 9a277de3-02be-4022-ad26-ec4e66d97bd1 | compute01 | 5c2d1374-8b20-4af6-b114-df15bbd3d9ca | power on    | active             | False       |
        +--------------------------------------+-----------+--------------------------------------+-------------+--------------------+-------------+
        
  • The deploy will continue onto the configuration management stage. Before Rocky, this process used os-collect-config (Heat). Starting with Rocky, this now uses config-download (Ansible).

2019-10-30 23:40:47Z [overcloud-AllNodesDeploySteps-5yoxyq2a4bgz]: UPDATE_COMPLETE  Stack UPDATE completed successfully
2019-10-30 23:40:47Z [AllNodesDeploySteps]: UPDATE_COMPLETE  state changed
2019-10-30 23:40:51Z [overcloud]: UPDATE_COMPLETE  Stack UPDATE completed successfully

 Stack overcloud UPDATE_COMPLETE

Deploying overcloud configuration
Enabling ssh admin (tripleo-admin) for hosts:
192.168.24.17 192.168.24.16
Using ssh user cloud-user for initial connection.
Using ssh key at /home/stack/.ssh/id_rsa for initial connection.
Inserting TripleO short term key for 192.168.24.17
Warning: Permanently added '192.168.24.17' (ECDSA) to the list of known hosts.
Inserting TripleO short term key for 192.168.24.16
Warning: Permanently added '192.168.24.16' (ECDSA) to the list of known hosts.
Starting ssh admin enablement workflow
Started Mistral Workflow tripleo.access.v1.enable_ssh_admin. Execution ID: 0a69a3a3-d9bb-43c6-8aed-0ef33f6336d7
ssh admin enablement workflow - RUNNING.
ssh admin enablement workflow - RUNNING.
ssh admin enablement workflow - COMPLETE.
Removing TripleO short term key from 192.168.24.17
Warning: Permanently added '192.168.24.17' (ECDSA) to the list of known hosts.
  • Once the deployment is complete, verify that the Overcloud was deployed successfully. If it was not, then troubleshoot any stack resources that failed.

    PLAY RECAP *********************************************************************
    overcloud-controller-0     : ok=257  changed=142  unreachable=0    failed=0
    overcloud-novacompute-0    : ok=178  changed=78   unreachable=0    failed=0
    undercloud                 : ok=21   changed=12   unreachable=0    failed=0
    
    Wednesday 13 February 2019  14:38:34 -0500 (0:00:00.103)       0:40:32.320 ****
    ===============================================================================
    
    Ansible passed.
    Overcloud configuration completed.
    Waiting for messages on queue 'tripleo' with no timeout.
    Host 192.168.24.23 not found in /home/stack/.ssh/known_hosts
    Overcloud Endpoint: http://192.168.24.23:5000
    Overcloud Horizon Dashboard URL: http://192.168.24.23:80/dashboard
    Overcloud rc file: /home/stack/overcloudrc
    Overcloud Deployed
    
    $ openstack stack list
    $ openstack stack failures list <OVERCLOUD_STACK_ID> --long
    $ openstack stack show <OVERCLOUD_STACK_ID>
    $ openstack stack resource list <OVERCLOUD_STACK_ID>
    $ openstack stack resource show <OVERCLOUD_STACK_ID> <RESOURCE_NAME>
    $ openstack overcloud failures list # Requires >= Rocky
    
  • Source the Overcloud admin credentials to manage it.

    $ source ~/overcloudrc
    
  • The nodes can be managed via SSH using the “heat-admin” user.

    $ openstack server list
    +--------------------------------------+-------------------------+--------+------------------------+----------------+---------+
    | ID                                   | Name                    | Status | Networks               | Image          | Flavor  |
    +--------------------------------------+-------------------------+--------+------------------------+----------------+---------+
    | 9a277de3-02be-4022-ad26-ec4e66d97bd1 | overcloud-novacompute-0 | ACTIVE | ctlplane=192.168.24.35 | overcloud-full | compute |
    | c1456e44-5245-4a4d-b551-3c6d6217dac4 | overcloud-controller-0  | ACTIVE | ctlplane=192.168.24.34 | overcloud-full | control |
    +--------------------------------------+-------------------------+--------+------------------------+----------------+---------+
    $ ssh -l heat-admin 192.168.24.34
    

[13][23]

  • Passwords for the Overcloud services can be found by running:

    • TripleO Queens:

      $ openstack object save overcloud plan-environment.yaml
      
  • In >= Rocky (or in Queens, if configured), the Ansible files used for the configuration management can be downloaded. Those files can then be imported into an external source such as Ansible Tower or AWX. The tripleo-ansible-inventory script is used to generate a dynamic inventory file for Ansible that contains the Overcloud hosts. [30]

    $ openstack overcloud config download
    
  • For a lab with a private network, use a proxy service from the hypervisor to access the dashboard and API IP address.

    $ sshuttle -r stack@undercloud 192.168.24.23
    

Overcloud (Pre-deployed/provisioned Nodes)

Introspection and the operating system provisioning can be skipped if the Overcloud nodes are already setup and running.

Pros:

  • Easier to deploy, subjectively.

  • Faster to deploy if using a pre-configured operating system snapshot.

  • No Nova or Ironic dependencies.

Cons:

  • All Overclouds nodes must be pre-provisioned. Ironic cannot manage any for provisioning.

  • Requires the operating system to already be installed.

  • Repositories have to be installed and enabled manually.

  • Validations are not supported.


Overcloud Nodes

  • Install CentOS or RHEL.

  • Create a stack user. Add the stack user’s SSH key from the Undercloud to allow access during deployment.

    • Alternatively, specify a different user for the deployment with openstack overcloud deploy --overcloud-ssh-user <USER> --overcloud-ssh-key <PRIVATE_KEY_FLIE>. This user is only used during the initial deployment to create a tripleo-admin user (or the user heat-admin in Queens release and older).

  • Enable the RDO or RHOSP repositories.

  • Install the Heat user agent (required only for <= Queens when not using config-download).

    $ sudo yum -y install python-heat-agent*
    

Undercloud/Director

  • For config-download scenarios on < Train, generate Heat templates for pre-provisioned nodes from a special roles data file. Starting in Train, it uses the default /usr/share/openstack-tripleo-heat-templates/roles_data.yaml file. Previously, roles such as ControllerDeployedServer and ComputeDeployedServer were used. These now use the standard Controller and Compute roles.

    $ cd /usr/share/openstack-tripleo-heat-templates/
    $ mkdir /home/stack/templates/
    $ /usr/share/openstack-tripleo-heat-templates/tools/process-templates.py --roles-data /usr/share/openstack-tripleo-heat-templates/deployed-server/deployed-server-roles-data.yaml --output /home/stack/templates/
    
  • TripleO needs a hostname and port mapping to know what IP addresses to connect to for the deployment. The NeutronPublicInterface (eth0 by default) will be converted into a bridge (br-ex by default). It will have static IP addressing set to what the fixed_ips and cidr are set to. The ControlPlaneDefaultRoute will set the default route in /etc/sysconfig/network-scripts/route-br-ex.

  • Scenario 1: Use the Undercloud control plane network.

    • The control plane IP address of each Overcloud node should be within the range of the network_cidr value defined in the undercloud.conf configuration. By default this is 192.168.24.0/24 with 192.168.24.{1,2,3} all being reserved/used by the Undercloud.

      ---
      resource_registry:
        # This allows the IPs for provisioning to be manually set via DeployedServerPortMap.
        OS::TripleO::DeployedServer::ControlPlanePort: /usr/share/openstack-tripleo-heat-templates/deployed-server/deployed-neutron-port.yaml
        # These role resources will convert the NeutronPublicInterface into the required br-ex bridge interface.
        ## Open vSwitch
        OS::TripleO::ControllerDeployedServer::Net::SoftwareConfig: net-config-static-bridge.yaml
        OS::TripleO::ComputeDeployedServer::Net::SoftwareConfig: net-config-static-bridge.yaml
      
      parameter_defaults:
        # The Overcloud NIC that has a default route.
        ## Specify the exact network interface name.
        ## Alternatively, use a Heat alias such as "nic1" (eth0) or "nic2" (eth1) if the NICs are named
        ## differently on the Overcloud nodes.
        NeutronPublicInterface: nic2
        # The default route for the Overcloud nodes.
        # Example: 192.168.24.1
        ControlPlaneDefaultRoute: <DEFAULT_ROUTE_IP_ADDRESS>
        EC2MetadataIp: <UNDERCLOUD_LOCAL_IP>
        DeployedServerPortMap:
          <CONTROLLER0_SHORT_HOSTNAME>-ctlplane:
            fixed_ips:
              - ip_address: <CONTROLLER0_IPV4>
            subnets:
              # Example = 192.168.24.0/24
              - cidr: <NETWORK_ADDRESS>/<PREFIX>
            network:
              tags:
                # Example = 192.168.24.0/24
                - <NETWORK_ADDRESS>/<PREFIX>
          <CONTROLLER1_SHORT_HOSTNAME>-ctlplane:
            fixed_ips:
              - ip_address: <CONTROLLER1_IPV4>
            subnets:
              - cidr: <NETWORK_ADDRESS>/<PREFIX>
            network:
              tags:
                - <NETWORK_ADDRESS>/<PREFIX>
          <CONTROLLER2_SHORT_HOSTNAME>-ctlplane:
            fixed_ips:
              - ip_address: <CONTROLLER2_IPV4>
            subnets:
              - cidr: <NETWORK_ADDRESS>/<PREFIX>
            network:
              tags:
                - <NETWORK_ADDRESS>/<PREFIX>
          <COMPUTE0_SHORT_HOSTNAME>-ctlplane:
            fixed_ips:
              - ip_address: <COMPUTE0_IPV4>
            subnets:
              - cidr: <NETWORK_ADDRESS>/<PREFIX>
            network:
              tags:
                - <NETWORK_ADDRESS>/<PREFIX>
          <COMPUTE1_SHORT_HOSTNAME>-ctlplane:
            fixed_ips:
              - ip_address: <COMPUTE1_IPV4>
            subnets:
              - cidr: <NETWORK_ADDRESS>/<PREFIX>
            network:
              tags:
                - <NETWORK_ADDRESS>/<PREFIX>
      
  • Scenario 2: Use a custom network (not on the Undercloud control plane).

    • The Undercloud must be configured to use a public host for API communication during provisioning. The only way to do that, for security reasons, is to enable a TLS certificate.

      • Set the undercloud_public_host in the undercloud.conf to an IP address or hostname that will be accessible by the Overcloud control plane IP addresses.

      • Create a YAML file with the Puppet Hiera data that forces the deployment to use the public API endpoint on the Undercloud instead of the internal one. Set the hieradata_override value to the file path of that YAML file in the undercloud.conf.

