Jump to: navigation, search

StarlingX/Installation Guide

< StarlingX
Revision as of 20:01, 20 June 2018 by David.b.kinder (talk | contribs) (Initializing Compute Host)

Intro

This section contains information about the StarlingX installation in a virtualized environment using Libvirt/QEMU.

Software Configurations

  • All In One
  • Standard Controller Storage
  • Duplex
  • Standard Dedicated Storage

Standard Controller Storage

Requirements

Different use cases require different configurations. For general StarlingX deployment, the recommended minimum requirements include:

Hardware Requirements

A workstation computer with:

  • Processor: x86_64 only supported architecture with hardware virtualization extensions
  • Memory: At least 32GB RAM
  • Hard Disk: 500GB HDD
  • Network: Two network adapters with active Internet connection

Software Requirements

A workstation computer with:

  • Operating System: Freshly installed Ubuntu 16.04 LTS 64-bit
  • Proxy settings configured (if applies)
  • Git
  • KVM/VirtManager
  • Libvirt Library
  • QEMU Full System Emulation Binaries
  • <stx-deployment> project
  • StarlingX ISO Image

Deployment Environment Setup

This section describes how to set up a StarlingX system in a workstation computer. After completing these steps, you will be able to deploy and run your StarlingX system on the following Linux distribution:

  • Ubuntu 16.04 LTS 64-bit

Updating Your Operating System

Before proceeding with the build, ensure your OS is up to date. You’ll first need to update the local database list of available packages:

$ sudo apt-get update

Installing Requirements and Dependencies

Install the required packages in an Ubuntu host system with:

$ sudo apt-get install git virt-manager libvirt-bin qemu-system

Installing Deployment Tool

Clone the <stx-deployment> project. Usually you’ll want to clone it under your user’s home directory.

$ cd $HOME
$ git clone <stx-deployment> 

Getting the StarlingX ISO Image

1. Get the StarlingX ISO Image from:

Tbd

2. Copy the StarlingX ISO Image to the <stx-deployment> libvirt project directory naming it as bootimage.iso:

$ cp <starlingx iso image> $HOME/<stx-deployment>/libvirt/bootimage.iso

Controller-0 Host Installation

Installing controller-0 involves initializing a host with software and then applying a configuration from the command line. The configured host becomes Controller-0.
Procedure:

  1. Using an ISO image of StarlingX, initialize the controller host via Libvirt/QEMU.
  2. Configure the controller using the config_controller script.

Initializing Controller-0

This section describes how to initialize StarlingX in host Controller-0. Except where noted, all the commands must be executed from a console of the Workstation.

Navigate to the <stx-deployment> libvirt project directory:

$ cd <stx-deployment>/libvirt

Run the install packages script:

$ bash install_packages.sh

Run the libvirt qemu setup script:

$ bash setup_tic.sh

From the KVM/VirtManager window, power on the host to be configured as Controller-0 and show the virtual machine console and details:

  • When the installer is loaded and the installer welcome screen appears in the Controller-0 host, select the type of installation "Standard Controller Configuration".
  • Select the "Graphical Console" as the console to use during installation.
  • Select "Standard Security Boot Profile" as the Security Profile.
  • Monitor the initialization until it is complete. When initialization is complete, a reboot is initiated on the Controller-0 host, briefly displays a GNU GRUB screen, and then boots automatically into the StarlingX image.

Log into Controller-0 as user wrsroot, with password wrsroot. The first time you log in as wrsroot, you are required to change your password. Enter the current password (wrsroot):

Changing password for wrsroot.
(current) UNIX Password:

Enter a new password for the wrsroot account:

New password:

Enter the new password again to confirm it:

Retype new password:

Controller-0 is initialized with StarlingX, and is ready for configuration.

Configuring Controller-0

This section describes how to perform the Controller-0 configuration interactively. Except where noted, all the commands must be executed from the console of the active controller (here assumed to be controller-0.

