Skip to content

synctree/terraform-aws-openshift

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

46 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

terraform-aws-openshift

This project shows you how to set up OpenShift Origin on AWS using Terraform. This the companion project to my article Get up and running with OpenShift on AWS.

OpenShift Sample Project

I am also adding some 'recipes' which you can use to mix in more advanced features:

Index

Overview

Terraform is used to create infrastructure as shown:

Network Diagram

Once the infrastructure is set up an inventory of the system is dynamically created, which is used to install the OpenShift Origin platform on the hosts.

Prerequisites

You need:

  1. Terraform - brew update && brew install terraform
  2. An AWS account, configured with the cli locally -
if [[ "$unamestr" == 'Linux' ]]; then
        dnf install -y awscli || yum install -y awscli
elif [[ "$unamestr" == 'FreeBSD' ]]; then
        brew install -y awscli
fi

Creating the Cluster

Create the infrastructure first:

# Make sure ssh agent is on, you'll need it later.
eval `ssh-agent -s`

# Create the infrastructure.
make infrastructure

You will be asked for a region to deploy in, use us-east-1 or your preferred region. You can configure the nuances of how the cluster is created in the main.tf file. Once created, you will see a message like:

$ make infrastructure
var.region
  Region to deploy the cluster into

  Enter a value: ap-southeast-1

...

Apply complete! Resources: 20 added, 0 changed, 0 destroyed.

That's it! The infrastructure is ready and you can install OpenShift. Leave about five minutes for everything to start up fully.

Installing OpenShift

To install OpenShift on the cluster, just run:

make openshift

You will be asked to accept the host key of the bastion server (this is so that the install script can be copied onto the cluster and run), just type yes and hit enter to continue.

It can take up to 30 minutes to deploy. If this fails with an ansible not found error, just run it again.

Once the setup is complete, just run:

make browse-openshift

To open a browser to admin console, use the following credentials to login:

Username: admin
Password: 123

Accessing and Managing OpenShift

There are a few ways to access and manage the OpenShift Cluster.

OpenShift Web Console

You can log into the OpenShift console by hitting the console webpage:

make browse-openshift

# the above is really just an alias for this!
open $(terraform output master-url)

The url will be something like https://a.b.c.d.xip.io:8443.

The Master Node

The master node has the OpenShift client installed and is authenticated as a cluter administrator. If you SSH onto the master node via the bastion, then you can use the OpenShift client and have full access to all projects:

$ make ssh-master # or if you prefer: ssh -t -A ec2-user@$(terraform output bastion-public_dns) ssh master.openshift.local
$ oc get pods
NAME                       READY     STATUS    RESTARTS   AGE
docker-registry-1-d9734    1/1       Running   0          2h
registry-console-1-cm8zw   1/1       Running   0          2h
router-1-stq3d             1/1       Running   0          2h

Notice that the default project is in use and the core infrastructure components (router etc) are available.

You can also use the oadm tool to perform administrative operations:

$ oadm new-project test
Created project test

The OpenShift Client

From the OpenShift Web Console 'about' page, you can install the oc client, which gives command-line access. Once the client is installed, you can login and administer the cluster via your local machine's shell:

oc login $(terraform output master-url)

Note that you won't be able to run OpenShift administrative commands. To administer, you'll need to SSH onto the master node. Use the same credentials (admin/123) when logging through the commandline.

Connecting to the Docker Registry

The OpenShift cluster contains a Docker Registry by default. You can connect to the Docker Registry, to push and pull images directly, by following the steps below.

First, make sure you are connected to the cluster with The OpenShift Client:

oc login $(terraform output master-url)

Now check the address of the Docker Registry. Your Docker Registry url is just your master url with docker-registry-default. at the beginning:

% echo $(terraform output master-url)
https://54.85.76.73.xip.io:8443

In the example above, my registry url is https://docker-registry-default.54.85.76.73.xip.io:8443. You can also get this url by running oc get routes -n default on the master node.

You will need to add this registry to the list of untrusted registries. The documentation for how to do this here https://docs.docker.com/registry/insecure/. On a Mac, the easiest way to do this is open the Docker Preferences, go to 'Daemon' and add the address to the list of insecure regsitries:

Docker Insecure Registries Screenshot

Finally you can log in. Your Docker Registry username is your OpenShift username (admin by default) and your password is your short-lived OpenShift login token, which you can get with oc whoami -t:

% docker login docker-registry-default.54.85.76.73.xip.io -u admin -p `oc whoami -t`
Login Succeeded

You are now logged into the registry. You can also use the registry web interface, which in the example above is at: https://registry-console-default.54.85.76.73.xip.io

Atomic Registry Screenshot

Additional Configuration

The easiest way to configure is to change the settings in the ./inventory.template.cfg file, based on settings in the OpenShift Origin - Advanced Installation guide.

When you run make openshift, all that happens is the inventory.template.cfg is turned copied to inventory.cfg, with the correct IP addresses loaded from terraform for each node. Then the inventory is copied to the master and the setup script runs. You can see the details in the makefile.

Choosing the OpenShift Version

To change the version, just update the version identifier in this line of the ./install-from-bastion.sh script:

git clone -b release-3.6 https://github.com/openshift/openshift-ansible

Available versions are listed here.

OpenShift 3.5 is fully tested, and has a slightly different setup. You can build 3.5 by checking out the release/openshift-3.5 branch.

Destroying the Cluster

Bring everything down with:

terraform destroy

Makefile Commands

There are some commands in the makefile which make common operations a little easier:

Command Description
make infrastructure Runs the terraform commands to build the infra.
make openshift Installs OpenShift on the infrastructure.
make browse-openshift Opens the OpenShift console in the browser.
make ssh-bastion SSH to the bastion node.
make ssh-master SSH to the master node.
make ssh-node1 SSH to node 1.
make ssh-node2 SSH to node 2.
make sample Creates a simple sample project.

Pricing

You'll be paying for:

  • 1 x m4.xlarge instance
  • 2 x t2.large instances

Recipes

Your installation can be extended with recipes.

Splunk

You can quickly add Splunk to your setup using the Splunk recipe:

Splunk Screenshot

To integrate with splunk, merge the recipes/splunk branch then run make splunk after creating the infrastructure and installing OpenShift:

git merge recipes/splunk
make infracture
make openshift
make splunk

There is a full guide at:

http://www.dwmkerr.com/integrating-openshift-and-splunk-for-logging/

You can quickly rip out container details from the log files with this filter:

source="/var/log/containers/counter-1-*"  | rex field=source "\/var\/log\/containers\/(?<pod>[a-zA-Z0-9-]*)_(?<namespace>[a-zA-Z0-9]*)_(?<container>[a-zA-Z0-9]*)-(?<conatinerid>[a-zA-Z0-9_]*)" | table time, host, namespace, pod, container, log

Troubleshooting

Image pull back off, Failed to pull image, unsupported schema version 2

Ugh, stupid OpenShift docker version vs registry version issue. There's a workaround. First, ssh onto the master:

$ ssh -A ec2-user@$(terraform output bastion-public_dns)

$ ssh master.openshift.local

Now elevate priviledges, enable v2 of of the registry schema and restart:

sudo su
oc set env dc/docker-registry -n default REGISTRY_MIDDLEWARE_REPOSITORY_OPENSHIFT_ACCEPTSCHEMA2=true
systemctl restart origin-master.service

You should now be able to deploy. More info here.

References

Releases

No releases published

Packages

No packages published

Languages

  • HCL 64.4%
  • Shell 26.4%
  • Makefile 9.2%