        ---
        heat_clients_endpoint_type: public
        heat::engine::default_deployment_signal_transport: TEMP_URL_SIGNAL
        
      • Set the generate_service_certificate to true in the undercloud.conf. This will generate a self-signed certificate.

      • Load the new Undercloud configuration by re-running openstack undercloud install.

    • Set a custom control plane virtual IP that will be used by the HAProxy load balancer.

      ---
      resource_registry:
        OS::TripleO::ControllerDeployedServer::Net::SoftwareConfig: net-config-static-bridge.yaml
        OS::TripleO::ComputeDeployedServer::Net::SoftwareConfig: net-config-static-bridge.yaml
        # These resources will allow for a custom control plane virtual IP to be used for controller node services.
        OS::TripleO::DeployedServer::ControlPlanePort: /usr/share/openstack-tripleo-heat-templates/deployed-server/deployed-neutron-port.yaml
        OS::TripleO::Network::Ports::ControlPlaneVipPort: /usr/share/openstack-tripleo-heat-templates/deployed-server/deployed-neutron-port.yaml
        OS::TripleO::Network::Ports::RedisVipPort: /usr/share/openstack-tripleo-heat-templates/network/ports/noop.yaml
        OS::TripleO::Network::Ports::OVNDBsVipPort: /usr/share/openstack-tripleo-heat-templates/network/ports/noop.yaml
      
      parameter_defaults:
        NeutronPublicInterface: <NIC>
        ControlPlaneDefaultRoute: <DEFAULT_ROUTE_IP_ADDRESS>
        EC2MetadataIp: <UNDERCLOUD_PUBLIC_HOST>
        DeployedServerPortMap:
          control_virtual_ip:
            fixed_ips:
              # This IP must be accessible by all of the Overcloud nodes and should be on the same network.
              # It must also must be a unique IP address and not conflict with any other IP addresses.
              - ip_address: <CONTROL_VIRTUAL_IP_ADDRESS>
            subnets:
              - cidr: <NETWORK_ADDRESS>/<PREFIX>
            network:
              tags:
                - <NETWORK_ADDRESS>/<PREFIX>
          <CONTROLLER0_SHORT_HOSTNAME>-ctlplane:
            fixed_ips:
              - ip_address: <CONTROLLER0_IPV4>
            subnets:
              # Example = 192.168.122.0/24
              - cidr: <NETWORK_ADDRESS>/<PREFIX>
            network:
              tags:
                # Example = 192.168.122.0/24
                - <NETWORK_ADDRESS>/<PREFIX>
          <COMPUTE0_SHORT_HOSTNAME>-ctlplane:
            fixed_ips:
              - ip_address: <COMPUTE0_IPV4>
            subnets:
              - cidr: <NETWORK_ADDRESS>/<PREFIX>
            network:
              tags:
                - <NETWORK_ADDRESS>/<PREFIX>
      
  • If config-download will be used, hostname maps have to be defined. These must be mapped to the short hostname of the servers that relate to the port mappings.

    ---
    parameter_defaults:
      HostnameMap:
        overcloud-controller-0: <CONTROLLER0_SHORT_HOSTNAME>
        overcloud-controller-1: <CONTROLLER1_SHORT_HOSTNAME>
        overcloud-controller-2: <CONTROLLER2_SHORT_HOSTNAME>
        overcloud-novacompute-0: <COMPUTE0_SHORT_HOSTNAME>
        overcloud-novacompute-1: <COMPUTE1_SHORT_HOSTNAME>
    
  • Start the deployment of the Overcloud using at least these arguments and templates. The Heat templates defining the hostname and port maps must also be included.

    • <= Stein:

      $ openstack overcloud deploy --disable-validations --templates ~/templates \
          -e ~/templates/environments/deployed-server-environment.yaml \
          -e ~/templates/environments/deployed-server-bootstrap-environment-[centos|rhel].yaml \
          -e ~/templates/environments/deployed-server-pacemaker-environment.yaml \
          -r /usr/share/openstack-tripleo-heat-templates/deployed-server/deployed-server-roles-data.yaml
      
    • >= Train (the default composable roles data can now be used):

      $ openstack overcloud deploy --disable-validations --templates ~/templates \
          -e ~/templates/environments/deployed-server-environment.yaml \
          -r /usr/share/openstack-tripleo-heat-templates/roles_data.yaml
      
    • >= Victoria:

      $ openstack overcloud deploy --disable-validations --templates ~/templates \
          --deployed-server
      

config-download (>= Rocky)

No further action is required.

config-download (Queens)

Add the --config-download -e ~/templates/environments/config-download-environment.yaml template after (not before) the predeployed server templates to properly enable config-download.

os-collect-config (Queens, Automatic)

  • When using Queens without config-download, the deployment will pause on the creation of the Overcloud nodes. The Heat agent on the Overcloud nodes need to be registered for the deployment to continue. For new deployments only (not scaling), automatic detection of the Heat agents can be used. Use the Overcloud node roles defined in the “roles_data.yaml” configuration file.

    2019-01-01 12:00:00Z [overcloud.Compute.0.NovaCompute]: CREATE_IN_PROGRESS  state changed
    2019-01-01 12:00:01Z [overcloud.Controller.0.Controller]: CREATE_IN_PROGRESS  state changed
    
  • Then run the get-occ-config script on the Undercloud to configure the service.

    $ export OVERCLOUD_ROLES="ControllerDeployedServer ComputeDeployedServer"
    $ export ControllerDeployedServer_hosts="<CONTROLLER0_IP> <CONTROLLER1_IP> <CONTROLLER2_IP>"
    $ export ComputeDeployedServer_hosts="<COMPUTE0_IP> <COMPUTE1_IP>"
    $ /usr/share/openstack-tripleo-heat-templates/deployed-server/scripts/get-occ-config.sh
    

os-collect-config (Queens, Manual)

  • Use the manual method if the automatic one does not work.

  • Generate metadata URLs for the Overcloud nodes.

    $ for STACK in $(openstack stack resource list -n5 --filter name=deployed-server -c stack_name -f value overcloud) ; do STACKID=$(echo $STACK | cut -d '-' -f2,4 --output-delimiter " ") ; echo "== Metadata URL for $STACKID ==" ; openstack stack resource metadata $STACK deployed-server | jq -r '.["os-collect-config"].request.metadata_url' ; echo ; done
    
  • On the Overcloud nodes, add the correct metadata URL to the os-collect-config configuration, and then restart the service.

    $ sudo rm /usr/libexec/os-apply-config/templates/etc/os-collect-config.conf
    $ sudo vi /usr/libexec/os-apply-config/templates/etc/os-collect-config.conf
    
    [DEFAULT]
    collectors=request
    command=os-refresh-config
    polling_interval=30
    
    [request]
    metadata_url=<METADATA_URL>
    
    $ sudo systemctl restart os-collect-config
    
  • If issues are encountered with the manual process, stop the service and then run the os-collect-config command and force it to use the primary configuration file.

$ sudo rm -rf /var/lib/heat-config/deployed/*
$ sudo systemctl stop os-collect-config
$ sudo os-collect-config --debug --force --one-time --config-file /etc/os-collect-config.conf

[36][37]

Operations

Updates (Minor)

Minor updates keep the cloud at the same major OpenStack version. These provide both bug and security fixes.

Undercloud:

  • Update the containers prepare parameters to use the new containers. The template that defines those values should be used for both the Undercloud and Overcloud.

  • Update the Undercloud.

$ openstack undercloud upgrade

Overcloud:

  • Regenerate the Heat templates if they were also manually generated for the initial deployment.

$ cd /usr/share/openstack-tripleo-heat-templates/
$ ./tools/process-templates.py -o ~/templates/
  • Update the Heat stack configuration using the same arguments from openstack overcloud deploy. This will disable tasks that should not run during an update.

$ opentack overcloud update prepare <OVERCLOUD_DEPLOYMENT_ARGUMENTS>
  • Run the update_tasks and post_update_tasks from config-download.

$ openstack overcloud update run
  • Re-enable the tasks that were disabled by the prepare step.

$ opentack overcloud upgrade converge <OVERCLOUD_DEPLOYMENT_ARGUMENTS>

[68]

Upgrades (Major)

Both the Undercloud and Overcloud must first be updated to the latest minor release before attempting an upgrade. The upgrade process is very similar to the update process.

Undercloud:

  • Update the containers prepare parameters to use the new containers. The template that defines those values should be used for both the Undercloud and Overcloud.

  • Upgrade the Undercloud.

$ openstack undercloud upgrade

Overcloud:

  • Regenerate the Heat templates if they were also manually generated for the initial deployment.

$ cd /usr/share/openstack-tripleo-heat-templates/
$ ./tools/process-templates.py -o ~/templates/
  • Update the Heat stack configuration using the same arguments from openstack overcloud deploy. This will disable tasks that should not run during an upgrade.

$ opentack overcloud upgrade prepare <OVERCLOUD_DEPLOYMENT_ARGUMENTS>
  • Upgrade each Controller node, one at a time.

$ openstack overcloud upgrade run --limit overcloud-controller-0
$ openstack overcloud upgrade run --limit overcloud-controller-1
$ openstack overcloud upgrade run --limit overcloud-controller-2
  • Upgrade all of the Compude nodes.

$ openstack overcloud upgrade run --limit Compute
  • Re-enable the tasks that were disabled by the prepare step.

$ opentack overcloud upgrade converge <OVERCLOUD_DEPLOYMENT_ARGUMENTS>

[69]

Add a Compute Node

  • From the Undercloud, create a instackenv.json file describing the new node. Import the file using Ironic.

$ source ~/stackrc
$ openstack baremetal import --json ~/instackenv.json
  • Automatically configure it to use the existing kernel and ramdisk for PXE booting.

$ openstack baremetal configure boot
  • Set the new node to the “manageable” state. Then introspect the new node so Ironic can automatically determine it’s resources and hardware information.

$ openstack baremetal node manage <NODE_UUID>
$ openstack overcloud node introspect <NODE_UUID> --provided
  • Configure the node to be a compute node.

$ openstack baremetal node set --property capabilities='profile:compute,boot_option:local' <NODE_UUID>
  • Update the compute node scale using a Heat template.

---
parameter_defaults:
  ComputeCount: <NEW_COMPUTE_COUNT>
  • Redeploy the Overcloud while specifying the number of compute nodes that should exist in total after it is complete. The ComputeCount parameter in the Heat templates should also be increased to reflect it’s new value.

$ openstack overcloud deploy --templates ~/templates <DEPLOYMENT_OPTIONS>

[19]

Remove a Compute Node

Disable the Nova services.

$ . ~/overcloudrc
$ openstack compute service set <NODE> nova-compute --disable

Delete the Compute node and include the templates used during deployment. [49]

$ . ~/strackrc
$ openstack overcloud node delete --stack overcloud --templates ~/templates <NODE>

Delete additional services related to the Compute node.

