What

This will provision one or more CentOS clusters, along with Portworx, in the cloud.

Supported platforms

Container

• Kubernetes
• Openshift
• Nomad
• Rancher
• DC/OS
• Swarm

Cloud

• AWS
• GCP

How

1. Install the CLI for your choice of cloud provider:

• AWS: https://docs.aws.amazon.com/cli/latest/userguide/cli-chap-configure.html
• GCP: https://cloud.google.com/sdk/docs/quickstarts

2. Install Vagrant.

3. Install the Vagrant plugin for your choice of cloud provider:

• AWS: vagrant plugin install vagrant-aws
• GCP: vagrant plugin install vagrant-google

Note: For AWS, also need to install a dummy box:

vagrant box add dummy https://github.com/mitchellh/vagrant-aws/raw/master/dummy.box

4. Clone this repo and cd to it.

5. Configure cloud-specific environment and project/VPC:

• AWS: Edit aws-create-vpc.sh and set the variables:
  • AWS_region (you need to ensure this matches the region set in $HOME/.aws/config until mitchellh/vagrant-aws#564 is merged)
  • AWS_owner_tag - tag all of the created AWS objects so the owner can be identified
  • AWS_keypair - name of keypair to be created
• GCP: Edit gcp-create-project.sh and set the variables:
  • GCP_REGION
  • GCP_owner_tag - tags all of the GCP objects so the owner can be identified

6. Create cloud-specific VPC/project:

• AWS: bash aws-create-vpc.sh
• GCP: bash gcp-create-project.sh

7. Edit Vagrantfile.

• clusters: number of clusters
• nodes: number of nodes per cluster
• disk_size: size of storage disk in GB
• cluster_name: name of cluster (except in Kubernetes and Openshift, where it will be $clustername-$n, where $n is the cluster number)
• version: Portworx version
  • leave blank to determine latest version from install.portworx.com
  • otherwise it should be the name as the Docker tag, eg 2.1.5
• journal:
  • false: no journal
  • true: provision 3GB volume for journal
• training: only applies to Kubernetes and Openshift
  • false: install Portworx
  • true:
    • do not install Portworx
    • install shellinabox (this will listen on port 443 of each node)
• cloud: set to one of aws, gcp
• platform: set to one of kubernetes, openshift, swarm, rancher, nomad, dcos
• dcos_license: DC/OS license hash

There are also some cloud-specific variables below this section that may need to be modified. They all begin with AWS_ and GCP_.

• AWS_type: t3.large is the default, t3.medium also works
• AWS_hostname_prefix: set prefix for hostnames on AWS

8. Source the cloud-specific environment:

• AWS: . aws-env.sh
• GCP: . gcp-env.sh

9. Start the cluster(s):

$ vagrant up

10. Destroy the cluster(s):

$ vagrant destroy -fp

11. Destroy cloud-specific VPC/project:

$ sh aws-delete-vpc.sh
$ sh gcp-delete-project.sh

Notes:

• The DC/OS UI username and password are admin/admin.
• The root SSH password is portworx.
• The status.sh script will output a list of master nodes and IP addresses, useful for training sessions:

$ sh status.sh
master-1 34.245.47.251 ec2-34-245-47-251.eu-west-1.compute.amazonaws.com
master-2 34.252.74.216 ec2-34-252-74-216.eu-west-1.compute.amazonaws.com
master-3 34.245.11.144 ec2-34-245-11-144.eu-west-1.compute.amazonaws.com
...

If you need a list of VM IDs for whitelisting, use:

aws ec2 describe-instances --filters "Name=vpc-id,Values=$AWS_vpc" --query 'Reservations[*].Instances[*].[InstanceId]' --output text