When run interactively, the config_controller script presents a series of prompts for initial configuration of StarlingX. The script is used to configure the first controller in the StarlingX cluster as controller-0. The prompts are grouped by configuration area. To start the script interactively, use the following command with no parameters and accept all the default values:

controller-0:~$ sudo config_controller

The output when config_controller script is run interactively is:

WARNING: Command should only be run from the console. Continuing with this
terminal may cause loss of connectivity and configuration failure
...
Apply the above configuration? [y/n]: y

Applying configuration (this will take several minutes):

01/08: Creating bootstrap configuration ... DONE
02/08: Applying bootstrap manifest ... DONE
03/08: Persisting local configuration ... DONE
04/08: Populating initial system inventory ... DONE
05:08: Creating system configuration ... DONE
06:08: Applying controller manifest ... DONE
07:08: Finalize controller configuration ... DONE
08:08: Waiting for service activation ... DONE

Configuration was applied

Please complete any out of service comissioning steps with system commands and unlock controller to proceed.

Controller-0 and System Provision

Configuring Provider Networks at Installation

You must set up provider networks at installation so that you can attach data interfaces and unlock the compute nodes.

On Controller-0, acquire Keystone administrative privileges:

controller-0:~$ source /etc/nova/openrc

Set up one provider network of the vlan type, named providernet-a:

[wrsroot@controller-0 ~(keystone_admin)]$ neutron providernet-create providernet-a --type=vlan
[wrsroot@controller-0 ~(keystone_admin)]$ neutron providernet-range-create --name providernet-a-range1 --range 100-400 providernet-a

Unlocking Controller-0

You must unlock controller-0 so that you can use it to install the remaining hosts. On Controller-0, acquire Keystone administrative privileges:

controller-0:~$ source /etc/nova/openrc

Use the system host-unlock command:

[wrsroot@controller-0 ~(keystone_admin)]$ system host-unlock controller-0

The host is rebooted. During the reboot, the command line is unavailable, and any ssh connections are dropped. To monitor the progress of the reboot, use the controller-0 console.

Verifying the Controller-0 Configuration

On Controller-0, acquire Keystone administrative privileges:

controller-0:~$ source /etc/nova/openrc

Verify that the Titanium Cloud controller services are running:

[wrsroot@controller-0 ~(keystone_admin)]$ nova service-list
+--------------------------------------+------------------+--------------+----------+---------+-------+ ...
| Id                                   | Binary           | Host         | Zone     | Status  | State | ...
+--------------------------------------+------------------+--------------+----------+---------+-------+ ...
| d7cdfaf0-9394-4053-b752-d609c837b729 | nova-conductor   | controller-0 | internal | enabled | up    | ...
| 692c2659-7188-42ec-9ad1-c88ac3020dc6 | nova-scheduler   | controller-0 | internal | enabled | up    | ...
| 5c7c9aad-696f-4dd1-a497-464bdd525e0c | nova-consoleauth | controller-0 | internal | enabled | up    | ...
+--------------------------------------+------------------+--------------+----------+---------+-------+ ...

Verify that controller-0 is unlocked, enabled, and available:

[wrsroot@controller-0 ~(keystone_admin)]$ system host-list
+----+--------------+-------------+----------------+-------------+--------------+
| id | hostname     | personality | administrative | operational | availability |
+----+--------------+-------------+----------------+-------------+--------------+
| 1  | controller-0 | controller  | unlocked       | enabled     | available    |
+----+--------------+-------------+----------------+-------------+--------------+

Compute Host Installation

After initializing and configuring an active controller, you can add and configure a backup controller and additional compute or storage hosts. Using the system host-add command, you add one or more host entries to the system inventory, assigning a personality, MAC address, IP address, and so on for each host, and then you power on the hosts, causing them to be recognized and configured according to the system inventory entry.

Initializing Compute Host

On Workstation, print information of virbr2 virtual interface associated to compute-N host:

$ sudo virsh domiflist compute-0 | grep virbr2
vnet5      bridge     virbr2     e1000       52:54:00:b6:1f:c7
$ sudo virsh domiflist compute-1 | grep virbr2
vnet9      bridge     virbr2     e1000       52:54:00:da:58:b4

On Controller-0, acquire Keystone administrative privileges:

controller-0:~$ source /etc/nova/openrc

Use the system host-add command to add compute-N host and specify their compute personality using their associated virbr2 virtual interfaces MAC addresses:

[wrsroot@controller-0 ~(keystone_admin)]$ system host-add -n compute-0 -p compute -m 52:54:00:15:7a:86
[wrsroot@controller-0 ~(keystone_admin)]$ system host-add -n compute-1 -p compute -m 52:54:00:aa:a2:46

On Workstation, start Compute-N host:

$ sudo virsh start compute-0
$ sudo virsh start compute-1

Once the message "Domain compute-N started" is displayed, from the KVM/VirtManager window, power on the host to be configured as compute-N and show the virtual machine console and details. The node is assigned the personality specified in the system host-add parameters. A display device menu appears on the console, with text customized for the personality (Controller, Storage, or Compute Node). You can start the installation manually by pressing Enter. Otherwise, it is started automatically after a few seconds.

