NOTE: NOT FOR PRODUCTION USE

Please note these scripts are intended to allow experimentation with Azure Container Service. They are not intended for production use.

A set of convenience scripts for creating and testing ACS clusters. These scripts can also be helpful in working out how to use the REST API interfaces for managing applicaitons on an ACS cluster.

• Azure CLI installed and configured to access the test subscription
  • install Node and NPM
  • `sudo npm install azure-cli -g'
• Install jq
  • apt-get install jq
• Install Python
  • apt-get install python
• Install required Python libraries
  • pip install -r requirements.txt
• Whitelisted for ACS preview

Create a cluster.ini by copying cluster.ini.tmpl and editing accoringly.

Create a service principle for the application and add the details to your cluster.ini.

$ ./acs.py --help
Usage: acs.py [options] command

Options:
  -h, --help            show this help message and exit
  -c CONFIG_FILE, --config_file=CONFIG_FILE
                        define the configuration file to use. Default is
                        'cluster.ini'

There are a number of accepted commands, as follows

The deploy command will create or update a deployment.

python acs.py deploy [-c CONFIG_FILE]

In order to create a new cluster ensure that the dns_prefix in CONFIG_FILE does not already exist.

In order to update a new cluster you will run the deployment using a dns_prefix in CONFIG_FILE that already exists. The cluster will be modified to match any updated parameters. For example, you can increase the agent count.

You can execute arbitrary docker commands on each agent using the doocker command. This is a pass through of the standard Docker CLI.

It can be used, for example, to pull a docker container to all nodes in a cluster. This is not a necessary step, Mesos will pull the container image automatically for you, however pulling the container ahead of time makes the first starup on each node faster.

python acs.py docker "pull CONTAINER_NAME"

Adds a feature to the ACS cluster. Possible features are described below. Features can be added at deployment time by specifiying them in a comma separated list in the cluster ini file.

./acs.py addFeature FEATURE_LIST

Where FEATURE_LIST is a comma separated list of features required.

Azure File Service is a storage driver that enables multiple Docker containers to read and write to a shared folder. To add this feature simply run:

python acs.py addFeature afs

This will create a Storage Account on Azure, crate a share and mount that share on each of the agents in your cluster.

Add the OMS monitoring solution to the cluster. You will first need to register for the OMS service and then complete the OMS section of the cluster.ini file. Finally, run the following command.

Visit http://mms.microsoft.com and sign up for a free OMS account. Then click "Settings" and then "Connected Sources". You will need to copy your Workspace ID and your Workspace Primary Key into your config file. Finally, run the following command to install OMS on each of your agents (please note that this command restarts the Docker Engine and thus any containers on it will be stopped).

python acs.py addFeature oms

If an agent is added to the cluster it will not have the Azure File Service feature added by default.

delete will delete the cluster and all associated resource.

python acs.py delete [-c CONFIG_FILE]

NOTE: for this command to work you must have opened an SSH Tunnel to your cluster, or you must run the command from a VM inside the clusters VNET.

test will deploy some test applications and ensure they are started correctly on the cluster. The tests will be run against a cluster defined in the cluster.ini file (or the file specified with -c).

python acs.py test [-c CONFIG_FILE]

This command performs various actions, such as deploying a multi-container application and verifying it is working correctly. The log outputs of these test scripts detail the commands being run and can therefore be useful as a learning excercise, as well as testing whether the cluster is correctly configured.