Vmcli aims to provide command line tool to allow virtual machine management and delivery by interacting with VMware's vSphere API. It implements simple methods and classes for modules expansion, making addition of new vmcli subcommands easy as inheriting classes in the Python programming language.
Other aim is to make as much directives configurable as possible. Options can be provided via default values, command line arguments, configuration file directives and environment variables. For the create or clone subcommands there are flavors too, these represents templates for virtual machine's hardware configuration.
To begin using vmcli you need to have python2.7, python-virtualenv and python-pip packages installed. Then execute the following steps:
virtualenv env
source env/bin/activate
pip install -r requirements.txt
vmcli.py --help
If you plan to use --net-cfg option during clone/create subcommands, ensure appropriate script is present in your template (/usr/share/vmcli/provision-interfaces.sh). Example of this script can be found in this repository in examples/provision-interfaces.sh
Features likes Tags require vsphere-automation-sdk-python library (Python3). Make sure to install if if you plan to use them.
Options can be provided and overriden in a variety of ways. Foremost being command line arguments. Specific ways to provide configuration to the vmcli program are explained below in the exact order as they are prioritized.
Arguments being most straightforward way to pass configuration options to the running program. All the available command line arguments can be found via --help argument passed to the program like this - vmcli.py --help.
Help argument (--help) can be passed to any subcommand as well. Available subcommands of the vmcli program are listed in the positional arguments section of the help output delivered by main program (shown with the command above).
Flavors are specific feature for virtual machine deploy operations such as create, create-empty or clone. They work similar to flavors used in any other tools or virtualization platforms (OpenStack's flavors, AWS's instance types, etc.) and can be defined by the user himself.
To define custom flavor, start with copying file examples/flavor.yml.example into flavors/some-name.yml. The name you choose will be later used when pointing vmcli program to flavor you wish to use. This file contains only Python dictionary representation of data.
Fill each row as you desire respecting formatting in example. If you want to omit some directives, simply delete entire row or type in None as a value: folder: None and that value will be fetched from other sources listed in this section.
When flavor is finally defined, you can provide --flavor some-name option to one of the subcommands which support flavors. For example: ./vmcli.py create --flavor m1_tiny
When vmcli program does not find any cmd line argument or flavor directive for particular setting, it tries to load its value from environment variable, which name is hardcoded in the source code itself. For now, to get list of available environment variables and use some of them, simply run:
grep -o "'VMCLI[_A-Z]*'" lib/config.py export VMCLI_LOG_PATH=/var/log/vmcli.log
If you are unsure what data to fill to a specific ENV variable you can study simple lib/config.py file and its get_config(category, name, env_variable, data_type, default_value) function calls.
There is one special environment variable VMCLI_CONFIG_FILE, which points program to a specific vmcli.yml config file. This way, you can define multiple config files (e.g. ~/.dev.yaml, ~/.prod.yaml).
You can place any executable file into callbacks directory and it gets executed right after running 'create' sub-command. This is a great way to hook up your configuration management or any other provisioning scripts you may have.
First argument to each callback will always be JSON dictionary containing all valid arguments passed to vmcli. Should you require additional arguments to be passed to your callbacks, use --callback option. Ordering of executables may be achieved by naming them 01script.sh, 02script.py and so on.
- implement cache to speed up get_obj calls