$ . ~/overcloudrc
$ openstack compute service delete <NODE>
$ openstack network agent delete <NODE>
$ openstack resource provider delete <NDOE>

Decrease the ComputeCount in the Heat parameters used for the deployment.

[50]

Rebooting the Cloud

Servers hosting the cloud services will eventually need to go through a reboot to load up the latest patches for kernels, glibc, and other vital system components. This is the order in which servers should be restarted, one node at a time.

  • Undercloud

  • Controller

    • Stop clustered services on a controller node before rebooting.

      $ sudo pcs cluster stop
      
    • Reconnect to the clustered services after the reboot.

      $ sudo pcs cluster start
      
  • Ceph

    • Disable rebalancing before rebooting.

      $ sudo ceph osd set noout
      $ sudo ceph osd set norebalance
      
    • Enable rebalancing after all of the nodes are back online.

      $ sudo ceph osd unset noout
      $ sudo ceph osd unset norebalance
      
  • Compute

    • Disallow new instances from spawning on a specific compute node.

      $ openstack compute service list
      $ openstack compute service set <COMPUTE_HOST> nova-compute --disable
      
    • Live migrate all instances off of that compute node.

      $ nova host-evacuate-live <COMPUTE_HOST>
      
    • Verify that all instances have been migrated off before rebooting.

      $ openstack server list --host <COMPUTE_HOST> --all-projects
      

[34]

Ansible Playbooks (config-download)

The Queens release of TripleO featured optional usage of Ansible configuration management via a feature called config-download. It has been the default method of deployment since Rocky where it also added official support for deploying Ceph and Octavia. TripleO will log into the Overcloud nodes and configure a tripleo-admin user that will be used by Ansible for running updates and upgrades [39]. Use these arguments to enable config-download on Queens.

$ openstack overcloud deploy --templates ~/templates --config-download -e /usr/share/openstack-tripleo-heat-templates/environments/config-download-environment.yaml --overcloud-ssh-user heat-admin --overcloud-ssh-key ~/.ssh/id_rsa

In Queens, for reverting back to using Heat for the deployment, remove the config-download arguments and include an environment file with these resource registries [56]:

---
resource_registry:
  OS::TripleO::SoftwareDeployment: OS::Heat::StructuredDeployment
  OS::TripleO::DeploymentSteps: OS::Heat::StructuredDeploymentGroup
  OS::Heat::SoftwareDeployment:  OS::TripleO::Heat::SoftwareDeployment
  OS::Heat::StructuredDeployment: OS::TripleO::Heat::StructuredDeployment

The latest playbooks and variables used to deploy the Overcloud can be downloaded to the current working directory.

$ openstack overcloud config download

All of that Ansible content is stored in a local git repository at /var/lib/mistral/overcloud/. The log files of the last config-download run are found at /var/lib/mistral/overcloud/ansible.log and /var/lib/mistral/overcloud/ansible-errors.json.

The deploy_steps_playbook.yaml file is the primary playbook that executes all of the deployment playbooks. Before running the playbook, the tripleo-admin account needs to be configured on the Overcloud nodes. This can be done manually if the playbooks for the deployment or scale-up are used manually (ex., not using openstack overcloud deploy) [36]:

  • Queens:

    $ export OVERCLOUD_HOSTS="<IP1> <IP2>"
    $ /usr/share/openstack-tripleo-heat-templates/deployed-server/scripts/enable-ssh-admin.sh
    
  • Train:

    $ openstack overcloud admin authorize
    

A static inventory can be created using the available dynamic inventory script tripleo-ansible-inventory.

$ tripleo-ansible-inventory --ansible_ssh_user tripleo-admin --static-yaml-inventory tripleo-ansible-inventory.yaml

Tags (as of Stein):

  • always

  • facts

  • common_roles

  • container_config

  • container_config_scripts

  • container_config_tasks

  • container_image_prepare

  • container_startup_configs

  • external_deploy_steps

  • external_post_deploy_steps

  • host_config

  • host_prep_steps

  • overcloud

  • pre_deploy_steps

  • step0

  • step1

  • step2

  • step3

  • step4

  • step5

  • tripleo_ssh_known_hosts

For only updating the Ansible playbooks based on the Heat templates, pass the --stack-only argument to the Overcloud deployment. They can then be downloaded and executed manually.

$ openstack overcloud deploy --stack-only

[41]

If the playbooks are already generated from a successful STACK_CREATE of the Overcloud, then the deployment can be ran again using only the playbooks (skipping the need to parse the Heat templates).

$ openstack overcloud deploy --config-download-only

Fact caching is enabled by default which can lead to issues with re-deployment. This can be manually cleared out on the Undercloud.

$ sudo rm -rf /var/lib/mistral/ansible_fact_cache/*

Force re-running tasks that only run during the initial deployment by using the force=true variable. The example below will run the network configuration tasks again.

$ ansible-playbook -i inventory.yaml --become --tags facts,post_deploy_steps deploy_steps_playbook.yaml -e force=true

Manual

The deployment of OpenStack services can be done manually outside of TripleO. The Ansible playbooks are generated, the inventory file is created, SSH keys are configured, and then the playbooks can be run to do the actual deployment. This is useful for debugging the deployment or running it on a server that is not the Undercloud.

  • Generate the Ansible playbooks by rendering the Overcloud Heat stack.

    $ openstack overcloud deploy --stack-only ...
    
  • Download the deployment playbooks. Then change into the directory it creates.

    $ openstack overcloud config download
    $ cd /home/stack/tripleo-config/overcloud/
    
  • Generate a static inventory file.

    $ tripleo-ansible-inventory --static-yaml-inventory tripleo-ansible-inventory.yaml
    
  • Copy the private key used to access the “tripleo-admin” user on the Overcloud.

    $ sudo cp /var/lib/mistral/overcloud/ssh_private_key ./
    $ sudo chown stack.stack ssh_private_key
    
  • Setup the authorized SSH keys for the “tripleo-admin” user on the Overcloud nodes.

    $ openstack overcloud admin authorize
    
  • Run the Overcloud deployment. The “deploy_steps_playbook.yaml” is the primary playbook that combines all of the other playbooks into one.

    $ ansible-playbook --inventory tripleo-ansible-inventory.yaml --key-file ssh_private_key --become deploy_steps_playbook.yaml
    

[41]

Configurations

These are configurations specific to Overcloud deployments using TripleO. Custom settings are defined using a YAML Heat template.

---
parameter_defaults:
  <KEY>: <VALUE>

Composable Roles

Roles in TripleO (not to be confused with Ansible roles) are used to define what Linux and OpenStack services will be configured on an Overcloud node. Each Overcloud node is assigned a role based on profile tagging.

View the default roles.

$ less /usr/share/openstack-tripleo-heat-templates/roles_data.yaml

View the available roles and see what services are set for each role.

$ openstack overcloud roles list
$ openstack overcloud roles show <ROLE>
$ ls -1 /usr/share/openstack-tripleo-heat-templates/roles/
$ less /usr/share/openstack-tripleo-heat-templates/roles/<ROLE>.yaml

Create a roles_data file that contains only the roles that will be used for deployment.

$ openstack overcloud roles generate <ROLE1> <ROLE2> > roles_data_custom.yaml

Use a specified roles_data file to generate a new set of TripleO Heat Templates (THT) based on the roles that are defined.

$ cd /usr/share/openstack-tripleo-heat-templates/
$ ./tools/process-templates.py -r roles_data.yaml -o ~/templates/

Services can be disabled from being deployed and configured on the Overcloud one of two ways.

  1. Remove the service entry from the relevant role in the roles_data.yaml. Then run process-templates.py again to regenerate the TripleO Heat Templates.

  2. Create a new Heat template file and map the service to OS::Heat::None.

---
resource_registry:
  OS::TripleO::Services::<SERVICE>: OS::Heat::None

Most services are mapped to a THT deployment template in /usr/share/openstack-tripleo-heat-templates/deployment/. Service deployment templates are named *-ansible.yaml or *-puppet.yaml based on what configuration management is used to deploy the service.

Even if a role has a service listed, the default may be set to have it be disabled. To re-enable the service, it must be mapped to the deployment template. A grep can help find the related mapping.

Syntax:

$ grep -r OS::TripleO::Services::<SERVICE>: /usr/share/openstack-tripleo-heat-templates/ | grep -v OS::Heat::None
---
resource_registry:
  OS::TripleO::Services::<SERVICE>: <DEPLOYMENT_TEMPLATE>

Example:

$ grep -r OS::TripleO::Services::MySQL: /usr/share/openstack-tripleo-heat-templates/ | grep -v OS::Heat::None
---
resource_registry:
  OS::TripleO::Services::MySQL: deployment/database/mysql-container-puppet.yaml

Additional service defaults are set in these files:

  • Undercloud: /usr/share/openstack-tripleo-heat-templates/environments/undercloud.yaml

  • Overcloud: /usr/share/openstack-tripleo-heat-templates/overcloud-resource-registry-puppet.j2.yaml

[63][64]

Some variables, such as RhsmVars, can be set on a per-role basis. These will override defaults values set by the service templates. These can be set in the composable role definition file or a Heat environment template.

Role:

---
- name: <COMPOSABLE_ROLE_NAME>
  RoleParametersDefault:
    <KEY>: <VALUE>

Heat:

---
parameter_defaults:
  <COMPOSABLE_ROLE_NAME>Parameters:
    <KEY>: <VALUE>

OpenStack Services

Configuration options for OpenStack services can be defined using ExtraConfig.

  • AllNodesExtraConfig or ExtraConfig = Apply the settings to all nodes.

  • <ROLE>ExtraConfig = Apply the settings to all nodes deployed using the composable role.

There are different deployment hooks used for configuration.

  • ExtraConfigPre = Before Puppet.

  • ExtraConfig = During Puppet. Use this one for changing Puppet settings.

  • ExtraConfigPost = After Puppet.

The configuration for OpenStack services are handled by Puppet (not Ansible).

$ grep -r <VARIABLE> /usr/share/openstack-tripleo-heat-templates/deployment/*/*-puppet.yaml
$ grep -r <VARIABLE> /usr/share/openstack-puppet/modules/

The variables defined in TripleO Heat templates will be rendered as Puppet Hieradata to this location inside the OpenStack service’s container and then applied by Puppet:

/var/lib/config-data/<OPENSTACK_SERVICE>/etc/puppet/hieradata/service_configs.json

This is the order in how to attempt modifying a variable. If it is not possible, then try the next one down. The deployment will fail if there is a duplicate declaration of a variable.

  1. Use a Heat parameter if exposed via a deployment template

  2. Use the Puppet class to override a value.

  3. Use the generic Puppet config class to manually override settings not exposed by Puppet.

Not all of the Puppet variables for OpenStack service configuration are exposed as Heat parameters. These can still be manually set. Puppet manifests define the default variables that are set. These also show what Puppet dictionary variables are used for each configuration.

