/cfg_ocp_deploy_cluster

cfg_ocp_deploy_cluster

Description:

This playbook deploys a customer OCP cluster

Behaviour:

Feature: Deploy OCP cluster

As a PaaS Operator I want to deploy a full OCP cluster for customer usage

  • Scenario: OCP Cluster is fully deployed
  • Given: at least one RHEL7 master node
  • Given: at least one RHEL7 infra node
  • Given: at least one RHEL7 compute node
  • Given: a correctly configured inventory file
  • When: the playbook is executed
  • Then: nodes are checked to ensure pre-configuration is correct
  • Then: the cluster is deployed

Dependencies:

RPMS:

  • ansible.x86_64.2.4.1.0-1.el7
  • java-1.8.0-openjdk-headless.x86_64.1:1.8.0.151-1.b12.el7_4
  • python-passlib.noarch.1.6.5-2.el7
  • httpd-tools.x86_64 0:2.4.6-67.el7_4.6

Checkout the ansible installer into the same directory as the cfg_ocp_deploy_cluster checkout:

$ git clone https://github.com/openshift/openshift-ansible
$ pushd openshift-ansible ; git checkout -b release-3.6 origin/release-3.6 ; popd

The playbook uses the ocp_configure_node and rhel_register_node roles. These need to be loaded by running:

$ ansible-galaxy install -r roles/requirements.yml --force

prior to running the playbook.

Configuration:

A list of the external variables used by the playbook.

Variable Description Example Where set
redhat_username A valid Red Hat portal usernam myuser vars/secrets.yml
redhat_password Red Hat portal password myuserpassword vars/secrets.yml
redhat_subscription_poolid Pool ID for an OCP subscription 908adcef098098 vars/secrets.yml
min_cpu List of minimum cpu requirements, by node_type (defaults in ocp_configure_node role) override in group_vars/all.yml if needed
min_storage List of minimum storage requirements, by node_type (defaults in ocp_configure_node role) override in group_vars/all.yml if needed
min_memory List of minimum memory requirements, by node_type (defaults in ocp_configure_node role) override in group_vars/all.yml if needed
docker_device Device to use as docker storage (defaults in ocp_configure_node role) override in group_vars/all.yml if needed

Inventory:

The inventory file (inventory/inventory.yml) is broadly similar to a standard OCP inventory file. The only differences is that there is a new group called 'compute'. The existing group 'nodes' is a superset of the 'masters' and 'compute' groups.

N.B For testing purposes, the installer resource checks are disabled (openshift_disable_check in the inventory).

Authentication:

Basic authentication is configured at deployment. This requires an htpasswd file with a password for the admin user in the 'templates/' directory.

Testing:

Testing is currently carried out on AWS servers. To test, do the following:

  1. Create four t2.large VMs, each with an additional 10GB of EBS storage for use as docker storage.
  2. Designate one node as master, one as infra, and the other two as compute.
  3. Configure a security group so that all nodes can talk to each other, and are externally accessible via ssh.
  4. Configure a security group for the master permitting external access on 8443.
  5. Configure a security group for the infra node permitting external access on 80 and 443.
  6. Update the inventory as required.
  7. Edit inventory/group_vars/all.yml to override minimum cpu, storage etc limits, and set docker and gluster device names. An example is provided.
  8. Create an htpasswd file for the admin password and save it in the files/ directory (htpasswd -c files/htpasswd admin)
  9. Ensure that vars/secrets.yaml contains your Red Hat portal credentials and a poolid.

Usage:

$ ansible-galaxy install -r requirements.yml --force
$ ansible-playbook playbook.yml --key-file <path_to_AWS_key>