Platform Installation

Why are you documentating the downstream version?

As we noted in the Project Overview we are documenting the downstream components initially for the deployment of the telemetry framework. We will note the upstream equivalent where possible.

The installation of the telemetry framework simply requires the deployment of OpenShift 3.11 or later and a bastion node for executing the supplied bash script to load the components into the sa-telemetry namespace.

Our reference implementation leverages 3 physical nodes using RHHI-V platform which provides us with virtualization and distributed storage.

The layers are shown in the following diagram.

_images/saf_high_level_overview.png

Installation of RHVH, RHHI-V, and OpenShift are beyond the scope of this documentation but we’ll link to existing documentation in the following sections, and describe any example configurations, or other useful information to make installation more reproducible.

We’ll build up the platform layer by installing a 3 node RHHI-V on top of RHVH and then creating the virtual machines in RHHI-V, setting up affinity, networking, etc prior to deploying OpenShift.

Installing Red Hat Virtualization Host (RHVH) Operating System

Upstream equivalent of RHVH

RHVH is the Red Hat Virtualization Host, a spin of RHEL with the specific repositories required to install RHHI-V. Installation of CentOS 7.6 or later should contain the required dependencies for an upstream installation.

First we need to install RHVH (Red Hat Virtualization Host). This needs to be done on 1 or 3 nodes (or groups of 3) so that we can do the installation of RHHI-V (Red Hat Hyperconverged Infrastructure for Virtualization) next.

Your node will need at least 2 physical disks; one for the installation of the operating system, and one blank disk for the GlusterFS system. If you have additional disks you can use these in either a RAID configuration for GlusterFS, or break the storage classes GlusterFS will setup across multiple disks.

See the official Red Hat Virtualization Hosts installation guide for obtaining RHVH and installing it.

The following procedure matches our reference architecture. Your procedure is likely to differ in order to match your environment, but the steps to building out the environment is likely to follow what we’re documenting.

Prepare nodes

  • Boot the RHVH 4.2.8 (or later) ISO for the node, or somehow automate the installation of RHVH with Satellite or other.

Now that you have three nodes with RHVH installed on them, let’s continue.

Prerequisite setup from bastian host

The bastian host is our initial terminal. It is likely your laptop, or some other dedicated jump-host in your environment.

  • Set a shell variable for the hostname you’re going to work with

    export SAFTLD=management.service-assurance.tld
    

The next steps will setup the SSH connections. We have a trusted environment and are not checking the fingerprints of the servers. Your environment may not allow for this trust.

  • Get host fingerprints and copy local SSH key onto the nodes

    for i in 1 2 3; do \
      ssh-keyscan -H virthost$i.${SAFTLD} >> ~/.ssh/known_hosts; \
      ssh-copy-id -i ~/.ssh/id_rsa.pub root@virthost$i.${SAFTLD}; \
    done
    

Prerequisite setup from the control host

Next we need to setup SSH connections into the local server and satisfy the subscriptions for the servers. If you don’t need to subscribe the servers then you can skip that step.

Login to one of the three servers you’re going to designate as the control server. If you pick the first server in your list (e.g. virthost1) then make sure when you login to the web interface to complete the installation you use that same server.

We’re going to create an SSH keypair on the control server and then copy that into the authorized_keys file on the remote machines. When we do our installation through the cockpit UI, Ansible will be executed against those nodes from our control server, and if we can’t SSH into the machines passwordless the Ansible-controlled install will fail.

  • Login to the control machine

    ssh root@virthost1.${SAFTLD}
    
  • Create an ssh key and make sure you can ssh into localhost without a password

    ssh-keygen -N '' -t rsa -b 2048 -q -f ~/.ssh/id_rsa
    ssh-copy-id -i ~/.ssh/id_rsa.pub root@$(hostname -f)
    
  • Next validate that the control machine can login to all networks that might be utilized during the installation process

    Tip

    # this one is for copying the key to all systems with various subdomains
    export SAFTLD=service-assurance.tld
    for i in 1 2 3; do ssh-keyscan -H virthost$i.${SAFTLD} >> ~/.ssh/known_hosts ; ssh-copy-id -i ~/.ssh/id_rsa root@virthost$i.${SAFTLD}; done
    

Prerequisite setup across all RHHI-V nodes

The rest of the commands need to be run on all systems.

  • (optional) Possibly setup the /etc/resolv.conf to point at your local DNS server if necessary and chattr +i /etc/resolv.conf to lock it (needed in my lab, no control of the DHCP server)

    Note

    In a proper environment this will not be necessary, but DNS is very important for proper operation of the environment as a whole

  • Register the nodes

    export SKU_NAME="SKU Name"
    subscription-manager register --username='<email>' --password='<password>'
    subscription-manager list --available --matches="${SKU_NAME}" --pool-only
    subscription-manager attach --pool=<pool from list>
    
    # alternatively do in a single command using the first item returned
    SKU_NAME="SKU Name" ; subscription-manager register --username='<email>' --password='<password>' subscription-manager attach --pool=$(subscription-manager list --available --matches="${SKU_NAME}" --pool-only | head -n1)
    
  • Enable subscriptions and update systems

    subscription-manager repos --enable=rhel-7-server-rhvh-4-rpms
    yum update -y
    
  • Wipe all the disks being used for GlusterFS to make sure they are blank.

    Warning

    Of course update the list of disk paths below to match your own environment.