All of the service manifests can be found here: /usr/share/openstack-puppet/modules/${OPENSTACK_SERVICE}/manifests/. The OpenStack services on OpenDev.org each have a related puppet-<SERVICE> repository that hosts the Puppet manifests. Miscellaneous service configuration are grouped into the puppet-tripleo project. Other Puppet dependencies and default settings are provided by the puppet-openstacklib project.

The Puppet class for a service is typically named using the convention <OPENSTACK_SERVICE>::<MANIFEST>. The actual class name will be listed in the manifest files. Below shows how awk can be used to extract the class names along with the exposed variables that can be modified.

$ cd /usr/share/openstack-puppet/modules/${OPENSTACK_SERVICE}/manifests/
$ awk '/^class/,/)/' ./*.pp # Search the top-level directory.
$ awk '/^class/,/)/' ./*/*.pp # Search in all of the sub-directories.

Once the correct class and variable are found, the setting can be defined in a Heat template.

Syntax:

---
parameter_defaults:
  <ROLE>ExtraConfig:
    # The primary manifest handles at least the primary configuration file.
    <PUPPET_CLASS>::<VARIABLE>:: <VALUE>

Example:

---
parameter_defaults:
  # Only apply this configuration to Overcloud nodes deployed using the "Controller" role.
  ControllerExtraConfig:
    # The class name is "keystone::wsgi::apache".
    # The variable exposed by the class is "workers".
    # The value is overridden and set to "4" (instead of the default "$::os_workers_keystone").
    keystone::wsgi::apache::workers: 4

Settings that are not handled by Puppet resources can be overridden manually. The dictionary name for each configuration file is defined in manifests/config.pp in the <OPENSTACK_SERVICE>::config class. For the main configuration file, the naming convention for the variable to set is <OPENSTACK_SERVICE>::config::<OPENSTACK_SERVICE>_config.

Syntax:

---
parameter_defaults:
  <ROLE>ExtraConfig:
    <PUPPET_CLASS>::<VARIABLE>:
      '<SECTION>/<KEY>':
        value: '<VALUE>'

Example:

---
parameter_defaults:
  UndercloudExtraConfig:
    heat::config::heat_config:
      'DEFAULT/stack_action_timeout':
        value: '7200'
    heat::config::heat_api_paste_ini:
      'filter:authtoken/admin_user':
        value: 'heat2'

[32]

The root MySQL account password can be configured for the Undercloud and/or Overcloud.

Undercloud:

# undercloud.conf
[auth]
undercloud_db_password=<PASSWORD>

Overcloud:

---
parameter_defaults:
  MysqlRootPassword: <PASSWORD>

Networks

When no network template is defined, VLANs are not used and instead each network will be assigned different subnets. Networks are only created using the STACK_CREATE phase and will not run during the STACK_UPDATE phase unless the Heat parameter NetworkDeploymentActions: ['CREATE','UPDATE'] is set.

Interfaces (os-net-config)

os-net-config is developed as part of TripleO and used to configure the network interfaces, DNS nameservers, IP addresses, and routes on all nodes (Undercloud and Overcloud).

Render the TripleO Heat Templates (THT) to view the static net-config files. These provide different layouts and examples for how to configure the networking interfaces.

$ cd /usr/share/openstack-tripleo-heat-templates/
$ mkdir ~/templates/
$ /usr/share/openstack-tripleo-heat-templates/tools/process-templates.py -o ~/templates/
$ cd ~/templates/
$ ls -1 net-config-*
net-config-bond.yaml
net-config-bridge.yaml
net-config-linux-bridge.yaml
net-config-noop.yaml
net-config-standalone.yaml
net-config-static-bridge-nic-two-only.yaml
net-config-static-bridge-two-nics.yaml
net-config-static-bridge-with-external-dhcp.yaml
net-config-static-bridge.yaml
net-config-static.yaml
net-config-undercloud.yaml

Set a custom net-config file on a per-role basis by overriding the resource registry for network configuration.

Syntax:

---
resource_registry:
  OS::TripleO::<ROLE>::Net::SoftwareConfig: <PATH_TO>/<NIC_CONFIG_TEMPLATE>.yaml

Example:

---
resource_registry:
  OS::TripleO::Controller::Net::SoftwareConfig: net-config-bond.yaml

A network environment template can be used to set related TripleO-provided net-config settings for all roles.

$ ls -1 environments/net-*
environments/net-2-linux-bonds-with-vlans.yaml
environments/net-bond-with-vlans-no-external.yaml
environments/net-bond-with-vlans.yaml
environments/net-dpdkbond-with-vlans.yaml
environments/net-multiple-nics-vlans.yaml
environments/net-multiple-nics.yaml
environments/net-noop.yaml
environments/net-single-nic-linux-bridge-with-vlans.yaml
environments/net-single-nic-with-vlans-no-external.yaml
environments/net-single-nic-with-vlans.yaml
$ openstack overcloud deploy -e ~/templates/environments/<NETWORK_ENVIRONMENT>.yaml

The $network_config dictionary stores the entire os-net-config configuration. The run-os-net-config.sh script will find and replace all references to interface_name with the Heat parameter value for NeutronPublicInterface and also replaces bridge_name with NeutronPhysicalBridge. The script will default any interface not defined in the os-net-config settings to use DHCP. If DHCP does not work, then the “network” service may fail to restart during the Overcloud deployment leading to an inaccessible Overcloud.

net-config THT template:

resources:
  OsNetConfigImpl:
    type: OS::Heat::SoftwareConfig
    properties:
      group: script
      config:
        str_replace:
          template:
            get_file: network/scripts/run-os-net-config.sh
          params:
            $network_config:

The configuration file is stored on every node at /etc/os-net-config/config.yaml. Settings from a custom file can be manually applied for testing by running os-net-config -c <OS_NET_CONFIG_FILE>.yaml -v --detailed-exit-codes --cleanup.

Every network object that can be managed is known as a type. Common types include: interface, ovs_bond, ovs_bridge, route_rule, team, and vlan. The full list of valid parameters are listed in the schema.yaml file of os-net-config.

The interface type accepts passing nic1 (eth0), nic2 (eth1), etc. as the name attribute for dynamically associating an interface. Alternatively, the actual name of the network interface, such as eth0 or eth1, can be defined.

Below are sample configurations that can be defined in a net-config THT template. They will render out the Heat parameters during the deployment.

DHCP:

$network_config:
  network_config:
    - type: interface
      name: nic1
      use_dhcp: true

Static:

$network_config:
  network_config:
    - type: interface
      name: network_interface
      addresses:
        - ip_netmask: 192.168.122.101/24
      dns_servers:
        get_param: DnsServers
      domain:
        get_param: DnsSearchDomains

Control Plane IP Address:

$network_config:
  network_config:
    - type: interface
      name: eth3
      addresses:
        - ip_netmask:
            list_join:
              - /
              - - get_param: ControlPlaneIp
                - get_param: ControlPlaneSubnetCidr
      routes:
        - default: true
          next_hop:
            get_param: ControlPlaneDefaultRoute

Bridge on a Bond:

$network_config:
  network_config:
    - type: ovs_bridge
      name: bridge_interface
      use_dhcp: true
      members:
        - type: ovs_bond
          name: bond0
          ovs_options:
            get_param: BondInterfaceOvsOptions
          members:
            - type: interface
              name: nic1
            - type: interface
              name: nic2

[61]

Only Open vSwitch (OVS) and Open Virtual Networking (OVN) are supported for bridge types. Linux Bridge is not tested in CI nor supported by Red Hat. RHEL 8 removed support for the legacy bridge utilities. [62]

VLANs

There are 6 different types of networks in a standard TripleO deployment using a VLANs template.

  • External = The external network that can access the Internet. This is used for the Horizon dashboard, public API endpoints, and floating IP addresses. Default VLAN: 10.

  • Internal = Default VLAN: 20.

  • Storage = Default VLAN: 30.

  • StorageMgmt = Default VLAN: 40

  • Tenant = Default VLAN: 50

  • Management = Default VLAN: 60.

The VLANs need to be trunked on the switch. A 7th native VLAN should also be configured on the switch for the provisioning network.

IP Addressing

Configure the network CIDRs, IP address ranges to allocation, and VLAN tags.

<NETWORK_TYPE>NetCidr: <CIDR>
<NETWORK_TYPE>AllocationPools:
  - start: <START_IP>
    end: <END_IP>
<NETWORK_TYPE>NetworkVlanID: <VLAN_ID>

Configure these settings to match the IP address that the Undercloud is configured to use for provisioning. The default value is 192.168.24.1.

ControlPlaneSubnetCidr: '24'
ControlPlaneDefaultRoute: <UNDERCLOUD_IP_OR_ROUTER>
EC2MetadataIp: <UNDERCLOUD_IP>

Public

Configure the Overcloud access to the public Internet. Define the default router for the External network, DNS resolvers, and the NTP servers. It is important the DNS is setup correctly because if chronyc fails to resolve the NTP servers then it will not try to resolve them again. DNS is also required to download and install additional TripleO packages.

ExternalInterfaceDefaultRoute: <PUBLIC_DEFAULT_GATEWAY_ADDRESS>
DnsServers:
  - 10.5.30.160
  - 10.11.5.19
NtpServer:
  - clock.redhat.com
  - clock2.redhat.com

Define the allowed network tag/tunnel types that Neutron networks use. The Neutron tunnel type is used for internal transmissions between the compute and network nodes. By default, the Neutron network bridge will be attached to br-int if left blank. This will configure a provider network. Otherwise, br-ex should be specified for self-service networks.

NeutronNetworkType: "vxlan,gre,vlan,flat"
NeutronTunnelTypes: "vxlan"
NeutronExternalNetworkBridge: "''"

Define the interface to use for public networks. The NeutronPublicInterface (nic1/eth0 by default) will be converted into the an Open vSwitch bridge named based on the value of NeutronPhysicalBridge (br-ex by default). Optionally, define a VLAN tag for it. If no IP addressing information is configured for this interface, TripleO will default to configuring DHCP.

NeutronPublicInterface: eth1
NeutronPhysicalBridge: br0
NeutronPublicInterfaceTag: 100

Configure bonding interface options, if applicable. Below is an example for LACP.

bonding_options: "mode=802.3ad lacp_rate=slow updelay=1000 miimon=100"

Configure the bridge that will be used for public routers and floating IPs. Map it to a user-friendly name that will be used by Neutron resources.

NeutronPhysicalBridge: br-ctlplane
NeutronBridgeMappings: datacentre:br-ctlplane

[31]

Containers

  • Configure the Undercloud to cache container images and serve them to the Overcloud nodes. This caching speeds up the deployment and lowers the amount of Internet bandwidth used. By default, Overcloud nodes will directly get images from the defined public registries. A private registry on the Undercloud will need to be configured as an insecure registry (it does not use a SSL/TLS certificate by default).

