myllynen / openshift-automation-tools

Basic OpenShift 3 automation tools

OpenShift Automation Tools

Tools to automate OpenShift installation and maintenance. These scripts and playbooks enable fully automated connected or disconnected OpenShift installation and updates for those parts not automated by the standard OpenShift playbooks. Can be used to quickly setup a PoC environment or to experiment with OpenShift installation parameters. Local registry use will avoid fetching gigabytes of data from external network. Automation for platform level tasks like OS upgrades.

Technical Overview

Official OpenShift documentation describes how to manually prepare nodes for installation, how to configure local registry for disconnected installations, and how to upgrade OpenShift cluster nodes' OS packages. However, currently (as of OCP 3.6 / 2017-09) no automated tools for these tasks are provided. Automation for additional tasks like complete uninstallation, setting up cluster wide Cockpit or configuring passwordless cross-node SSH access are also provided.

The following sections will describe the playbooks and scripts in detail. The provided Ansible inventory files provide a quick starting point for OpenShift installation with most commonly used parameters included. The playbooks and scripts have been tested both with RHEL and RHEL Atomic.

This repository contains branches for each tested OCP release, the first being OpenShift Container Platform 3.4. Before running any of the scripts or playbooks it is required to investigate their contents and adjust them for the local environment as needed. Playbooks are provided in standalone form to make it easier to copy and run them anywhere.

For introduction to OpenShift and official documentation, see

https://docs.openshift.com/container-platform/latest/welcome/index.html

For related supported and more feature-rich alternatives suitable for production environments, please consider Red Hat Satellite and Red Hat CloudForms.

Prerequisites

• Environment suitable for OpenShift installation
  • https://docs.openshift.com/container-platform/latest/install_config/install/prerequisites.html
• Basic understanding of OpenShift Container Platform, Ansible, Kubernetes, and shell scripts
  • For OpenShift documentation see the OpenShift documentation
  • For Kubernetes basics see http://kubernetesbyexample.com/

Detailed Description

• Generic playbooks and scripts
  • local-registry-setup-v2 - a simple script to setup a secured local Docker v2 registry. Creates and uses a self- signed cert (/etc/docker-distribution/registry/cert.crt).
  • local-registry-sync - a script to sync all the needed container images needed for disconnected installations. This script will sync only the latest available versions to avoid downloading old and unnecessary images. After running the script the local registry contents (/etc/docker-distribution and /var/lib/registry) can be copied to other systems for example with a USB stick in case of a fully disconnected environment. For a full-featured and supported local registry, see: https://docs.openshift.com/container-platform/latest/install_config/install/stand_alone_registry.html.
  • ansible.cfg - Ansible example configuration with some optimizations to slightly reduce installation time and provide additional control and diagnostics.
  • rhel-7-ocp.ks - an example RHEL kickstart file to install OpenShift cluster hosts.
  • ansible.hosts.compact and ansible.hosts.full - Ansible example inventory files for OpenShift Container Platform installation, compact and full variants.
  • prep.yml - a playbook to automate host preparation as described in https://docs.openshift.com/container-platform/latest/install_config/install/host_preparation.html. Docker storage configuration needs to be adjusted in the template file for local environment. Supports either Red Hat Subscription-Manager or .repo file based repository configuration (for .repo file based configuration the repo/ directory must be populated locally with suitable .repo files). This playbook expects to find the local registry cert as conf/cert.crt if using a local registry.
  • install-os-updates.yml - playbook to automatically and transparently apply available OS updates on nodes on live OpenShift clusters. Automates the steps described in https://docs.openshift.com/container-platform/latest/install_config/upgrading/os_upgrades.html.
  • cross-node-cockpit.yml - a playbook to configure Cockpit for all nodes in the OpenShift cluster with masters running the Cockpit Web UI. Note that this setup is unsupported and for PoC environments only. For a supported monitoring solution, use Red Hat CloudForms. Other alternatives for Kubernetes/OpenShift monitoring include https://blog.openshift.com/full-cluster-capacity-management-monitoring-openshift/ and https://github.com/kubernetes/dashboard.
  • uninst.yml - a playbook to complete OpenShift uninstallation. See also openshift/openshift-ansible#3082.
• Less useful playbooks and scripts
  • auth.yml and post.yml - trivial playbooks to finalize OpenShift installation.
  • cross-node-ssh.yml - a playbook to configure cross-node SSH access across all the cluster nodes.
  • local-registry-setup-v1 - a simple script to setup a secured local Docker v1 registry (v1 registry is deprecated).
  • speed-up.yml - a playbook to (maybe) speed up OpenShift installations by prefetching all the needed images on selected nodes prior installation.
  • wipe-nfs.yml and wipe-docker.yml - playbooks to wipe out all NFS and Docker data. Use with caution!

There are also several helper scripts under bin/ for installing OCP on KVM/libvirt. These scripts will require some customization prior using and thus are not ready to be used out of the box but they may be helpful in some cases. In a properly prepared environment, the openshift-install or openshift-install-atomic allow for a one-step installation of the entire OpenShift platform starting from RHEL/Atomic installation on VMs and then installing and finalizing OpenShift cluster. Scripts utilize VM snapshots for faster reinstall cycles. For more serious tools see the earlier mentioned supported products.

License

Apache v2

openshift-automation-tools's Issues

test to make work with 3.7 and 3.9

clustered ssh expects keys on server

I install the OCP normally with ssh -A -l cloud-user bastion. So there are no local keys stored, but forwarded with agent. In such case you should create the keys locally while doing clustered ssh.

clustered ssh could better use sudo instead of su

Like Ansible normally uses sudo, it would make more sense for this also use sudo.

su only works when you are root everywhere. In my case I install in cloud images, so I'm typically cloud-user, which can not hop between user accounts using su. But sudo as normal ansible way, works totally fine as any user.

I commented out the following everywhere, and it works fine as cloud-user:
#become_method: su
#become_flags: '-s /bin/sh'

Use OSEv3 instead of all in uninst playbook?

Just borrowed your playbook and ran it against our new playground environment. Accidentally made a bunch of changes to our ceph nodes, cause they were in our active node list and so got grabbed by all. The main uninstall.yml in openshift-ansible targets OSEv3, might be a good idea to do it here too?