Please note that this project hasn't been updated since 2016 and hence we've decided to archive it. The code is likely obsolete, as it uses old versions of Terraform, Google Cloud SDK and Mesos.
- This module requires Terraform 0.6.16 or greater
- Follow the instructions on https://www.terraform.io/intro/getting-started/install.html to set up Terraform on your machine.
Authenticating with Google Cloud services requires a JSON file which we call the account file. This file is downloaded directly from the Google Developers Console. Follow these steps:
- Log into the Google Developers Console and select
your google project
. - The API Manager view should be selected, click on "Credentials" on the left, then "Create credentials", and finally "Service account key".
- Select "Compute Engine default service account" in the "Service account" dropdown, and select "JSON" as the key type.
- Clicking "Create" will download your account_file.
- Visit https://cloud.google.com/sdk/
- Install the SDK, login and authenticate with your Google Account.
- Back in the Developer Console, go to Compute - Compute Engine - Metadata and click the SSH Keys tab. Add your public SSH key there.
- Use the path to the private key and the username in the next step as
gce_ssh_user
andgce_ssh_private_key_file
Create a file mesos.tf
containing something like this:
module "mesos" {
source = "github.com/ContainerSolutions/terraform-mesos"
account_file = "/path/to/your.key.json"
project = "your google project ID"
region = "europe-west1"
zone = "europe-west1-d"
gce_ssh_user = "user"
gce_ssh_private_key_file = "/path/to/private.key"
name = "mymesoscluster"
masters = "3"
slaves = "5"
subnetwork = "10.20.30.0/24"
domain = "example.com"
mesos_version = "0.28.2"
image = "rhel-7-v20160418"
distribution = "redhat"
slave_machine_type = "n1-standard-2"
# slave_resources = "cpus(*):0.90; disk(*):7128"
}
See the variables.tf
file for the available variables and their defaults
If you set image
to the standard Ubuntu 15.04 GCE image name, you get the standard Mesos version distributed with this operating system.
image = "ubuntu-1504-vivid-v20151120"
If you decide to use a specific version of Mesos, which does exist as an Ubuntu package, enter the version number to the optional mesos_version
configuration option.
image = "ubuntu-1504-vivid-v20151120"
mesos_version = "0.25.0-0.2.70.ubuntu1504"
You might want to try Mesos installed from a specific commit (e.g. "69d4cf654", or "master"). In order to do it, build a GCE virtual machine image (see images/README.md) with Mesos installed and use the GCE_IMAGE_NAME
you give it as the image
configuration option, e.g.:
image = "ubuntu-1404-trusty-mesos"
Download the module
terraform get -update
Create the plan and save it to a file. Use module-depth 1 to show the configuration of the resources inside the module.
terraform plan -out my.plan -module-depth=1
Once you are satisfied with the plan, apply it.
terraform apply my.plan
Ports 80, 443 and 22 are open on all the machines within the cluster. Accessing other ports, e.g. Mesos GUI (port 5050) or Marathon GUI (port 8080) is only possible with VPN connection set up.
Use the following commands to download client.ovpn
file. Then use it to establish VPN with the cluster.
OVPNFILE=`terraform output -module mesos openvpn`
scp -o BatchMode=yes -o StrictHostKeyChecking=no -o UserKnownHostsFile=/dev/null $OVPNFILE .
sudo openvpn --config client.ovpn
Once the VPN is up, you can access all machines within the cluster using their private IP addresses. Open a second tab to execute the following commands
export INTERNAL_IP_M0=`gcloud compute instances list --regexp .*master-0.* --format='value(networkInterfaces[].networkIP.list())'`
open http://$INTERNAL_IP_M0:5050 # Mesos Console
open http://$INTERNAL_IP_M0:8080 # Marathon Console
mesos config master zk://$INTERNAL_IP_M0:2181/mesos # for those who have the local mesos client installed
curl -s $INTERNAL_IP_M0:5050/master/slaves | python -mjson.tool | grep -e pid -e disk -e cpus
When you're done, clean up the cluster with
terraform destroy
- Cannot reach the log files of the Mesos slave nodes from the web interface on the leading master
The installation and configuration used in this module is based on this excellent howto: https://www.digitalocean.com/community/tutorials/how-to-configure-a-production-ready-mesosphere-cluster-on-ubuntu-14-04