# undercloud.conf
[DEFAULT]
container_images_file = /home/stack/containers-prepare-parameter.yml
container_insecure_registries = ['192.168.24.1:8787']
---
# containers-prepare-parameters.yml
parameter_defaults:
  DockerInsecureRegistryAddress:
    - 192.168.24.1:8787
    - undercloud.ctlplane.localdomain:8787
  ContainerImagePrepare:
    - push_destination: true
  • Authenticate with a registry. For example, the Red Hat repository that contains the RHOSP container images. [42]

---
parameter_defaults:
ContainerImageRegistryCredentials:
  registry.redhat.io:
    <RED_HAT_USERNAME>: <RED_HAT_PASSWORD>
  • Information on how to define custom registries, set container names, version tags to use, and other related settings can be found here.

  • Custom package repositories and RPMs installed in containers are handled by the tripleo-modify-image Ansible role.

ContainerImagePrepare:
  - push_destination: true
    includes:
      - <CONTAINER_NAME>
    modify_role: tripleo-modify-image
    modify_append_tag: "-hotfix"
    modify_vars:
      <TRIPLEO_MODIFY_IMAGES_ROLE_VARIABLES>

Packages

By default, TripleO will not install packages. The standard Overcloud image from RDO already has all of the OpenStack packages installed. When using a custom image or not using Ironic for deploying Overcloud nodes, packages can be configured to be installed.

EnablePackageInstall: true

A different repository for Overcloud service containers can be configured (>= Pike).

DockerNamespace: registry.example.tld/rocky
DockerNamespaceIsRegistry: true
DockerInsecureRegistryAddress: registry.example.tld
DockerNamespaceIsInsecureRegistry: true

Ceph

Releases

Ceph is fully supported as a back-end for Overcloud storage services. If Ceph is enabled in TripleO, it will be used by default for Glance and Cinder. Before Pike, puppet-ceph was used to manage Ceph. Experimental support for using ceph-ansible was added in Pike. [17] It is fully supported via config-download as of Rocky. In Train, it uses the same Ansible inventory as config-download. Ceph updates are handled during the external_deploy_steps_tasks stage of config-download.

Red Hat Ceph Storage (RHCS) is the supported enterprise version of Ceph. RHCS 3.2 added official support for BlueStore. Using Ceph’s FileStore mechanism has been deprecated since RHOSP 14. FileStore to BlueStore migration is supported by Red Hat. Customers must first update to RHCS 4 and then each OSD node is upgraded one at a time. [16]

RHCS releases and supported platforms:

  • RHCS 2 (Jewel) = RHOSP 10, 11, and 12.

  • RHCS 3 (Luminous) = RHOSP 13 and 14.

  • RHCS 4 (Nautilus) = RHOSP >= 15.

Deployment Types

TripleO can use an existing/independent external Ceph cluster. This is not managed by TripleO, and only provides connection details for OpenStack to communicate with the Ceph cluster. This requires the environments/ceph-ansible-external.yaml template. For a managed internal cluster, TripleO can deploy and manage the life-cycle of Ceph by using the environments/ceph-ansible.yaml template.

Packages

There are package and container requirements for both internal and external deployments of Ceph. The ceph-ansible package has to be installed for either the internal or external use case. For RHOSP, this is provided by the ceph-tools repository. As of Pike, the ceph-container has to be used to manage the Ceph services (even only as a client). This means that troubleshooting must be done inside the container. All OSD daemons run through a single container on each OSD node.

Architecture

TripleO puts the ceph-mon[itors] on the Overcloud Controller nodes. The OSDs are recommended to be placed on dedicated hardware. Hyperconverged infrastructure (HCI) is supported to run OSDs on the Compute nodes alongside the OpenStack services. For the Edge deployments, the ceph-mons live on the OSD nodes.

If the specified disks for deployment are clean, TripleO will create the LVMs required for the Ceph OSDs.

Pools for each OpenStack service are automatically created.

  • images = Glance

  • metrics = Gnocchi

  • backups = Cinder

  • vms = Nova

  • volumes = Cinder

One keyring at /etc/ceph/ceph.client.openstack.keyring is created by default to access all of the pool/rbds.

Deployment (Internal)

Use the environments/ceph-ansible.yaml Heat template. The command output of ceph-ansible is saved in the config-download directory at ceph-ansible/ceph-ansible-command.log.

~/templates/environments/ceph-ansible.yaml = Enables Ceph ~/ceph.yaml = Specify a custom file with your own overrides

Configure the object storage back-end: bluestore or filestore.

---
parameter_defaults:
  osd_objectstore: <BACKEND>

Example configuration of letting ceph-volume automatically determine which disks to use for what purpose (OSD or metadata). LVM is the recommended scenario.

---
parameter_defaults:
  CephAnsibleDisksConfig:
    devices:
      - /dev/sdb
      - /dev/sdc
      - /dev/nvme1n1
    osd_scenario: lvm
    osd_objectstore: bluestore

Manually created LVMs can also be defined to skip the usage of ceph-volume.

---
parameter_defaults:
  CephAnsibleDisksConfig:
    lvm_volumes:
      - data: data-lv2
        data_vg: vg2
        db: db-lv2
        db_vg: vg2
    osd_scenario: lvm
    osd_objectstore: bluestore

If the initial deployment of TripleO with internal Ceph fails, the storage devices used for Ceph should be cleaned. If the undercloud.conf has clean_nodes = True set then the cleaning will be done automatically when Ironic chances a node state from active to available or manageable to available.

Example of common settings for Ceph in RHOSP:

---
parameter_defaults:
  CephAnsiblePlaybookVerbosity: 3
  CephPoolDefaultSize: 1
  CephPoolDefaultPgNum: 32
  LocalCephAnsibleFetchDirectoryBackup: /tmp/fetch_dir
  CephAnsibleDisksConfig:
    osd_scenario: lvm
    osd_objectstore: bluestore
    lvm_volumes:
      - data: data-lv2
        data_vg: vg2
        db: db-lv2
        db_vg: vg2
  CephAnsibleExtraConfig:
    mon_host_v1:
      enabled: False
  # Required on RHOSP 15 until RHCS 4 becomes GA.
  EnableRhcs4Beta: true
  • CephAnsiblePlaybookVerbosity = If set to > 0, then the playbooks are kept (and the verbosity is enabled for the playbook).

  • CephAnsiblePoolDefaultSize = Set the replica size for each pool. Default: 3. Lab recommended: 1.

  • CephAnsibleDefaultPgNum = For a production environment, use PGCalc to determine the optimal value. Set to a low number for a lab with 1 disk. Lab recommended: 32.

  • CephAnsibleExtraConfig: mon_host_v1: enabled: false = Force msgr2 (messenger v2). By default, both v1 and v2 are used, which causes issues in lab environments such as Standalone.

Deployment (External)

Use the environments/ceph-ansible-external.yaml Heat template.

TripleO Queens:

---
parameter_defaults:
  NovaEnableRbdBackend: true
  CinderEnableRbdBackend: true
  CinderBackupBackend: ceph
  GlanceBackend: rbd
  GnocchiBackend: rbd
  NovaRbdPoolName: vms
  CinderRbdPoolName: volumes
  CinderBackupRbdPoolName: backups
  GlanceRbdPoolName: images
  GnocchiRbdPoolName: metrics
  CephClientUserName: openstack
  CephClusterFSID: '<CLUSTER_FILE_SYSTEM_ID>'
  CephClientKey: '<CEPHX_USER_KEY>'
  CephExternalMonHost: '<CEPH_MONITOR_1>, <CEPH_MONITOR_2>, <CEPH_MONITOR_3>'

[29]

Overcloud (cloud-init)

Any Overcloud node that is provisioned and managed by Ironic and Nova can be configured using cloud-init configuration data.

---
parameter_defaults:
  OS::TripleO::NodeUserData: <CLOUD_INIT_CONFIG>.yml

Minions

Introduced in the Train release [47], the Undercloud can be scaled horizontally by using minion nodes to help with a large Overcloud deployment, update, or upgrade. This runs the “heat-engine” and “ironic-conductor” services on additional nodes. There are no limits to the number of minions that can be used. When not in use, minion nodes can be turned off. The only requirement is that all of the nodes are on the Control Plane network. The framework for the minion installer is based on the standalone installer.

  • Copy the required files from the Undercloud to the minions.

$ scp tripleo-undercloud-outputs.yaml tripleo-undercloud-passwords.yaml <USER>@<MINION_MACHINE>
  • Install the TripleO packages.

$ sudo yum install python3-tripleoclient

Configure the Minion node. The minion_local_ip and the minion_local_interface should be on the Overcloud control plane / provisioning network. The container_images_file should also use the same custom container-image-prepare.yaml file that the Undercloud uses (if applicable).

$ cp /usr/share/python-tripleoclient/minion.conf.sample ~/minion.conf

Install the Minion services.

$ openstack undercloud minion install
  • Verify the services are running as expected.

$ source ~/stackrc
$ openstack orchestration service list
$ openstack baremetal conductor list

[44]

Scaling (Large Overcloud)

RHOSP 13 (Queens) supports deploying 500 Overcloud nodes. The Undercloud needs resource allocations related directly to how many Overcloud nodes will be deployed: 1 CPU core and 8GB RAM per every 15 nodes. Undercloud services require a few other configuration tweaks to handle the large capacity. [66] Set the NodeCreateBatchSize Heat parameter to a value equal to the number of Overcloud nodes. This will greatly decrease the initial Heat template processing time.

Puppet

Puppet is used for configuring the OpenStack and operating system services.

  • Train = Puppet 5

  • Queens = Puppet 4

  • Newton = Puppet 3

Lab Tips

These are trips and tricks for setting up a full, yet basic, TripleO cloud for testing the deployment.

  • Use the Standalone deployment or at least the minimum amount of nodes required for TripleO: 1 Undercloud, 1 Controller, and 1 Compute.

  • Use the most minimally required resources for each node role. These values have been verified on the Train release.

    • Undercloud: 4 vCPUs and 8GB RAM

    • Controller: 2 vCPUs and 12GB RAM

    • Compute: 1 vCPU

      • Pre-deployed: 2GB RAM

      • Ironic provisioned: 4GB RAM

  • Enable swap. This is especially required for environments with limited RAM.

    ---
    resource_registry:
      OS::TripleO::AllNodesExtraConfig: extraconfig/all_nodes/swap.yaml
    parameter_defaults:
      swap_size_megabytes: '8192'
    
  • If using OpenStack as the lab infrastructure, disable port security to allow any MAC and IP address to be used. Also disable security groups to avoid further connection issues.

    $ openstack port set --disable-port-security --no-security-group <PORT>
    
  • Use the low resource usage template: environments/low-memory-usage.yaml. This sets the worker count to 1 for all of the OpenStack services, lowers the Apache resource utilization (used as the CGI handler for OpenStack services), and configures low defaults for (optional) Ceph services.