    • Wipe the disks

      wipefs -f -a /dev/sdb
      wipefs -f -a /dev/sdc
      wipefs -f -a /dev/sdd
      
      # this will wipe out any master boot record (MBR) data
      dd if=/dev/zero of=/dev/sdb bs=512 count=1 conv=notrunc
      
  • Note the size of the disks in GB, which you’ll need for the next section

    fdisk -s /dev/sdd | awk '{$1=$1/(1024^2); print $1,"GB";}'
    
    # or...
    lsblk
    

Load the web interface to start installation

Go to the web interface on your control host at https://virthost1.management.service-assurance.tld:9090 to start installation of RHHI-V.

Installing Red Hat Hyperconverged Infrastructure for Virtualization (RHHI-V)

Upstream equivalent of RHHI-V

More information about deploying oVirt hyperconverged in a 1 or 3 node configuration is available at oVirt Gluster-Hyperconverged documentation.

Official documentation for installation of RHHI-V can be found at Chapter 6. Configure Red Hat Gluster Storage For Hosted Engine Using The Cockpit UI

In the previous section titled Installing Red Hat Virtualization Host (RHVH) Operating System we mentioned the creation of SSH keys and populating them among the various hosts. That is required during the installation of RHHI-V since it executes Ansible from the web interface to build the virtual machine for the engine, and to create the GlusterFS storage domains.

Download RHEL 7.6 KVM Image Onto Bastian Host

With RHHI-V now installed, we need to download a copy of the RHEL 7.6 KVM image which will be the source operating system during installation of OpenShift. You can download the image from https://access.redhat.com.

Get your download link (which is time sensitive and must be obtained each time you wish to download a new image) and download it onto your bastian host (or any other host where you can install guestfish which we’ll discuss next).

cd /tmp
curl ‘http://access.cdn.redhat.com/…’ -o rhel-server-7.6-x86_64-kvm.qcow2

Tip

For an upstream-only deployment, download CentOS 7.6 qcow2 image for your base image.

Modify RHEL 7.6 KVM Image and Upload To RHV-M Engine

Before importing the template into the system, you’ll need to run some guestfish commands to strip out the default 192.168.122.1 nameserver which can cause us issues as we’ll be running the virtual machines with bridged network interfaces.

yum install guestfish -y
systemctl start libvirtd.service
virt-edit --expr 's/nameserver 192.168.122.1//g' \
    -a /tmp/rhel-server-7.6-x86_64-kvm.qcow2 /etc/resolv.conf

We can now upload our modified virtual machine image to the RHV-M engine with scp or another method. Our Ansible for creating the rhel76_template in RHHI-V will expect the virtual machine image file to be located in the /root directory on the RHV-M engine.

scp /tmp/rhel-server-7.6-x86_64-kvm.qcow2 root@engine.rhhi-v.tld:/root

Installing OpenShift

Upstream equivalent of OpenShift

The upstream OpenShift project is known as OKD and is the Red Hat distribution of Kubernetes. Documentation for installation of OKD 3.11 is available at docs.okd.io.

Installation of OpenShift within RHHI-V is done with Ansible playbooks and roles as created within the rhhi-v/ subdirectory of the telemetry-framework repository.

The first step is creating an inventory file that will result in the creation of the virtual machines within the RHHI-V environment and then subsequently execute openshift-ansible to install the OpenShift platform within the virtual machines.

Both upstream and downstream deployment methods use the same automation and the primary difference is that when installing the downstream version of OpenShift an extra variables file will be required that contains the information for registration of OpenShift and pulling the components from another repository.

Creating Virtual Machine and OpenShift Inventory Files

A few example inventory files for our lab configuration exist within the rhhi-v/inventory/ directory. We’ll create a new inventory configuration based on a working example.

On your bastian host clone the telemetry-framework repository and create a new directory in the rhhi-v/inventory/ directory.

mkdir -p ~/src/github.com/redhat-service-assurance
cd ~/src/github.com/redhat-service-assurance
git clone https://github.com/redhat-service-assurance/telemetry-framework
cd telemetry-framework/rhhi-v
mkdir inventory/my_lab

With our new directory created, we need two inventory files to deploy our infrastructure on top of RHHI-V; hosts.yml and openshift.yml.

The hosts.yml file will contain the information required to instantiate and configure the virtual machines in RHHI-V in preparation for our OpenShift installation.

Virtual Machines Inventory File

The hosts.yml file is relatively long, but most of it is boilerplate that you can use to build out your own hosts file. An example inventory file is available in rhhi-v/inventory/nfvha-lab/hosts.yml.

OpenShift Inventory File

The openshift.inventory file is used by openshift-ansible to configure your OpenShift cluster. An example openshift.inventory file that works with the hosts.yml file for creating the virtual machines, is available in the rhhi-v/inventory/nfvha-lab/openshift.inventory file.

Create Variables Files

We need a file that contains our login information for registering our RHEL virtual machines.

Importing RHEL 7.6 Template Into RHHI-V

ansible-galaxy install -r requirements.yml
ansible-playbook -i inventory/nfvha-lab/ \
     --ask-vault-pass playbooks/rhel-template.yml

Instantiate The OpenShift Cluster on RHHI-V

# make sure you edit your inventory files first
cd telemetry-framework/rhhi-v/
ansible-playbook -i inventory/nfvha-lab/ \
    --ask-vault-pass -e “@./vars/rhsub.vars” playbooks/vm-infra.yml