On Controller-0, you can monitor the installation progress by running the system host-show command for the host periodically. Progress is shown in the install_state field.

[wrsroot@controller-0 ~(keystone_admin)]$ system host-show compute-0 | grep install
| install_output      | text                                 |
| install_state       | booting                              |
| install_state_info  | None                                 |
[wrsroot@controller-0 ~(keystone_admin)]$ system host-show compute-1 | grep install
| install_output      | text                                 |
| install_state       | booting                              |
| install_state_info  | None                                 |
[wrsroot@controller-0 ~(keystone_admin)]$ 

Wait while the compute-N is configured and rebooted. Up to 20 minutes may be required for a reboot, depending on hardware. When the reboot is complete, the compute-N is reported as Locked, Disabled, and Online.

[wrsroot@controller-0 ~(keystone_admin)]$ system host-list
+----+--------------+-------------+----------------+-------------+--------------+
| id | hostname     | personality | administrative | operational | availability |
+----+--------------+-------------+----------------+-------------+--------------+
| 1  | controller-0 | controller  | unlocked       | enabled     | available    |
| 2  | compute-0    | compute     | locked         | disabled    | online       |
| 3  | compute-1    | compute     | locked         | disabled    | online       |
+----+--------------+-------------+----------------+-------------+--------------+
[wrsroot@controller-0 ~(keystone_admin)]$ 

Compute Host Provision

You must configure the network interfaces and the storage disks on a host before you can unlock it.

On Controller-0, acquire Keystone administrative privileges:

controller-0:~$ source /etc/nova/openrc

Provisioning Network Interfaces on a Compute Host

Provision the data interfaces

[wrsroot@controller-0 ~(keystone_admin)]$ system host-if-modify -p providernet-a -nt data compute-0 ens6
[wrsroot@controller-0 ~(keystone_admin)]$ system host-if-modify -p providernet-a -nt data compute-1 ens6

Provisioning Storage on a Compute Host

Ensure that provider networks are available for the data interfaces. Provision the data interfaces:

system host-list --nowrap &> /dev/null && NOWRAP="--nowrap"
ALL_COMPUTE=`system host-list $NOWRAP | grep compute- | cut -d '|' -f 3`
# for each compute node, we should run the followings
for compute in $ALL_COMPUTE; do
    system host-cpu-modify ${compute} -f vswitch -p0 1
    system host-lvg-add ${compute} nova-local
    system host-pv-add ${compute} nova-local $(system host-disk-list ${compute} $NOWRAP | grep /dev/sdb | awk '{print $2}')
    system host-lvg-modify -b image -s 10240 ${compute} nova-local
done

Unlocking a Compute Host

Use the system host-unlock command to unlock the node:

[wrsroot@controller-0 ~(keystone_admin)]$ system host-unlock compute-0
[wrsroot@controller-0 ~(keystone_admin)]$ system host-unlock compute-1

Wait while the compute-N is rebooted. Up to 10 minutes may be required for a reboot, depending on hardware. The host is rebooted, and its Availability State is reported as In-Test.

System Health Check

After a few minutes, all nodes shall be reported as Unlocked, Enabled, and Available:

[wrsroot@controller-0 ~(keystone_admin)]$ system host-list
+----+--------------+-------------+----------------+-------------+--------------+
| id | hostname     | personality | administrative | operational | availability |
+----+--------------+-------------+----------------+-------------+--------------+
| 1  | controller-0 | controller  | unlocked       | enabled     | available    |
| 2  | compute-0    | compute     | unlocked       | enabled     | available    |
| 3  | compute-1    | compute     | unlocked       | enabled     | available    |
+----+--------------+-------------+----------------+-------------+--------------+