  • Avoid using complex network templates such as environments/network-isolation.yaml and environments/network-environment.yaml. By default, TripleO will use flat networking for all of the services and separate traffic using different subnets.

  • For virtual machines without nested virtualization, set the parameter NovaComputeLibvirtType to qemu.

  • Use this template (designed for Train) as a reference to prevent deploying unnecessary services on the Overcloud. That template will disable everything except Keystone. Alternatively, remove services from the roles_data.yaml file.

  • Disable Swift. Then use NFS as the back-end for Cinder, Glance, Gnocchi, and Nova based off of the configuration files from /usr/share/openstack-tripleo-heat-templates/environments/storage/*-nfs.yaml. [51] Add the NFS mount option ‘nosharecache’ to address this bug.

    ---
    # Disable Swift
    resource_registry:
      OS::TripleO::Services::SwiftProxy: OS::Heat::None
      OS::TripleO::Services::SwiftStorage: OS::Heat::None
      OS::TripleO::Services::SwiftRingBuilder: OS::Heat::None
    
    parameter_defaults:
      ControllerExtraConfig:
        tripleo::haproxy::swift_proxy_server: false
    
    ---
    # Enable NFS back-ends
    resource_registry:
      OS::TripleO::Services::CephMgr: ~/templates/docker/services/ceph-ansible/ceph-mgr.yaml
      OS::TripleO::Services::CephMon: ~/templates/docker/services/ceph-ansible/ceph-mon.yaml
      OS::TripleO::Services::CephOSD: ~/templates/docker/services/ceph-ansible/ceph-osd.yaml
      OS::TripleO::Services::CephClient: ~/templates/docker/services/ceph-ansible/ceph-client.yaml
    
    parameter_defaults:
      # Cinder
      CinderBackupBackend: nfs
      CinderEnableIscsiBackend: false
      CinderEnableNfsBackend: true
      CinderEnableRbdBackend: false
      # docker (Queens)
      #CinderNfsMountOptions: 'rw,sync,vers=4,minorversion=2,nosharecache,context=system_u:object_r:cinder_var_lib_t:s0'
      # Podman (Train)
      CinderNfsMountOptions: 'rw,sync,vers=4,minorversion=2,nosharecache,context=system_u:object_r:container_file_t:s0'
      CinderNfsServers: '<NFS_SERVER_IP>:/exports/cinder'
      # Glance
      GlanceBackend: file
      GlanceNfsEnabled: true
      # docker (Queens)
      #GlanceNfsOptions: 'rw,sync,vers=4,minorversion=2,nosharecache,context=system_u:object_r:glance_var_lib_t:s0'
      # Podman (Train)
      GlanceNfsOptions: 'rw,sync,vers=4,minorversion=2,nosharecache,context=system_u:object_r:container_file_t:s0'
      GlanceNfsShare: '<NFS_SERVER_IP>:/exports/glance'
      # Gnocchi
      GnocchiBackend: file
      # Nova
      NovaNfsEnabled: true
      NovaEnableRbdBackend: false
      # docker and Podman
      NovaNfsOptions: 'rw,sync,vers=4,minorversion=2,nosharecache,context=system_u:object_r:nfs_t:s0'
      NovaNfsShare: '<NFS_SERVER_IP>:/exports/nova'
    
    • TripleO will not create or manage an NFS server. If using the Undercloud as the NFS server, the firewall ports for the service will need to be opened.

      $ sudo iptables -I INPUT -p tcp -m tcp --dport 111 -j ACCEPT
      $ sudo iptables -I INPUT -p tcp -m tcp --dport 2049 -j ACCEPT
      $ sudo iptables -I INPUT -p udp -m udp --dport 111 -j ACCEPT
      $ sudo iptables -I INPUT -p udp -m udp --dport 2049 -j ACCEPT
      $ sudo iptables-save | sudo tee /etc/sysconfig/iptables
      

Troubleshooting

Service Telemetry Framework

The Red Hat Service Telemetry Framework (STF) provides a standardized way to log and monitor all of the services in RHOSP. This was previously known as the Service Assurance Framework (SAF). STF is hosted on the Red Hat OpenShift Container Platform (RHOCP).

Services:

  • RHOSP nodes:

    • ceilometer = Collects metrics and events from OpenStack services.

    • collectd = Collects metrics and events from the infrastructure and non-OpenStack services.

    • Red Hat AMQ = Sends metrics and events to AMQ Interconnect on RHOCP.

  • RHOCP operators:

    • AMQ Interconnect = Receives metrics and events from Red Hat AMQ on RHOSP. These are then sent to the Smart Gateway.

    • Smart Gateway = Routes metrics to Prometheus and events to Elasticsearch.

    • Prometheus = Stores all of the time-series metrics onto a persistent storage backend. Provides a dashboard and API to search the data.

    • AlertManager = Waits for an alert rule to be triggered by Prometheus. Once an alert is received, the AlertManager sends communication to the system administrators about possible issues.

    • Elasticsearch = Stores all of the events onto a persistent storage backend. Provides a dashboard and API to search the data.

  • Visual dashboard:

    • Grafana = Not officially part of the STF. Available as a useful community plugin to visualize the data.

[75]

Tips

  • Disable the Extra Packages for Enterprise Linux (EPEL) and Puppet Labs repositories if these are available. These will cause package conflicts and result in the installation of wrong dependencies.

  • If a deployment fails, view the config-download playbook errors: $ openstack overcloud failures list.

  • If highly-available (HA) services on the Controller nodes are stopped or not working, cleanup and restart the affected resources managed by Pacemaker.

$ sudo pcs status
$ sudo crm_resource -C <RESOURCE_BUNDLE>
$ sudo pcs resource restart <RESOURCE_BUNDLE>
$ sudo systemctl stop tripleo_<SERVICE>
$ sudo vim /var/lib/config-data/puppet-generated/<CONTAINER_NAME>/etc/foo/bar.conf
$ sudo vim /var/lib/config-data/puppet-generated/<CONTAINER_NAME>/etc/httpd/conf.d/10-<SERVICE>_wsgi.conf
$ sudo systemctl start tripleo_<SERVICE>
  • Puppet variables can be retrieved from Hieradata on both the Undercloud and Overcloud nodes. Example:

$ sudo hiera -c /etc/puppet/hiera.yaml mysql::server::root_password
  • Enable various parameters to assist with debugging.

    ---
    parameter_defaults:
      ContainerHealthcheckDisabled: true
      ContainerImagePrepareDebug: true
      # Enable 'Debug' logging for all OpenStack services.
      Debug: true
      SELinuxMode: permissive
    

Errors

No valid host was found” when running openstack overcloud deploy.

<OVERCLOUD_STACK_NAME>.<ROLE_TYPE>.<NODE_INDEX>.<ROLE_NAME>: resource_type: OS::TripleO::<ROLE_TYPE>Server physical_resource_id: <RESOURCE_ID> status: CREATE_FAILED status_reason: | ResourceInError: resources.<ROLE_NAME>: Went to status ERROR due to "Message: No valid host was found. There are not enough hosts available., Code: 500"
overcloud.Controller.0.Controller: resource_type: OS::TripleO::ControllerServer physical_resource_id: 2e41f61b-8f3c-4fed-a523-0e56a7a88ecc status: CREATE_FAILED status_reason: | ResourceInError: resources.Controller: Went to status ERROR due to "Message: No valid host was found. There are not enough hosts available., Code: 500"
overcloud.Compute.0.NovaCompute: resource_type: OS::TripleO::ComputeServer physical_resource_id: ab29fe63-4103-4afd-bc95-9a9e720920ed status: CREATE_FAILED status_reason: | ResourceInError: resources.NovaCompute: Went to status ERROR due to "Message: No valid host was found. There are not enough hosts available., Code: 500"
overcloud.BlockStorage.0.BlockStorage: resource_type: OS::TripleO::BlockStorageServer physical_resource_id: 7640f169-7790-4b73-9006-8171ddd450e4 status: CREATE_FAILED status_reason: | ResourceInError: resources.BlockStorage: Went to status ERROR due to "Message: No valid host was found. There are not enough hosts available., Code: 500"
  • All Overcloud nodes are required to have been successfully introspected.

    • If introspection is failing with a kernel panic, ensure the nodes have at least 4GB of RAM.

  • All nodes must have “Maintenance” mode set to “False” and be in the “Provisioning State” of “available” .

    $ openstack baremetal node list
    $ openstack baremetal node maintenance unset <BAREMETAL_NODE>
    $ openstack baremetal node provide <BAREMETAL_NODE>
    
  • All Overcloud nodes require a profile tag that will determine what type of node it will be.

    • Manually set the profile type for each node: block-storage, ceph-storage, compute, control, or swift-storage.

      $ openstack baremetal node set --property capabilities='profile:<PROFILE_TYPE>,boot_option:local' <BAREMETAL_NODE>
      
  • Ensure that each <ROLE_NAME>Count: Heat parameter is correctly set to the number of nodes that are available.


StackValidationFailed: The Resource Type (OS::TripleO::<ROLE>::<RESOURCE>) could not be found.” when running openstack overcloud deploy.

  • The <ROLE> needs to be set to a valid role defined in the roles_data.yaml file.

  • The resource_registry mapping must link a resource type to a file ending with the .yaml extension (.yml will not work). Example:

    ---
    resource_registry:
      OS::TripleO::ControllerDeployedServer::Net::SoftwareConfig: net-config-static-bridge.yaml
    
  • The resource type may be undefined. That means a THT template is missing and needs to be included as part of the deployment.

  • The resource type may be misspelled.


Config download already in progress with execution id <MISTRAL_EXECUTION_ID> for stack overcloud” when running openstack overcloud deploy.

  • Check the status of the Mistral execution: openstack workflow execution show <MISTRAL_EXECUTION_ID>.

  • Check if the deployment is still running: ps faux | grep ansible-playbook.

    • kill any lingering Ansible processes.

  • Manually set the workflow execution state to CANCELLED or FAILED.

    $ openstack workflow execution update -s FAILED <MISTRAL_EXECUTION_ID>
    
  • Re-run openstack overcloud deploy.


ERROR configuring <OPENSTACK_SERVICE>

  • A configuration section may be misconfigured to use an invalid section or key.

  • A duplicate configuration entry may exist. This usually means a different Puppet manifest is trying to declare the same value for a specified section and key. When the operator defines the same variable elsewhere, the deployment will error out. The example below showcases that the configuration is already handled by the Puppet class oslo::messaging::default.

$ less /var/log/containers/stdouts/container-puppet-<OPENSTACK_SERVICE>.log
2020-01-01T16:15:00.000000000-04:00 stderr F <13>Jan 01 16:12:38 puppet-user: Error: Evaluation Error: Error while evaluating a Resource Statement, Evaluation Error: Error while evaluating a Function Call, Duplicate declaration: <OPENSTACK_SERVICE>_config[<SECTION>/<KEY>] is already declared at (file: /etc/puppet/modules/<OPENSTACK_SERVICE>/manifests/config.pp, line: 36); cannot redeclare (file: /etc/puppet/modules/oslo/manifests/messaging/default.pp, line: 47) (file: /etc/puppet/modules/oslo/manifests/messaging/default.pp, line: 47, column: 3) (file: /etc/puppet/modules/<OPENSTACK_SERVICE>/manifests/init.pp, line: 445) on node undercloud.localhost.localdomain

Failed to pull image <CONTAINER_REGISTRY_NAMESPACE>/<CONTAINER_IMAGE_NAME>:<CONTAINER_IMAGE_TAG>

  • Verify that (1) the registry works, (2) the container name exists, and (3) that the container tag exists.

$ skopeo inspect --tls-verify=False docker://<CONTAINER_REGISTRY_NAMESPACE>/<CONTAINER_NAME_PERFIX><CONTAINER_IMAGE_NAME>
$ skopeo inspect --tls-verify=False docker://<CONTAINER_REGISTRY_NAMESPACE>/<CONTAINER_NAME_PERFIX><CONTAINER_IMAGE_NAME>:<CONTAINER_IMAGE_TAG>
    "RepoTags": [
     "latest",

Object GET failed: https://192.168.24.2:13808/v1/AUTH_<PROJECT_ID>/overcloud/plan-environment.yaml 404 Not Found [first 60 chars of response] <html><h1>Not Found</h1><p>The resource could not be found.<

  • Delete the overcloud Swift container.

    $ openstack container delete --recursive overcloud
    
  • Re-run the deployment with the --templates directory. All of the TripleO Heat templates used in the deployment should be stored in that directory.

    $ openstack overcloud deploy --templates ~/templates
    

[67]


Error: mistralclient.api.base.APIException: Duplicate entry for WorkbookDefinition [\\\’name\\\’, \\\’namespace\\\’, \\\’project_id\\\’]: tripleo.deployment.v1, , None\\n\’\n[2020-01-23 00:00:00,000]

  • There is a duplicate Mistral workflow on the Undercloud. In this example, it is tripleo.deployment.v1.

  • Look for files that may have been copied and/or renamed.

    $ find /usr/share/tripleo-common/workbooks/ ! -name "*.yaml"
    
  • Look for files containing the workflow definition.

    $ grep -rl 'name: tripleo.deployment.v1' /usr/share/tripleo-common/workbooks/
    
  • Delete any files found based on the previous two commands or at least move them out of the “workbooks” directory.

OpenShift on OpenStack

Red Hat provides a way to install the Red Hat OpenShift Container Platform (RHOCP, their Kubernetes product) using infrastructure from a RHOSP Overcloud.

OpenStack integration:

  • Ceilometer = Provides metrics of OpenShift resources.

  • Cinder = Block storage can be used for persistent storage on OpenShift.

  • Designate = Native DNS integration between OpenStack and OpenShift.

  • Heat = Automatic scaling of OpenShift Nodes based on capacity and usage.

  • Keystone = Authentication providers used in OpenStack, such as LDAP, can be used in OpenShift.

  • Load-Balancing-as-a-Service (Octavia) = A public load balancer than can be used by OpenShift.

  • Neutron = Kuryr is a networking plugin for Kubernetes that allows it to natively use Neutron. This removes the double overhead of having a Kubernetes overlay network ontop of OpenStack’s overlay network.

  • Nova = OpenShift Nodes can be deployed as virtual machines or on baremetal.

The first tool to accomplish this was openshift-openstack-1-2-3 which was an unofficial tool written by a Red Hat employee. All of the infrastructure is defined via Heat templates. Then the official openshift-ansible tool is used to install OpenShift. [70] Much of that work has been integrated into the openshift-ansible project for the 3.11 release.

OpenShift 4 now uses the concept of User Provisioned Infrastructure (UPI). [71] This means that the infrastructure is no longer managed by OpenShift but reference architecture material is available. [72] However, Ansible content is being developed to automate those steps (without Heat) in the new openshift/installer project. The deployment is highly opinionated to provide an out-of-the-box solution following best practices. It supports deploying infrastracture on RHOSP 13 and 16 and installing OpenShift 4 ontop of it. [73]

History

Bibliography

  1. “Red Hat OpenStack Platform Life Cycle.” Red Hat Support. Accessed June 17, 2020. https://access.redhat.com/support/policy/updates/openstack/platform

  2. “Frequently Asked Questions.” RDO Project. Accessed December 21, 2017. https://www.rdoproject.org/rdo/faq/

  3. “Director Installation and Usage.” Red Hat OpenStack Platform 13 Documentation. September 26, 2018. Accessed September 26, 2018. https://access.redhat.com/documentation/en-us/red_hat_openstack_platform/13/pdf/director_installation_and_usage/Red_Hat_OpenStack_Platform-13-Director_Installation_and_Usage-en-US.pdf

  4. “Packstack: Create a proof of concept cloud.” RDO Project. Accessed March 19, 2018. https://www.rdoproject.org/install/packstack/

  5. “Neutron with existing external network. RDO Project. Accessed September 28, 2017. https://www.rdoproject.org/networking/neutron-with-existing-external-network/

  6. “Error while installing openstack ‘newton’ using rdo packstack.” Ask OpenStack. October 25, 2016. Accessed September 28, 2017. https://ask.openstack.org/en/question/97645/error-while-installing-openstack-newton-using-rdo-packstack/

  7. “TripleO quickstart.” RDO Project. Accessed March 26, 2018. https://www.rdoproject.org/tripleo/

  8. “[TripleO] Minimum System Requirements.” TripleO Documentation. September 7, 2016. Accessed March 26, 2018. https://images.rdoproject.org/docs/baremetal/requirements.html

  9. [RDO] Recommended hardware.” RDO Project. Accessed September 28, 2017. https://www.rdoproject.org/hardware/recommended/

  10. “[TripleO] Virtual Environment.” TripleO Documentation. Accessed September 28, 2017. http://tripleo-docs.readthedocs.io/en/latest/environments/virtual.html

  11. “Getting started with TripleO-Quickstart.” OpenStack Documentation. Accessed December 20, 2017. https://docs.openstack.org/tripleo-quickstart/latest/getting-started.html

  12. “TripleO Documentation.” OpenStack Documentation. Accessed September 12, 2017. https://docs.openstack.org/tripleo-docs/latest/

  13. “Basic Deployment (CLI).” TripleO Documentation. November 23, 2020. Accessed December 3, 2020. https://docs.openstack.org/project-deploy-guide/tripleo-docs/latest/deployment/install_overcloud.html

  14. “Bug 1466744 - Include docker.yaml and docker-ha.yaml environment files by default.” Red Hat Bugzilla. December 13, 2017. Accessed January 12, 2018. https://bugzilla.redhat.com/show_bug.cgi?id=1466744

  15. “Baremetal Environment.” TripleO OpenStack Documentation. October 25, 2019. Accessed October 28, 2019. https://docs.openstack.org/project-deploy-guide/tripleo-docs/latest/environments/baremetal.html

  16. “Does Red Hat Ceph support migration from FileStore to BlueStore with the release of RHCS 3.2?” Red Hat Customer Portal. May 23, 2019. Accessed October 28, 2019. https://access.redhat.com/articles/3793241

  17. “Configuring Ceph with Custom Config Settings.” OpenStack Documentation. October 25, 2019. Accessed October 28, 2019. https://docs.openstack.org/project-deploy-guide/tripleo-docs/latest/features/ceph_config.html

  18. “[Ironic] Enabling drivers.” OpenStack Documentation. March 15, 2018. Accessed March 15, 2018. https://docs.openstack.org/ironic/queens/admin/drivers.html

  19. “CHAPTER 8. SCALING THE OVERCLOUD.” Red Hat Documentation. Accessed January 30, 2018. https://access.redhat.com/documentation/en-us/red_hat_openstack_platform/10/html/director_installation_and_usage/sect-scaling_the_overcloud

  20. “Containers based Undercloud Deployment.” OpenStack Documentation. October 25, 2019. Accessed October 28, 2019. https://docs.openstack.org/project-deploy-guide/tripleo-docs/latest/deployment/undercloud.html

  21. “[TripleO Quickstart] Networking.” TripleO Documentation. September 7, 2016. Accessed April 9, 2018. https://images.rdoproject.org/docs/baremetal/networking.html

  22. “Repository Enablement.” OpenStack TripleO Documentation. October 25, 2019. Accessed October 28, 2019. https://docs.openstack.org/project-deploy-guide/tripleo-docs/latest/repositories.html

  23. “TripleO: Using the fake_pxe driver with Ironic.” Leif Madsen Blog. November 11, 2016. Accessed June 13, 2018. http://blog.leifmadsen.com/blog/2016/11/11/tripleo-using-the-fake_pxe-driver-with-ironic/

  24. “Bug 1535214 - baremetal commands that were deprecated in Ocata have been removed in Queens.” Red Hat Bugzilla. Accessed June 13, 2018. https://bugzilla.redhat.com/show_bug.cgi?id=1535214

  25. “OpenStack lab on your laptop with TripleO and director.” Tricky Cloud. November 25, 2015. Accessed June 13, 2018. https://trickycloud.wordpress.com/2015/11/15/openstack-lab-on-your-laptop-with-tripleo-and-director/

  26. “Director Installation and Usage.” Red Hat OpenStack Platform 10 Documentation. Accessed July 18, 2018. https://access.redhat.com/documentation/en-us/red_hat_openstack_platform/10/html/director_installation_and_usage/

  27. “Director Installation and Usage.” Red Hat OpenStack Platform 13 Documentation. Accessed July 18, 2018. https://access.redhat.com/documentation/en-us/red_hat_openstack_platform/13/html/director_installation_and_usage/

  28. “Red Hat OpenStack Platform 13 Release Notes.” Red Hat OpenStack Platform 13 Documentation. September 20, 2018. Accessed September 26, 2018. https://access.redhat.com/documentation/en-us/red_hat_openstack_platform/13/pdf/release_notes/Red_Hat_OpenStack_Platform-13-Release_Notes-en-US.pdf

  29. “Use an external Ceph cluster with the Overcloud.” TripleO Documentation. October 25, 2019. Accessed October 28, 2019. https://docs.openstack.org/project-deploy-guide/tripleo-docs/latest/features/ceph_external.html

  30. “TRIPLEO AND ANSIBLE: CONFIG-DOWNLOAD WITH ANSIBLE TOWER (PART 3).” Slagle’s Blog. June 1, 2018. Accessed October 3, 2018. https://blogslagle.wordpress.com/2018/06/01/tripleo-and-ansible-config-download-with-ansible-tower-part-3/

  31. “Configuring Network Isolation.” TripleO Documentation. January 30, 2020. Accessed February 5, 2020. https://docs.openstack.org/project-deploy-guide/tripleo-docs/latest/features/network_isolation.html

  32. “Modifying default node configuration.” TripleO Documentation. September 21, 2020. Accessed September 23, 2020. https://docs.openstack.org/project-deploy-guide/tripleo-docs/latest/features/node_config.html

  33. “Containers based Overcloud Deployment.” OpenStack Documentation. October 25, 2019. Accessed October 28, 2019. https://docs.openstack.org/project-deploy-guide/tripleo-docs/latest/deployment/overcloud.html

  34. CHAPTER 12. REBOOTING NODES.” Red Hat OpenStack Platform 13 Documentation. Accessed January 28, 2019. https://access.redhat.com/documentation/en-us/red_hat_openstack_platform/13/html/director_installation_and_usage/sect-rebooting_the_overcloud

  35. “Bootstrap.” InfraRed Documentation. Accessed February 8, 2019. https://infrared.readthedocs.io/en/stable/bootstrap.html

  36. “CHAPTER 8. CONFIGURING A BASIC OVERCLOUD USING PRE-PROVISIONED NODES.” Red Hat Documentation. Accessed January 28, 2020. https://access.redhat.com/documentation/en-us/red_hat_openstack_platform/13/html/director_installation_and_usage/chap-configuring_basic_overcloud_requirements_on_pre_provisioned_nodes

  37. “Using Already Deployed Servers.” OpenStack Documentation. January 30, 2020. Accessed February 4, 2020. https://docs.openstack.org/project-deploy-guide/tripleo-docs/latest/features/deployed_server.html

  38. “CHAPTER 4. INSTALLING THE UNDERCLOUD.” Red Hat Documentation. Accessed April 1, 2019. https://access.redhat.com/documentation/en-us/red_hat_openstack_platform/13/html/director_installation_and_usage/installing-the-undercloud

  39. “CHAPTER 10. CONFIGURING THE OVERCLOUD WITH ANSIBLE.” Red Hat Documentation. Accessed May 14, 2019. https://access.redhat.com/documentation/en-us/red_hat_openstack_platform/13/html/director_installation_and_usage/configuring-the-overcloud-with-ansible

  40. “Evaluating OpenStack: Single-Node Deployment.” Red Hat Knowledgebase. October 5, 2018. Accessed May 15, 2019. https://access.redhat.com/articles/1127153

  41. “TripleO config-download User’s Guide: Deploying with Ansible.” TripleO Documentation. November 23, 2020. Accessed December 8, 2020. https://docs.openstack.org/project-deploy-guide/tripleo-docs/latest/deployment/ansible_config_download.html

  42. “CHAPTER 3. PREPARING FOR DIRECTOR INSTALLATION.” Red Hat RHOSP 15 Documentation. Accessed September 26, 2019. https://access.redhat.com/documentation/en-us/red_hat_openstack_platform/15/html/director_installation_and_usage/preparing-for-director-installation

  43. “The road ahead for the Red Hat OpenStack Platform.” Red Hat Blog. August 20, 2019. Accessed September 26, 2019. https://www.redhat.com/en/blog/road-ahead-red-hat-openstack-platform

  44. “Installing a Undercloud Minion.” OpenStack Documentation. October 29, 2019. Accessed November 1, 2019. https://docs.openstack.org/project-deploy-guide/tripleo-docs/latest/features/undercloud_minion.html

  45. “CHAPTER 3. PLANNING YOUR OVERCLOUD.” Red Hat Documentation. Accessed November 20, 2019. https://access.redhat.com/documentation/en-us/red_hat_openstack_platform/13/html/director_installation_and_usage/chap-planning_your_overcloud

  46. “Configuring High Availability.” tripleo-docs. November 20, 2019. Accessed November 20, 2019. https://docs.openstack.org/project-deploy-guide/tripleo-docs/latest/features/high_availability.html

  47. “Scale Undercloud with a Minion.” tripleo-docs. May 3, 2019. Accessed December 3, 2019. https://specs.openstack.org/openstack/tripleo-specs/specs/train/undercloud-minion.html

  48. “Understanding undercloud/standalone stack updates.” tripleo-docs. December 19, 2019. Accessed December 19, 2019. https://docs.openstack.org/project-deploy-guide/tripleo-docs/latest/post_deployment/updating-stacks-notes.html

  49. “Deleting Overcloud Nodes.” TripleO Documentation. January 30, 2020. Accessed January 30, 2020. https://docs.openstack.org/project-deploy-guide/tripleo-docs/latest/post_deployment/delete_nodes.html

  50. “Scaling the Overcloud. Red Hat OpenStack Platform 13 Documentation. Accessed January 30, 2020. https://access.redhat.com/documentation/en-us/red_hat_openstack_platform/13/html/director_installation_and_usage/sect-scaling_the_overcloud

  51. “Chapter 19. Storage Configuration.” Red Hat OpenStack Platform 13 Documentation. Accessed February 5, 2020. https://access.redhat.com/documentation/en-us/red_hat_openstack_platform/13/html/advanced_overcloud_customization/storage_configuration

  52. “TripleO Architecture.” TripleO Documentation. August 16, 2019. Accessed February 6, 2020. https://docs.openstack.org/tripleo-docs/latest/install/introduction/architecture.html

  53. “Overview of available RDO repos.” RDO Project. January 27, 2018. Accessed February 6, 2020. https://www.rdoproject.org/what/repos/

  54. “Workflow: RDO Trunk repo.” RDO Project. May 24, 2019. Accessed February 6, 2020. https://www.rdoproject.org/what/trunk-repos/

  55. “Director Installation and Usage.” Red Hat OpenStack Platform 16.0 Documentation. Accessed February 7, 2020. https://access.redhat.com/documentation/en-us/red_hat_openstack_platform/16.0/html/director_installation_and_usage/index

  56. “Ensure config-download mappings are unset on ceph-upgrade.” OpenDev openstack/tripleo-heat-templates. April 27, 2018. Accessed February 10, 2020. https://opendev.org/openstack/tripleo-heat-templates/commit/24469e3c02747b7b6de6d61fcf2a8b9be67b370b

  57. “TripleO Project Specifications.” TripleO Documentation. October 16, 2019. Accessed February 17, 2020. https://specs.openstack.org/openstack/tripleo-specs/

  58. “Blueprints for tripleo.” tripleo Launchpad. Accessed February 17, 2020. https://blueprints.launchpad.net/tripleo

  59. “CHAPTER 7. CONFIGURING A BASIC OVERCLOUD WITH CLI TOOLS.” Red hat RHOSP 16 Documentation. Accessed April 21, 2020. https://access.redhat.com/documentation/en-us/red_hat_openstack_platform/16.0/html/director_installation_and_usage/creating-a-basic-overcloud-with-cli-tools

  60. “Building a Single Image.” TripleO Documentation. April 20, 2020. Accessed April 21, 2020. https://docs.openstack.org/project-deploy-guide/tripleo-docs/latest/deployment/build_single_image.html

  61. “Network Configuration with os-net-config.” Ales Nosek - The Software Practitioner. September 28, 2015. Accessed May 4, 2020. http://alesnosek.com/blog/2015/09/28/network-configuration-with-os-net-config/

  62. “Linux Bridge Container Permission Issues.” Launchpad Bugs tripleo. May 4, 2020. Accessed May 4, 2020. https://bugs.launchpad.net/tripleo/+bug/1862179

  63. “Deploying with Custom Roles.” TripleO Documentation. May 7, 2020. Accessed May 7, 2020. https://docs.openstack.org/project-deploy-guide/tripleo-docs/latest/features/custom_roles.html

  64. “Deploying with Composable Services.” TripleO Documentation. May 7, 2020. Accessed May 7, 2020. https://docs.openstack.org/project-deploy-guide/tripleo-docs/latest/features/composable_services.html

  65. “Promotion Stages.” TripleO Documentation. August 16, 2019. Accessed May 26, 2020. https://docs.openstack.org/tripleo-docs/latest/ci/stages-overview.html

  66. “Scaling Red Hat OpenStack Platform to more than 500 Overcloud Nodes.” Red Hat Blog. October 28, 2019. Accessed May 26, 2020. https://www.redhat.com/en/blog/scaling-red-hat-openstack-platform-more-500-overcloud-nodes

  67. “Bug 1607453 - Deployment fails with: Object GET failed: https://…/overcloud/plan-environment.yaml 404 Not Found.” Red Hat Bugzilla. November 13, 2019. Accessed May 28, 2020. https://bugzilla.redhat.com/show_bug.cgi?id=1607453

  68. “Minor version update.” TripleO Upgrade Developer Documentation. January 20, 2020. Accessed June 19, 2020. https://docs.openstack.org/tripleo-docs/latest/upgrade/developer/upgrades/minor_update.html

  69. “Upgrading to a Next Major Release.” TripleO Upgrade Documentation. June 8, 2020. Accessed June 19, 2020. https://docs.openstack.org/project-deploy-guide/tripleo-docs/latest/post_deployment/upgrade/major_upgrade.html

  70. “OpenShift on OpenStack 1-2-3: Bringing IaaS and PaaS Together.” Red Hat OpenShift Blog. March 12, 2018. Accessed June 30, 2020. https://www.openshift.com/blog/openshift-openstack-1-2-3-bringing-iaas-paas-together

  71. “Deploying a UPI environment for OpenShift 4.1 on VMs and Bare Metal.” Red Hat OpenShift Blog. July 15, 2019. Accessed June 30, 2020. https://www.openshift.com/blog/deploying-a-upi-environment-for-openshift-4-1-on-vms-and-bare-metal

  72. “OpenStack UPI.” GitHub openshift/enhancements. September 18, 2019. Accessed June 30, 2020. https://github.com/openshift/enhancements/blob/master/enhancements/installer/openstack-upi.md

  73. “Reference Architectures 2020 Deploying Red Hat OpenShift Container Platform 4.4 on Red Hat OpenStack Platform 13 and 16.0.” Red Hat OpenShift Resources. Accessed June 30, 2020. https://www.redhat.com/cms/managed-files/cl-openshift-container-platform-4-4-on-openstack-platform-13-16-reference-architecture-f23768-202005-en.pdf

  74. “Quick Start Guide Red Hat OpenStack Platform 16.1-Beta.” Red Hat Documentation. Accessed July 23, 2020. https://access.redhat.com/documentation/en-us/red_hat_openstack_platform/16.1-beta/html-single/quick_start_guide/index

  75. “SERVICE TELEMETRY FRAMEWORK.” Red Hat OpenStack Platform 16.0 Service Telemetry Framework. Accessed October 21, 2020. https://access.redhat.com/documentation/en-us/red_hat_openstack_platform/16.0/html-single/service_telemetry_framework/index

  76. “Container Image Prepare.” TripleO Documentation. October 30, 2020. Accessed November 16, 2020. https://docs.openstack.org/project-deploy-guide/tripleo-docs/latest/deployment/container_image_prepare.html